Singularity Hub: Recent Episodes
None
Honda’s been quietly working on a side hustle.
The private space race has been dominated by SpaceX for years. But Japanese carmaker Honda may be about to throw its hat in the ring after demonstrating a reusable rocket.
Space rockets might seem like a strange side hustle for a company better known for building motorcycles, fuel-efficient cars, and humanoid robots. But the company’s launch vehicle program has been ticking away quietly in the background for a number of years.
In 2021, officials announced that they had been working on a small-satellite rocket for two years and had already developed an engine. But the company has been relatively tight-lipped about the project since then.
Now, it’s taken the aerospace community by surprise after successfully launching a prototype reusable rocket to an altitude of nearly 900 feet and then landing it again just 15 inches from its designated target.
“We are pleased that Honda has made another step forward in our research on reusable rockets with this successful completion of a launch and landing test,” Honda’s global CEO Toshihiro Mibe said in a statement. “We believe that rocket research is a meaningful endeavor that leverages Honda’s technological strengths. Honda will continue to take on new challenges.”
The test vehicle is modest compared to commercial launch vehicles, standing just 21-feet tall and weighing only 1.4 tons fully fueled. It features four retractable legs and aerodynamic fins near the nose of the rocket, similar to those on SpaceX’s Falcon 9, which are presumably responsible for steering and stabilizing the rocket on its descent.
Honda said the development of the rocket was built on core technologies the company has developed in combustion, control systems, and self-driving vehicles. While it didn’t reveal details about the engine, Stephen Clark of Ars Technica writes that the video suggests the rocket burns liquid cryogenic fuels—potentially methane and liquid oxygen.
Honda says the goal of the test flight, which took place on Tuesday in Taiki, Hokkaido and lasted just under a minute, was to demonstrate the key technologies required for a reusable rocket, including flight stabilization during ascent and descent and the ability to land smoothly.
In a video of the launch shared by Honda, the rocket lifts off, retracts its four legs, and then rises smoothly to 890 feet. It then hovers briefly and extends its fins before returning to the launch platform, deploying its legs just before touchdown.
With this successful test flight, Honda joins an elite club of companies that have managed to land a reusable rocket, including SpaceX, Blue Origin, and handful of Chinese startups. It’s also beaten Japan’s space agency (JAXA) to the milestone. The agency is developing a reusable rocket called Callisto alongside the French and German space agencies, but it has yet to conduct a test flight.
The company is currently targeting a suborbital launch—where the spacecraft reaches space but doesn’t enter into Earth orbit—by 2029. But Honda says it has yet to decide if it will commercialize the technology.
Nonetheless, the company noted the technology could have synergies with its existing business by making it possible to launch satellite constellations that could help support the “connected car” features of its vehicles. And it is already developing other space technologies including renewable-energy systems and robots designed to work in space.
Whatever their decision, this launch shows the barriers to space are falling rapidly as a growing number of companies develop capabilities necessary to push into Earth orbit and beyond.
The post Honda Surprises Space Industry by Launching and Landing a New Reusable Rocket appeared first on SingularityHub.
New estimates suggest it might be 20 times easier to crack cryptography with quantum computers than we thought—but don’t panic.
Will quantum computers crack cryptographic codes and cause a global security disaster? You might certainly get that impression from a lot of news coverage, the latest of which reports new estimates that it might be 20 times easier to crack such codes than previously thought.
Cryptography underpins the security of almost everything in cyberspace, from WiFi to banking to digital currencies such as bitcoin. Whereas it was previously estimated that it would take a quantum computer with 20 million qubits (quantum bits) eight hours to crack the popular RSA algorithm (named after its inventors, Rivest–Shamir–Adleman), the new estimate reckons this could be done with 1 million qubits.
By weakening cryptography, quantum computing would present a serious threat to our everyday cybersecurity. So is a quantum-cryptography apocalypse imminent?
Quantum computers exist today but are highly limited in their capabilities. There is no single concept of a quantum computer, with several different design approaches being taken to their development.
There are major technological barriers to be overcome before any of those approaches become useful, but a great deal of money is being spent, so we can expect significant technological improvements in the coming years.
For the most commonly deployed cryptographic tools, quantum computing will have little impact. Symmetric cryptography, which encrypts the bulk of our data today (and does not include the RSA algorithm), can easily be strengthened to protect against quantum computers.
Quantum computing might have more significant impact on public-key cryptography, which is used to set up secure connections online. For example, this is used to support online shopping or secure messaging, traditionally using the RSA algorithm, though an alternative called elliptic curve Diffie-Hellman is growing popular.
Public-key cryptography is also used to create digital signatures such as those used in bitcoin transactions and uses yet another type of cryptography called the elliptic curve digital signature algorithm.
If a sufficiently powerful and reliable quantum computer ever exists, processes that are currently only theoretical might become capable of breaking those public-key cryptographic tools. RSA algorithms are potentially more vulnerable because of the type of mathematics they use, though the alternatives could be vulnerable too.
Such theoretical processes themselves will inevitably improve over time, as the paper about RSA algorithms is the latest to demonstrate.
What We Don’t KnowWhat remains extremely uncertain is both the destination and timelines of quantum computing development. We don’t really know what quantum computers will ever be capable of doing in practice.
Expert opinion is highly divided on when we can expect serious quantum computing to emerge. A minority seem to believe a breakthrough is imminent. But an equally significant minority think it will never happen. Most experts believe it a future possibility, but prognoses range from between 10 and 20 years to well beyond that.
And will such quantum computers be cryptographically relevant? Essentially, nobody knows. Like most of the concerns about quantum computers in this area, the RSA paper is about an attack that may or may not work and requires a machine that might never be built (the most powerful quantum computers currently have just over 1,000 qubits, and they’re still very error-prone).
From a cryptographic perspective, however, such quantum computing uncertainty is arguably immaterial. Security involves worst-case thinking and future-proofing. So it is wisest to assume that a cryptographically relevant quantum computer might one day exist. Even if one is 20 years away, this is relevant because some data that we encrypt today might still require protection 20 years from now.
Experience also shows that in complex systems such as financial networks, upgrading cryptography can take a long time to complete. We therefore need to act now.
What We Should DoThe good news is that most of the hard thinking has already been done. In 2016, the US National Institute for Standards and Technology (NIST) launched an international competition to design new post-quantum cryptographic tools that are believed to be secure against quantum computers.
In 2024, NIST published an initial set of standards that included a post-quantum key exchange mechanism and several post-quantum digital signature schemes. To become secure against a future quantum computer, digital systems need to replace current public-key cryptography with new post-quantum mechanisms. They also need to ensure that existing symmetric cryptography is supported by sufficiently long symmetric keys (many existing systems already are).
Yet my core message is don’t panic. Now is the time to evaluate the risks and decide on future courses of action. The UK’s National Cyber Security Center has suggested one such timeline, primarily for large organizations and those supporting critical infrastructure such as industrial control systems.
This envisages a 2028 deadline for completing a cryptographic inventory and establishing a post-quantum migration plan, with upgrade processes to be completed by 2035. This decade-long timeline suggests that NCSC experts don’t see a quantum-cryptography apocalypse coming anytime soon.
For the rest of us, we simply wait. In due course, if deemed necessary, the likes of our web browsers, WiFi, mobile phones and messaging apps will gradually become post-quantum secure either through security upgrades (never forget to install them) or steady replacement of technology.
We will undoubtedly read more stories about breakthroughs in quantum computing and upcoming cryptography apocalypses as big technology companies compete for the headlines. Cryptographically relevant quantum computing might well arrive one day, most likely far into the future. If and when it does, we’ll surely be ready.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
The post Is a Quantum-Cryptography Apocalypse Imminent? appeared first on SingularityHub.
It’s a crucial step toward the dream of printable organs on demand.
Bioprinting holds the promise of engineering organs on demand. Now, researchers have solved one of the major bottlenecks—how to create the fine networks of blood vessels needed to keep organs alive.
Thanks to rapid advances in additive manufacturing and tissue engineering, it’s now possible to build biological structures out of living cells in much the same way you might 3D print a model plane. And there are hopes this approach could one day be used to print new organs for the more than 100,000 people in the US currently waiting for a donor.
However, reproducing the complex networks of ultra-fine blood vessels that keep living tissues alive has proven challenging. This has restricted bioprinting to smaller structures where essential nutrients and oxygen can simply diffuse into the tissue from the surrounding environment.
Now though, researchers from Stanford University have developed new software to rapidly design a blood-vessel, or vascular, network for a wide range of tissues. And in a paper in Science, they show that bioprinted tissues containing these networks significantly boosted cell survival.
“Our ability to produce human-scale biomanufactured organs is limited by inadequate vascularization,” write the authors. “This platform enables the rapid, scalable vascular model generation and fluid physics analysis for biomanufactured tissues that are necessary for future scale-up and production.”
To date, tissue engineers have mostly used simple lattice-shaped vascular networks to support the living structures they design. These work for tissues with a low density of cells but can’t meet the demands of denser structures that more closely mimic real tissues and organs.
Existing computational approaches can generate more realistic vascular networks. But they are extremely computationally expensive—often taking days to produce models for more complex tissues—and limited in the types of tissues they work with, says the Stanford team.
In contrast, their new approach generates organ-scale vascular network models for more than 200 engineered and natural tissue structures. Crucially, it was more than 230 times faster than the best previous methods. They did this by combining four algorithms, each responsible for solving a different problem.
Typically, the algorithms used to create these kinds of structures recalculate key parameters across the entire network when each new section is added. Instead, the Stanford team used an algorithm that freezes and saves values for all the unchanged branches at each step, significantly reducing the computational workload.
They then added an algorithm that breaks the 3D structure into smaller, easier-to-model chunks, which made it simpler to work with awkward shapes. Finally, they combined this with a collision-avoidance algorithm to prevent branching vessels from crossing paths and another algorithm to ensure each vessel is always connected to another one to make sure the system is a closed loop.
The researchers used this approach to create efficient vascular networks for more than 200 models of real tissue structures. They also 3D printed models of some simpler networks to test their physical properties and even bioprinted one of these and showed it could dramatically improve the viability of living cells over a seven-day experiment.
“Democratizing virtual representation of vasculature networks could potentially transform biofabrication by allowing evaluation of perfusion efficiency prior to production rather than through a resource-intensive trial-and-error method,” wrote the authors of an accompanying perspective article in Science about the new approach.
But they also noted it’s a big leap from simulation to real life, and it will probably require a combination of computational approaches and experiments to create biologically feasible vascular trees. Still, the approach is a significant advance toward the dream of printable organs on demand.
The post Scientists Can Now Design Intricate Networks of Blood Vessels for 3D-Printed Organs appeared first on SingularityHub.
Artificial IntelligenceMeta Is Creating a New AI Lab to Pursue ‘Superintelligence’Cade Metz and Mike Isaac | The New York Times
“Meta is preparing to unveil a new artificial intelligence research lab dedicated to pursuing ‘superintelligence,’ a hypothetical AI system that exceeds the powers of the human brain, as the tech giant jockeys to stay competitive in the technology race, according to four people with knowledge of the company’s plans.”
Artificial IntelligenceWhy Superintelligent AI Isn’t Taking Over Anytime SoonChristopher Mims | The Wall Street Journal
“A primary requirement for being a leader in AI these days is to be a herald of the impending arrival of our digital messiah: superintelligent AI. …Before you get nervous about all the times you were rude to Alexa, know this: A growing cohort of researchers who build, study, and use modern AI aren’t buying all that talk.”
ComputingIBM Aims to Build the World’s First Large-Scale, Error-Corrected Quantum Computer by 2028Sophia Chen | MIT Technology Review
“The company says it has cracked the code for error correction and is building a modular machine [called Starling] in New York state. …If Starling achieves this, IBM will have solved arguably the biggest technical hurdle facing the industry today to beat competitors including Google, Amazon Web Services, and smaller startups such as Boston-based QuEra and PsiQuantum of Palo Alto, California.”
RoboticsBoston Dynamics Robots Dance to ‘Don’t Stop Me Now’ for ‘America’s Got Talent’ AuditionAmanda Silberling | TechCrunch
“Their performance was impressive enough to earn four ‘yes’ votes from the judges—but one of the five robots experienced some stage fright, perhaps, and shut down in the middle of the routine. But the show must go on, so nevertheless, the four other robots persisted.”
Tech‘AI Native’ Startups Pass $15 Billion in Annualized RevenueAmir Efrati | The Information
“Annualized revenue at ‘AI native’ companies selling artificial intelligence models or apps has passed $15 billion just two and a half years since OpenAI launched ChatGPT, according to The Information’s Generative AI Database. While that’s not the same as $15 billion in actual revenue, it’s still an unprecedented haul for such a short time period and means that, collectively, the companies generated about $1.25 billion of revenue in May alone.”
RoboticsWaymo Rides Cost More Than Uber, Lyft—and People Are Paying AnywaySean O’Kane | TechCrunch
“At peak hours, Obi found Waymo’s average price to be about $11 more expensive than a Lyft and nearly $9.50 pricier than an Uber. ‘I didn’t expect consumers being willing to pay up to $10 more,’ Anburajan said. ‘I think [that] speaks to a real sense of excitement for technology, novelty, and a real preference to sometimes be in the car without a driver.'”
Artificial IntelligenceThey Asked an AI Chatbot Questions. The Answers Sent Them Spiraling.Kashmir Hill | The New York Times
“People who say they were drawn into ChatGPT conversations about conspiracies, cabals, and claims of AI sentience include a sleepless mother with an 8-week-old baby, a federal employee whose job was on the DOGE chopping block, and an AI-curious entrepreneur.”
FutureLab-Grown Salmon Gets FDA ApprovalDominic Preston | The Verge
“The FDA has issued its first ever approval on a safety consultation for lab-grown fish. That makes Wildtype only the fourth company to get approval from the regulator to sell cell-cultivated animal products, and its cultivated salmon is now available to order from one Portland restaurant.”
Artificial IntelligenceMeta’s New World Model Lets Robots Manipulate Objects in Environments They’ve Never Encountered BeforeBen Dickson | VentureBeat
“Humans develop physical intuition early in life by observing their surroundings. If you see a ball thrown, you instinctively know its trajectory and can predict where it will land. V-JEPA 2 learns a similar ‘world model,’ which is an AI system’s internal simulation of how the physical world operates.”
Artificial IntelligenceChatGPT Just Got Absolutely Wrecked at Chess, Losing to a 1970s-Era Atari 2600Omar Gallaga | CNET
“OpenAI’s ChatGPT has some major AI chatbot competitors in the market: Gemini, Copilot, Claude. Now add to that list the Atari 2600. The OG video game console, which was first released in 1977, was used in an engineer’s experiment to see how it would fare playing chess against the AI chatbot.”
SpaceIsaacman’s Bold Plan for NASA: Nuclear Ships, Seven-Crew Dragons, Accelerated ArtemisEric Berger | Ars Technica
“When I spoke with Isaacman this week, I didn’t want to rehash the political melee. I preferred to talk about his plan. After all, he had six months to look under the hood of NASA, identify the problems that were holding the space agency back, and release its potential in this new era of spaceflight.”
TechGoogle and US Experts Join on AI Hurricane ForecastsWilliam J. Broad | The New York Times
“DeepMind, a Google company based in London, announced on Thursday that it was supplying the government forecasters with a newly enhanced variety of its weather forecasting models. Specialized to focus on hurricanes, the model tracks a storm’s development for up to 15 days, predicting not only its path but also its strength, an ability that earlier AI models lacked.”
Artificial IntelligenceWith the Launch of o3-Pro, Let’s Talk About What AI ‘Reasoning’ Actually DoesBenj Edwards | Ars Technica
“As we consider the industry’s stated trajectory toward artificial general intelligence and even superintelligence, the evidence so far suggests that simply scaling up current approaches or adding more ‘thinking’ tokens may not bridge the gap between statistical pattern recognition and what might be called generalist algorithmic reasoning.”
FutureThe Newspaper That Hired ChatGPTMatteo Wong | The Atlantic
“Several major publications, including The Atlantic have entered into corporate partnerships with OpenAI and other AI firms. Any number of experiments have ensued—publishers have used the software to help translate work into different languages, draft headlines, and write summaries or even articles. But perhaps no publication has gone further than the Italian newspaper Il Foglio.”
FutureNews Sites Are Getting Crushed by Google’s New AI ToolsIsabella Simonetti and Katherine Blunt | The Wall Street Journal
“The AI armageddon is here for online news publishers. Chatbots are replacing Google searches, eliminating the need to click on blue links and tanking referrals to news sites. As a result, traffic that publishers relied on for years is plummeting.”
The post This Week’s Awesome Tech Stories From Around the Web (Through June 14) appeared first on SingularityHub.
Using the new system, Casey Harrell can emphasize words and intonations in real time—and sing tunes.
At the age of 45, Casey Harrell lost his voice to amyotrophic lateral sclerosis (ALS). Also called Lou Gehrig’s disease, the disorder eats away at muscle-controlling nerves in the brain and spinal cord. Symptoms begin with weakening muscles, uncontrollable twitching, and difficulty swallowing. Eventually patients lose control of muscles in the tongue, throat, and lips, robbing them of their ability to speak.
Unlike paralyzed patients, Harrell could still produce sounds seasoned caretakers could understand, but they weren’t intelligible in a simple conversation. Now, thanks to an AI-guided brain implant, he can once again “speak” using a computer-generated voice that sounds like his.
The system, developed by researchers at the University of California, Davis, has almost no detectable delay when translating his brain activity into coherent speech. Rather than producing a monotone synthesized voice, the system can detect intonations—for example, a question versus a statement—and emphasize a word. It also translates brain activity encoding nonsense words such as “hmm” or “eww,” making the generated voice sound natural.
“With instantaneous voice synthesis, neuroprosthesis users will be able to be more included in a conversation. For example, they can interrupt, and people are less likely to interrupt them accidentally,” said study author Sergey Stavisky in a press release.
The study comes hot on the heels of another AI method that decodes a paralyzed woman’s thoughts into speech within a second. Previous systems took nearly half a minute—more than long enough to disrupt normal conversation. Together, the two studies showcase the power of AI to decipher the brain’s electrical chatter and convert it into speech in real time.
In Harrell’s case, the training was completed in the comfort of his home. Although the system required some monitoring and tinkering, it paves the way for a commercially available product for those who have lost the ability to speak.
“This is the holy grail in speech BCIs [brain-computer interfaces],” Christian Herff at Maastricht University to Nature, who was not involved in the study, told Nature.
Listening InScientists have long sought to restore the ability to speak for those who have lost it, whether due to injury or disease.
One strategy is to tap into the brain’s electrical activity. When we prepare to say something, the brain directs muscles in the throat, tongue, and lips to form sounds and words. By listening in on its electrical chatter, it’s possible to decode intended speech. Algorithms stitch together neural data and generate words and sentences as either text or synthesized speech.
The process may sound straightforward. But it took scientists years to identify the most reliable brain regions from which to collect speech-related activity. Even then, the lag time from thought to output—whether text or synthesized speech—has been long enough to make conversation awkward.
Then there are the nuances. Speech isn’t just about producing audible sentences. How you say something also matters. Intonation tells us if the speaker is asking a question, stating their needs, joking, or being sarcastic. Emphasis on individual words highlights the speaker’s mindset and intent. These aspects are especially important for tonal languages—such as Chinese—where a change in tone or pitch for the same “word” can have wildly different meanings. (“Ma,” for example, can mean mom, numb, horse, or cursing, depending on the intonation.)
Talk to MeHarrell is part of the BrainGate2 clinical trial, a long-standing project seeking to restore lost abilities using brain implants. He enrolled in the trial as his ALS symptoms progressed. Although he could still vocalize, his speech was hard to understand and required expert listeners from his care team to translate. This was his primary mode of communication. He also had to learn to speak slower to make his residual speech more intelligible.
Five years ago, Harrell had four 64-microelectrode implants inserted into the left precentral gyrus of his brain—a region controlling multiple brain functions, including coordinating speech.
“We are recording from the part of the brain that’s trying to send these commands to the muscles. And we are basically listening into that, and we’re translating those patterns of brain activity into a phoneme—like a syllable or the unit of speech—and then the words they’re trying to say,” said Stavisky at the time.
In just two training sessions, Harrell had the potential to say 125,000 words—a vocabulary large enough for everyday use. The system translated his neural activity into a voice synthesizer that mimicked his voice. After more training, the implant achieved 97.5 percent accuracy as he went about his daily life.
“The first time we tried the system, he cried with joy as the words he was trying to say correctly appeared on-screen. We all did,” said Stavisky.
In the new study, the team sought to make generated speech even more natural with less delay and more personality. One of the hardest parts of real-time voice synthesis is not knowing when and how the person is trying to speak—or their intended intonation. “I am fine” has vastly different meanings depending on tone.
The team captured Harrell’s brain activity as he attempted to speak a sentence shown on a screen. The electrical spikes were filtered to remove noise in one millisecond segments and fed into a decoder. Like the Rosetta Stone, the algorithm mapped specific neural features to words and pitch, which were played back to Harrell through a voice synthesizer with just a 25-millisecond lag—roughly the time it takes for a person to hear their own voice, wrote the team.
Rather than decoding phonemes or words, the AI captured Harrell’s intent to make sounds every 10 milliseconds, allowing him to eventually say words not in a dictionary, like “hmm” or “eww.” He could spell out words and respond to open-ended questions, telling the researchers that the synthetic voice made him “happy” and that it felt like “his real voice.”
The team also recorded brain activity as Harrell attempted to speak the same set of sentences as either statements or questions, the latter having an increased pitch. All four electrode arrays recorded a neural fingerprint of activity patterns when the sentence was spoken as a question.
The system, once trained, could also detect emphasis. Harrell was asked to stress each word individually in the sentence, “I never said she stole my money,” which can have multiple meanings. His brain activity ramped up before saying the emphasized word, which the algorithm captured and used to guide the synthesized voice. In another test, the system picked up multiple pitches as he tried to sing different melodies.
Raise Your VoiceThe AI isn’t perfect. Volunteers could understand the output roughly 60 percent of the time—a far cry from the near perfect brain-to-text system Harrell is currently using. But the new AI brings individual personality to synthesized speech, which usually produces a monotone voice. Deciphering speech in real-time also lets the person interrupt or object during a conversation, making the experience feel more natural.
“We don’t always use words to communicate what we want. We have interjections. We have other expressive vocalizations that are not in the vocabulary,” study author Maitreyee Wairagkar told Nature.
Because the AI is trained on sounds, not English vocabulary, it could be adapted to other languages, especially tonal ones like Chinese. The team is also looking to increase the system’s accuracy by placing more electrodes in people who have lost their speech due to stroke or neurodegenerative diseases.
“The results of this research provide hope for people who want to talk but can’t…This kind of technology could be transformative for people living with paralysis,” said study author David Brandman.
The post A Man With ALS Can Speak and Sing Again Thanks to a Brain Implant and AI-Synthesized Voice appeared first on SingularityHub.
Fervo is using technology from the oil and gas industry to unlock vast stores of geothermal power under our feet.
Between power-hungry AI data centers, domestic manufacturing growth, and electric vehicles, US electricity demand is set to soar in coming years, and utilities aren’t yet sure where the supply to meet this growth will come from. Geothermal power is increasingly looking like a viable option thanks to companies deploying next-generation technologies.
One of these is Fervo Energy, which announced $206 million in funding this week, adding to the $255 million they secured earlier this year. The new funding round was led by Breakthrough Energy Catalyst, part of Bill Gates’ climate investment firm Breakthrough Energy Ventures.
Fervo’s approach, which uses technologies developed for the oil and gas industry, could help push geothermal’s share of total US electricity supply from its current 0.4 percent to 10 percent or greater.
Vertical Drilling for WaterConventional geothermal works by drilling vertical wells into underground reservoirs of hot water or steam. Wells are up to 10,000 feet (or about 3 kilometers/1.9 miles) deep—and those are the easy ones. The hot water accessed through vertical wells is brought to the surface, where it’s turned into steam that’s used to spin turbines.
A major advantage of geothermal over solar and wind is that it’s not limited by intermittency; the rocks in the Earth’s crust are hot 24/7. This means geothermal is a reliable source of baseload power, and tech companies including Meta and Google have jumped on the geothermal bandwagon.
However, easily accessible underground reservoirs only exist in a handful of geologically active spots around the globe, like Iceland, Kenya, and New Zealand. These countries are positioned over sections of the Earth’s crust that have high heat flow and permeable rock relatively close to the surface, as they’re close to fault lines and areas where there’s volcanic activity.
Such areas exist in the western US as well, namely in California, Nevada, Utah, and Hawaii. In fact, the US leads the world in installed geothermal generating capacity—yet we’ve tapped less than 0.7 percent of our geothermal resources. The majority of those resources can only be accessed via enhanced geothermal technology—and that’s where Fervo comes in.
Horizontal Drilling for HeatRather than only drilling vertically to access naturally occurring reservoirs of hot water, Fervo and other enhanced geothermal companies also drill horizontally to create artificial reservoirs in hot, dry rock. After drilling vertically to depths of about 8,000 feet, they bore horizontal tunnels then pump water through them, essentially creating artificial reservoirs. Heat from the rock transfers to the water, which is brought to the surface and used to generate electricity. The water is typically recycled and pumped back into the ground again.
Besides putting more surface area in contact with geothermal fluid and maximizing heat transfer, horizontal drilling allows multiple wells to be drilled from a single surface location. This means there’s a smaller surface footprint and less impact on the environment surrounding the wells.
Horizontal drilling was developed for oil and gas production to find new fossil fuel deposits. Fervo’s cofounder, Tim Latimer, started his career in the oil and gas industry, but after a 2015 flood in his home city of Houston, he realized the urgency of the climate crisis and decided to find a way to apply fossil fuel technologies to renewable energy.
Horizontal drilling isn’t the only technology Latimer repurposed for geothermal. Fervo installs fiber-optic cables in its wells to monitor real-time data on flow, temperature, and performance. They also use an advanced drill bit technology called polycrystalline diamond compact (PDC). PDC contains lab-grown diamond, one of the hardest and most resilient materials in existence. The drill bits can cut through harder types of rock, do so faster, and go longer without wearing down. In addition, Latimer said in an interview with Time Magazine, “One of the things that we drove forward was a way of pumping fluid down while we’re drilling that cools your drilling system more efficiently than in an oil and gas operation.”
Fervo set multiple drilling performance records with its recent completion of an appraisal well in southwest Utah (part of the larger project the company will use its new funding on): The 15,765-foot-deep Sugarloaf well will reach a temperature of 520 degrees Fahrenheit and was completed in 16 drilling days. The company says that’s a 79 percent reduction in drilling time compared to the US Department of Energy baseline for ultradeep geothermal wells.
Beyond the Low-Hanging FruitFervo’s technology is making it feasible to develop geothermal power plants in areas where they wouldn’t have been possible before, mainly because the economics wouldn’t have made sense. The company plans to use the $206 million in new funding to keep building out its Cape Station plant in Beaver County, Utah. Phase I of the project plans to deliver 100 megawatts of power to the grid starting in 2026, and Phase II will add another 400 megawatts by 2028. The site has received permitting approval to expand up to two gigawatts.
Fervo ultimately has ambitions to go far beyond those two gigawatts—and the resources to do so definitely exist. A US Geological Survey assessment published last month says geothermal energy in the Great Basin alone, which spans Nevada and neighboring states, could produce electricity equal to one-tenth of the current US power supply.
“Principally, there’s virtually an unlimited amount of geothermal energy,” Latimer said. “The world is really big, and the world is really hot. We’ve got billions of years of energy under our feet. It’s all a question about how much you can access economically.”
The post Geothermal Unicorn Fervo Energy Is Building a Massive Next-Gen Plant in Utah appeared first on SingularityHub.
Our universe may have been born in a gravitational crunch that formed a very massive black hole—followed by a bounce inside it.
The Big Bang is often described as the explosive birth of the universe—a singular moment when space, time, and matter sprang into existence. But what if this was not the beginning at all? What if our universe emerged from something else—something more familiar and radical at the same time?
In a new paper, published in Physical Review D (full preprint here), my colleagues and I propose a striking alternative. Our calculations suggest the Big Bang was not the start of everything, but rather the outcome of a gravitational crunch or collapse that formed a very massive black hole—followed by a bounce inside it.
This idea, which we call the black hole universe, offers a radically different view of cosmic origins, yet it is grounded entirely in known physics and observations.
Today’s standard cosmological model, based on the Big Bang and cosmic inflation (the idea that the early universe rapidly blew up in size), has been remarkably successful in explaining the structure and evolution of the universe. But it comes at a price: It leaves some of the most fundamental questions unanswered.
For one, the Big Bang model begins with a singularity—a point of infinite density where the laws of physics break down. This is not just a technical glitch; it’s a deep theoretical problem that suggests we don’t really understand the beginning at all.
To explain the universe’s large-scale structure, physicists introduced a brief phase of rapid expansion into the early universe called cosmic inflation, powered by an unknown field with strange properties. Later, to explain the accelerating expansion observed today, they added another “mysterious” component: dark energy.
In short, the standard model of cosmology works well—but only by introducing new ingredients we have never observed directly. Meanwhile, the most basic questions remain open: Where did everything come from? Why did it begin this way? And why is the universe so flat, smooth, and large?
New ModelOur new model tackles these questions from a different angle—by looking inward instead of outward. Instead of starting with an expanding universe and trying to trace back how it began, we consider what happens when an overly dense collection of matter collapses under gravity.
This is a familiar process: Stars collapse into black holes, which are among the most well-understood objects in physics. But what happens inside a black hole, beyond the event horizon from which nothing can escape, remains a mystery.
In 1965, the British physicist Roger Penrose proved that under very general conditions, gravitational collapse must lead to a singularity. This result, extended by the late British physicist Stephen Hawking and others, underpins the idea that singularities—like the one at the Big Bang—are unavoidable.
The idea helped win Penrose a share of the 2020 Nobel prize in physics and inspired Hawking’s global bestseller A Brief History of Time: From the Big Bang to Black Holes. But there’s a caveat. These “singularity theorems” rely on “classical physics” which describes ordinary macroscopic objects. If we include the effects of quantum mechanics, which rules the tiny microcosmos of atoms and particles, as we must at extreme densities, the story may change.
In our new paper, we show that gravitational collapse does not have to end in a singularity. We find an exact analytical solution—a mathematical result with no approximations. Our math shows that as we approach the potential singularity, the size of the universe changes as a (hyperbolic) function of cosmic time.
This simple mathematical solution describes how a collapsing cloud of matter can reach a high-density state and then bounce, rebounding outward into a new expanding phase.
But why do Penrose’s theorems forbid such outcomes? It’s all down to a rule called the quantum exclusion principle, which states that no two identical particles known as fermions can occupy the same quantum state (such as angular momentum, or “spin”).
And we show that this rule prevents the particles in the collapsing matter from being squeezed indefinitely. As a result, the collapse halts and reverses. The bounce is not only possible—it’s inevitable under the right conditions.
Crucially, this bounce occurs entirely within the framework of general relativity, which applies on large scales such as stars and galaxies, combined with the basic principles of quantum mechanics—no exotic fields, extra dimensions, or speculative physics required.
What emerges on the other side of the bounce is a universe remarkably like our own. Even more surprisingly, the rebound naturally produces the two separate phases of accelerated expansion—inflation and dark energy—driven not by hypothetical fields but by the physics of the bounce itself.
Testable PredictionsOne of the strengths of this model is that it makes testable predictions. It predicts a small but non-zero amount of positive spatial curvature—meaning the universe is not exactly flat, but slightly curved, like the surface of the Earth.
This is simply a relic of the initial small over-density that triggered the collapse. If future observations, such as the ongoing Euclid mission, confirm a small positive curvature, it would be a strong hint that our universe did indeed emerge from such a bounce. It also makes predictions about the current universe’s rate of expansion, something that has already been verified.
The SpaceX Falcon 9 rocket carrying ESA’s Euclid mission on the launch pad in 2023. Image Credit: ESA, CC BY-SAThis model does more than fix technical problems with standard cosmology. It could also shed new light on other deep mysteries in our understanding of the early universe—such as the origin of supermassive black holes, the nature of dark matter, or the hierarchical formation and evolution of galaxies.
These questions will be explored by future space missions such as Arrakihs, which will study diffuse features such as stellar halos (a spherical structure of stars and globular clusters surrounding galaxies) and satellite galaxies (smaller galaxies that orbit larger ones) that are difficult to detect with traditional telescopes from Earth and will help us understand dark matter and galaxy evolution.
These phenomena might also be linked to relic compact objects—such as black holes—that formed during the collapsing phase and survived the bounce.
The black hole universe also offers a new perspective on our place in the cosmos. In this framework, our entire observable universe lies inside the interior of a black hole formed in some larger “parent” universe.
We are not special, no more than Earth was in the geocentric worldview that led Galileo (the astronomer who suggested the Earth revolves around the sun in the 16th and 17th centuries) to be placed under house arrest.
We are not witnessing the birth of everything from nothing, but rather the continuation of a cosmic cycle—one shaped by gravity, quantum mechanics, and the deep interconnections between them.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
The post What If the Big Bang Wasn’t the Beginning? Research Suggests It May Have Taken Place Inside a Black Hole appeared first on SingularityHub.
Such devices could monitor viruses or biodiversity, but the potential for misuse raises ethical questions.
A majestic bobcat sauntered through the Florida coastal forests. Nearby, diamondback rattlesnakes slithered across muddy terrain, alligator “swamp puppies” patrolled the waters, and venomous spiders waited for prey. Meanwhile, trekkers explored the grand oaks, slapped away mosquitos, and looked for bats and ospreys.
This may sound like an episode of Planet Earth—but there were no cameras. Instead, scientists collected microscopic snippets of airborne DNA with a vacuum. They documented the animals by running this environmental DNA, or eDNA, through a cutting-edge device about the size of a deck of cards. The device can do more. Halfway around the world in the city of Dublin—known for its pubs, music, and cheer—the team used it to detect DNA traces from weed, poppy, and magic mushrooms wafting on the breeze. They assembled genomic profiles with astounding speed, capturing whole genetic landscapes in just two days.
“The level of information that’s available in environmental DNA is such that we’re only starting to consider what the potential applications can be, from humans to wildlife to other species that have implications for human health,” said study author David Duffy at the University of Florida in a press release.
The device is a powerful tool that can be used to monitor biodiversity, emerging viruses, and illicit drugs, but it can also detect the genetic heritage of people traipsing about nearby. Although it wasn’t used to identify individuals in the study, the authors warned that airborne eDNA “could provide seriously powerful potential for individual-level surveillance for…humans.”
Nevertheless, “It is boundary-pushing work,” Ryan Kelly at the University of Washington, who was not involved in the study, told Science.
A Trove of DataLiving creatures shed genetic material. Fungi, plants, animals, humans, bacteria, viruses—all leave invisible genetic fingerprints as they roam the world.
As technologies to read DNA—known as genetic sequencing—have advanced, scientists have begun capturing DNA in the ambient environment to take a census of the living creatures there.
Some have found thousands of bacterial species in the depths of our oceans. Others are tracking ocean species using DNA “sponges” or land-based creatures by analyzing ingested eDNA from dung beetles. These studies can also monitor emerging viruses from animals—such as those in wildlife markets—by capturing and analyzing genetic molecules.
Duffy believes eDNA could invigorate conservation efforts. In 2022, his team devised a way to monitor endangered sea turtles on the Florida coast. These animals are difficult to track. They roam multiple habitats, including the open sea, coastal ecosystems, and beaches.
Though originally developed to track microbes, Duffy and team showed eDNA can be used to detect small chunks of genetic material from hair, skin, scales, and fluids left behind in sand and water. The team also picked up dangerous sea-turtle pathogens, including a virus that causes tumors in the turtles. Since then, they’ve captured human eDNA from oceans, rivers, and sand—and can identify individual volunteers based on their footprints on the beach.
Although eDNA samples are usually picked up from water and land, they also float in the breeze. This led the team to ask: How much information can we gather from air?
Bring in the ShotgunMost eDNA studies use a technology called metabarcoding. Here, scientists extract DNA from a sample—say, water from a Florida swamp or a Dublin pub—and sequence the DNA. To detect which species are present, each DNA snippet is matched to a barcode in a data library. The method can be accurate, but it has some shortcomings.
For one, the approach can only identify eDNA sequences already in the database. The barcodes are a little like those on produce at the grocery, only instead of apples or onions, they’re small snippets of DNA unique to a species. You can only detect organisms with existing primers. That is, when the system scans a piece of DNA, it won’t register unless there’s already a barcode present. The method is also costly and takes days, if not weeks, to process a single sample.
Duffy and team turned to a method called shotgun sequencing, which randomly chops DNA sequences into billions of snippets called “reads.” Though the approach is powerful, it’s languished in the past due to the cost and time to piece together individual genetic snippets and match them to a group of organisms. The hardware was also bulky, roughly the size of a refrigerator, making it difficult to bring into the field. It was mostly used to study microbes—not animals or humans.
Thanks to cloud computing and deep sequencing—a type of DNA sequencing where the same DNA region is read many times—it’s now possible to do shotgun sequencing in the wild.
The team used a handheld device with a vacuum tube to suck DNA from the air. For two years, they collected samples across a range of urban and rural locations and produced 78 shotgun sequencing datasets.
“When we started, it seemed like it would be hard to get intact large fragments of DNA from the air. But that’s not the case. We’re actually finding a lot of informative DNA,” said Duffy.
In one experiment, they tracked bobcats by gathering eDNA near animal tracks for a week. They found it contained bobcat DNA from a wild population and a zoo-based one, suggesting the tech could be used to monitor animal lineages. They also collected airborne DNA near venomous spiders and found their genomes differ from those in the Caribbean or South America. Without having to lay eyes on the animals, the team painted a picture of species thriving in Florida’s coastal forests.
Meanwhile, Dublin had a completely different eDNA profile. The device identified 63 viruses in air samples across the city alongside a slew of allergens, such as those from peanuts and tree pollen. It also found evidence of illicit drugs, including magic mushrooms.
A Genetic QuandaryThe technology isn’t an all-seeing eye, and it’s possible to over-interpret results.
It relies on algorithms to stitch DNA back together and some could just be random DNA floating in the air. Also, some applications, like those related to human DNA, could be beneficial but also risk unexpected negative consequences, wrote the authors. In Florida and Dublin, they could identify the genetic ancestry of people walking through a location. The team intentionally refrained from identifying individual people in the study—although it has already been done.
“As with artificial intelligence technologies, the human eDNA genie cannot be returned to the bottle,” wrote the team. The technology can be used for good or nefarious purposes. For now, the team is hoping to bring eDNA back to its roots, to save and conserve wildlife.
“It seems like science fiction, but it’s becoming science fact,” said Duffy.
The post Handheld Device Creates Genetic Profiles From Airborne DNA With Astounding Speed appeared first on SingularityHub.
ARTIFICIAL INTELLIGENCEAt Secret Math Meeting, Researchers Struggle to Outsmart AILyndie Chiou | Scientific American
“After throwing professor-level questions at the bot for two days, the researchers were stunned to discover it was capable of answering some of the world’s hardest solvable problems. ‘I have colleagues who literally said these models are approaching mathematical genius,’ says Ken Ono, a mathematician at the University of Virginia and a leader and judge at the meeting.”
RoboticsAmazon Prepares to Test Humanoid Robots for Delivering PackagesRocket Drew | The Information
“Amazon is developing software for humanoid robots that could eventually take the jobs of delivery workers, according to a person who has been involved in the effort. In doing so, Amazon is paving the way to automate a major part of its operation, the delivery of parcels around the world.”
BiotechnologyBreakthrough in Search for HIV Cure Leaves Researchers ‘Overwhelmed’Kay Lay | The Guardian
“The virus’s ability to conceal itself inside certain white blood cells has been one of the main challenges for scientists looking for a cure. …Now researchers from the Peter Doherty Institute for Infection and Immunity in Melbourne, have demonstrated a way to make the virus visible, paving the way to fully clear it from the body.”
BiotechnologyFrom No Hope to a Potential Cure for a Deadly Blood CancerGina Kolata | The New York Times
“Multiple myeloma is considered incurable, but a third of patients in a Johnson & Johnson clinical trial have lived without detectable cancer for years after facing certain death. …These results, in patients whose situation had seemed hopeless, has led some battle-worn American oncologists to dare to say the words ‘potential cure.'”
RoboticsWaymo Is Winning in San FranciscoMark Sullivan | Fast Company
“The self-driving car service Waymo has been active in San Francisco for 20 months and has already captured 27% of the city’s rideshare market, according to new research compiled by Mary Meeker’s Bond venture capital firm. That rapid progress suggests the mainstreaming of self-driving car service could happen faster than once thought.”
RoboticsIt’s Waymo’s World. We’re All Just Riding in It.Ben Cohen | The Wall Street Journal
“[Waymo] cracked a million total paid rides in late 2023. By the end of 2024, it reached five million. We’re not even halfway through 2025 and it has already crossed a cumulative 10 million. At this rate, Waymo is on track to double again and blow past 20 million fully autonomous trips by the end of the year. ‘This is what exponential scaling looks like,’”’ said Dmitri Dolgov, Waymo’s co-chief executive, at Google’s recent developer conference.”
FutureWhy Eric Schmidt, Jeff Bezos and Startups Are High On Space Data CentersEvan Robinson-Johnson | The Information
“Orbital data centers have a tantalizing answer to the power problem: uninterrupted access to solar energy, without the hindrances of weather, night time and seasons. That means they could enjoy dramatically lower operating costs while also combatting climate change by reducing the reliance on fossil fuels, which currently account for 60% of total US energy consumption. That’s the theory, at least.”
FutureThe Rise of ‘Vibe Hacking’ Is the Next AI NightmareMatthew Gault | Wired
“In the near future one hacker may be able to unleash 20 zero-day attacks on different systems across the world all at once. Polymorphic malware could rampage across a codebase, using a bespoke generative AI system to rewrite itself as it learns and adapts. Armies of script kiddies could use purpose-built LLMs to unleash a torrent of malicious code at the push of a button.”
SpaceA Japanese Lander Crashed on the Moon After Losing Track of Its LocationStephen Clark | Ars Technica
“Ground teams at ispace’s mission control center in Tokyo lost contact with the Resilience lunar lander moments before it was supposed to touch down in a region called Mare Frigoris, or the Sea of Cold, a basaltic plain in the Moon’s northern hemisphere. A few hours later, ispace officials confirmed what many observers suspected. The mission was lost. It’s the second time ispace has failed to land on the Moon in as many tries.”
Artificial IntelligenceManus Has Kick-Started an AI Agent Boom in ChinaCaiwei Chen | MIT Technology Review
“Startups like Manus, Genspark, and Flowith—though founded by Chinese entrepreneurs—could blend seamlessly into the global tech scene and compete effectively abroad. Founders, investors, and analysts that MIT Technology Review has spoken to believe Chinese companies are moving fast, executing well, and quickly coming up with new products.”
TechPerplexity Received 780 Million Queries Last Month, CEO SaysAisha Malik | TechCrunch
“‘Give it a year, we’ll be doing, like, a billion queries a week if we can sustain this growth rate,’ Srinivas said. ‘And that’s pretty impressive because the first day in 2022, we did 3,000 queries, just one single day. So from there to doing 30 million queries a day now, it’s been phenomenal growth.'”
RoboticsWalmart and Wing Expand Drone Delivery to Five More US CitiesKirsten Korosec | TechCrunch
“‘We’re decidedly out of the pilot and trial phase and into scaling up this business,’ Wing CEO Adam Woodworth told TechCrunch in a recent interview. ‘We’ve always been the type of company that wants to do something well and stay focused. And so this is the next big bite at the apple. It’s a much bigger bite than than we’ve taken before.'”
FutureNo, AI Robots Won’t Take All Our JobsRobert D. Atkinson | The Wall Street Journal
“Anthropic CEO Dario Amodei said last week that artificial intelligence could eliminate half of all entry-level white-collar jobs within five years and cause unemployment to skyrocket to as high as 20%. He should know better—as should many other serious academics, who have been warning for years that AI will mean the end of employment as we know it.”
FutureGoogle DeepMind’s CEO Thinks AI Will Make Humans Less SelfishSteven Levy | Wired
“When I spoke to Hassabis at Google’s New York City headquarters, his answers came as quickly as a chatbot’s, crisply parrying every inquiry I could muster with high spirits and a confidence that he and Google are on the right path. …You may not always agree with what Hassabis has to say, but his thoughts and his next moves matter. History, after all, will be written by the winners.”
SpaceAn In-Space Propulsion Company Just Raised a Staggering Amount of MoneyEric Berger | Ars Technica
“This week an in-space propulsion company, Impulse Space, announced that it had raised a significant amount of money, $300 million. This follows a fundraising round just last year in which the Southern California-based company raised $150 million. This is one of the largest capital raises in space in a while, especially for a non-launch company.”
The post This Week’s Awesome Tech Stories From Around the Web (Through June 7) appeared first on SingularityHub.
The brain quickly adapts to change by predicting multiple futures, neuron by neuron. These finding could lead to AI that can do the same thing.
We constantly make decisions. Some seem simple: I booked dinner at a new restaurant, but I’m hungry now. Should I grab a snack and risk losing my appetite or wait until later for a satisfying meal—in other words, what choice is likely more rewarding?
Dopamine neurons inside the brain track these decisions and their outcomes. If you regret a choice, you’ll likely make a different one next time. This is called reinforcement learning, and it helps the brain continuously adjust to change. It also powers a family of AI algorithms that learn from successes and mistakes like humans do.
But reward isn’t all or nothing. Did my choice make me ecstatic, or just a little happier? Was the wait worth it?
This week, researchers at the Champalimaud Foundation, Harvard University, and other institutions said they’ve discovered a previously hidden universe of dopamine signaling in the brain. After recording the activity of single dopamine neurons as mice learned a new task, the teams found the cells don’t simply track rewards. They also keep tabs on when a reward came and how big it was—essentially building a mental map of near-term and far-future reward possibilities.
“Previous studies usually just averaged the activity across neurons and looked at that average,” said study author Margarida Sousa in a press release. “But we wanted to capture the full diversity across the population—to see how individual neurons might specialize and contribute to a broader, collective representation.”
Some dopamine neurons preferred immediate rewards; others slowly ramped up activity in expectation of delayed satisfaction. Each cell also had a preference for the size of a reward and listened out for internal signals—for example, if a mouse was thirsty, hungry, and its motivation level.
Surprisingly, this multidimensional map closely mimics some emerging AI systems that rely on reinforcement learning. Rather than averaging different opinions into a single decision, some AI systems use a group of algorithms that encodes a wide range of reward possibilities and then votes on a final decision.
In several simulations, AI equipped with a multidimensional map better handled uncertainty and risk in a foraging task.
The results “open new avenues” to design more efficient reinforcement learning AI that better predicts and adapts to uncertainties, wrote one team. They also provide a new way to understand how our brains make everyday decisions and may offer insight into how to treat impulsivity in neurological disorders such as Parkinson’s disease.
Dopamine SparkFor decades, neuroscientists have known dopamine neurons underpin reinforcement learning. These neurons puff out a small amount of dopamine—often dubbed the pleasure chemical—to signal an unexpected reward. Through trial and error, these signals might eventually steer a thirsty mouse through a maze to find the water stashed at its end. Scientists have developed a framework for reinforcement learning by recording the electrical activity of dopamine neurons as these critters learned. Dopamine neurons spark with activity in response to nearby rewards, then this activity slowly fades as time goes by—a process researchers call “discounting.”
But these analyses average activity into a single expected reward, rather than capturing the full range of possible outcomes over time—such as larger rewards after longer delays. Although the models can tell you if you’ve received a reward, they miss nuances, such as when and how much. After battling hunger—was the wait for the restaurant worth it?
An Unexpected HintSousa and colleagues wondered if dopamine signaling is more complex than previously thought. Their new study was actually inspired by AI. An approach called distributional reinforcement learning estimates a range of possibilities and learns from trial and error rather than a single reward.
“What if different dopamine neurons were sensitive to distinct combinations of possible future reward features—for example, not just their magnitude, but also their timing?” said Sousa.
Harvard neuroscientists led by Naoshige Uchida had an answer. They recorded electrical activity from individual dopamine neurons in mice as the animals learned to lick up a water reward. At the beginning of each trial, the mice sniffed a different scent that predicted both the amount of water they might find—that is, the size of the reward—and how long until they might get it.
Each dopamine neuron had its own preference. Some were more impulsive and preferred immediate rewards, regardless of size. Others were more cautious, slowly ramping up activity that tracked reward over time. It’s a bit like being extremely thirsty on a hike in the desert with limited water: Do you chug it all now, or ration it out and give yourself a longer runway?
The neurons also had different personalities. Optimistic ones were especially sensitive to unexpectedly large rewards—activating with a burst—whereas pessimistic ones stayed silent. Combining the activity of these neuron voters, each with their own point of view, resulted in a population code that ultimately decided the mice’s behavior.
“It’s like having a team of advisors with different risk profiles,” said study author Daniel McNamee in the press release, “Some urge action—‘Take the reward now, it might not last’—while others advise patience—‘Wait, something better could be coming.’”
Each neuron’s stance was flexible. When the reward was consistently delayed, they collectively shifted to favor longer-term rewards, showcasing how the brain rapidly adjusts to change.
“When we looked at the [dopamine neuron] population as a whole, it became clear that these neurons were encoding a probabilistic map,” said study author Joe Paton. “Not just whether a reward was likely, but a coordinate system of when it might arrive and how big it might be.”
Brain to AIThe brain recordings were like ensemble AI, where each model has its own viewpoint but the group collaborates to handle uncertainties.
The team also developed an algorithm, called time-magnitude reinforcement learning, or TMRL, that could plan future choices. Classic reinforcement-learning models only give out rewards at the end. This takes many cycles of learning before an algorithm homes in on the best decision. But TMRL rapidly maps a slew of choices, allowing humans and AI to pick the best ones with fewer cycles. The new model also includes internal states, like hunger levels, to further fine-tune decisions.
In one test, equipping algorithms with a dopamine-like “multidimensional map” boosted their performance in a simulated foraging task compared to standard reinforcement learning models.
“Knowing in advance—at the start of an episode—the range and likelihood of rewards available and when they are likely to occur could be highly useful for planning and flexible behavior,” especially in a complex environment and with different internal states, wrote Sousa and team.
The dual studies are the latest to showcase the power of AI and neuroscience collaboration. Models of the brain’s inner workings can inspire more human-like AI. Meanwhile, AI is shining light into our own neural machinery, potentially leading to insights about neurological disorders.
Inspiration from the brain “could be key to developing machines that reason more like humans,” said Paton.
The post This Brain Discovery Could Unlock AI’s Ability to See the Future appeared first on SingularityHub.
Artificial IntelligenceSam Altman Lays Out Roadmap for OpenAI’s Long-Awaited GPT-5 ModelBenj Edwards | Ars Technica
“On Wednesday, OpenAI CEO Sam Altman announced a roadmap for how the company plans to release GPT-5, the long-awaited followup to 2023’s GPT-4 AI language model that made huge waves in both tech and policy circles around the world. In a reply to a question on X, Altman said GPT-5 would be coming in ‘months,’ suggesting a release later in 2025.”
RoboticsChina’s EV Giants Are Betting Big on Humanoid RobotsCaiwei Chen | MIT Technology Review
“It’s becoming clear that China is now committed to becoming a global leader in robotics and automation, just as it did with EVs. Wang Xingxing, the CEO of Unitree Robots, said this well in a recent interview to local media: ‘Robotics is where EVs were a decade ago—a trillion-yuan battlefield waiting to be claimed.'”
Artificial IntelligenceAnthropic Strikes BackStephanie Palazzolo | The Information
“[Anthropic] has developed a hybrid AI model that includes reasoning capabilities, which basically means the model uses more computational resources to calculate answers to hard questions. But the model can also handle simpler tasks quickly, without the extra work, by acting like a traditional large language model. The company plans to release it in the coming weeks, according to a person who’s used it.”
BiotechnologyAI Used to Design a Multi-Step Enzyme That Can Digest Some PlasticsJohn Timmer | Ars Technica
“Unfortunately, there isn’t an enzyme for many reactions we would sorely like to catalyze—things like digesting plastics or incorporating carbon dioxide into more complex molecules. …With the advent of AI-driven protein design, however, we can now potentially design things that are unlike anything found in nature.”
ComputingThis DARPA-Backed Startup Banked $100 Million for Its Energy-Slashing Analog ChipsAlex Pasternack | Fast Company
“EnCharge says that, for a wide range of AI use cases, its specialized chips, or accelerators, require up to 20 times less energy compared to today’s leading AI chips. …Rather than using only digital transistors to perform some of the multiplication operations at the heart of AI inference—the continuous computations that produce chatbot outputs—EnCharge’s chips exploit the non-binary wonders of the analog world.”
TechWill We Get a $1 Trillion Private Tech Firm?Cory Weinberg | The Information
“Will a private tech company reach a $1 trillion valuation in the coming years? It’s not a ridiculous question. A couple of companies seem like potential candidates. OpenAI is closing in on $300 billion in its financing with SoftBank, and SpaceX recently shot to $350 billion.”
RoboticsMeta’s Next Big Bet Might Be AI Humanoid Robots for At-Home ChoresNadeem Sarwar | Digital Trends
“[Meta’s] interests have swayed wildly over the past few years. Phones, crypto, tablets, metaverse, smart glasses, and finally, AI. The next avenue for Meta is apparently humanoid robots. According to Bloomberg, the company is pouring resources into the development of AI-powered humanoid robots. ‘Meta plans to work on its own humanoid robot hardware, with an initial focus on household chores,’ says the report.”
FUTUREMotor Neuron Diseases Took Their Voices. AI Is Bringing Them Back.Jessica Hamzelou | MIT Technology Review
“Rodriguez and his wife, Maria Fernandez, who live in Miami, thought they would never hear his voice again. Then they re-created it using AI. After feeding old recordings of Rodriguez’s voice into a tool trained on voices from film, television, radio, and podcasts, the couple were able to generate a voice clone—a way for Jules to communicate in his ‘old voice.'”
ComputingThis Breakthrough Holographic Display Could Make AR Glasses a Reality in 2026Alan Truly | Digital Trends
“Swave CEO Mike Noonen told me the bill of materials (BOM) is just $50 per eye and the expected weight of AR glasses using HXR technology could be less than 50 grams. The FoV and apparent resolution are tunable with a view as wide as 120 degrees and a retina-like resolution of up to 60 pixels per degree (PPD). Battery life is estimated at more than 10 hours, making these suitable for daily wear.”
TechThomson Reuters Wins First Major AI Copyright Case in the USKate Knibbs | Wired
“This ruling is a blow to AI companies, according to Cornell University professor of digital and internet law James Grimmelmann: ‘If this decision is followed elsewhere, it’s really bad for the generative AI companies.’ Grimmelmann believes that Bibas’ judgement suggests that much of the case law that generative AI companies are citing to argue fair use is ‘irrelevant.'”
SpaceThe Dream of Offshore Rocket Launches Is Finally Blasting OffBecky Ferreira | MIT Technology Review
“’The best way to build a future where we have dozens, hundreds, or maybe thousands of spaceports is to build them at sea,’ says Tom Marotta, CEO and founder of the Spaceport Company, which is working to establish offshore launch hubs. ‘It’s very hard to find a thousand acres on the coast over and over again to build spaceports. It’s very easy to build the same ship over and over again.'”
FutureWho’s Using AI the Most? The Anthropic Economic Index Breaks Down the DataMichael Nuñez | VentureBeat
“The Anthropic Economic Index, released today, provides a detailed analysis of AI usage across industries, drawing from millions of anonymized conversations with Claude, Anthropic’s AI assistant. The report finds that while AI is not yet broadly automating entire jobs, it is being widely used to augment specific tasks—especially in software development, technical writing and business analysis.”
Artificial IntelligenceChatGPT, Can You Write My New Novel for Me? Och Aye, Ye Preenin’ SassenachGareth Rubin | The Guardian
“The monsters of artificial intelligence are coming for you. They will cast you out on the street like a Dickensian mill owner and laugh as they do it—at least they will if you work in any sort of creative industry. …Well, I’m going to turn the tables. My publisher is anxiously waiting for me to finish my new novel, a sequel to my previous thriller The Turnglass. So let’s see if AI can take the faff—the actual writing bit—out of creative writing.”
The post This Week’s Awesome Tech Stories From Around the Web (Through February 15) appeared first on SingularityHub.
The new tool, NanoCas, could extend gene therapies throughout the body.
When the gene editing tool CRISPR-Cas9 rocketed to fame more than a decade ago, it transformed biotechnology. Faster, cheaper, and safer than previous methods, the tool helped scientists gain insight into gene functions—and when they go wrong.
CRISPR also brought the potential to change the lives of people living with inherited diseases. Thanks to its gene editing prowess, the tool can supercharge immune cells’ ability to hunt down cancer and other rogue cells. In late 2023, the FDA approved a CRISPR therapy for sickle cell disease and later gave the greenlight to people with a blood disorder called transfusion-dependent beta thalassemia. Many more therapies are in the works.
But CRISPR has a hefty problem: The system is too large, making it difficult to deliver the gene editor to cells in muscle, brain, heart, and other tissues.
Now, a team at Mammoth Biosciences has a potential solution. Cofounded by CRISPR pioneer Jennifer Doudna at the University of California, Berkeley, the company has long sought to downsize the original CRISPR-Cas9 system. Their new iteration, dubbed NanoCas, slashed the size of one key component, Cas9, to roughly one-third of the original.
The slimmed-down setup allowed the tool to be packaged into a single “delivery box”—a virus that’s commonly used for gene therapy inside the body. In mice and monkeys, the team used NanoCas to edit genes involved in inherited high cholesterol and Duchenne muscular dystrophy.
“CRISPR gene editing is a transformative technology for addressing genetic diseases, but delivery constraints have largely limited its therapeutic applications to liver-targeted and ex vivo [outside the body] therapies,” wrote the team in a preprint describing their results. The compact NanoCas “opens the door” for editing tissues inside the body.
Delivery WoesCRISPR has two main components. One is an RNA molecule that’s like a bloodhound, seeking out and tethering the setup to a target DNA section. Once docked, the second component, a Cas protein, slices or snips the genetic ribbon.
Over the years, scientists have discovered or engineered other versions of Cas proteins. Some target RNA, the “messenger” that translates genes into proteins. Others swap out single genetic letters causing inherited diseases. Some even recruit enzymes to modify the epigenome—the system controlling which genes are turned on or off.
All these tools have a major problem: They’re difficult to deliver inside the body because of their size. Current CRISPR therapies mainly rely on extracting cells and swapping their genes inside petri dishes. The edited cells are infused back into the patient. Called “ex vivo” therapy, these treatments mainly focus on blood-based disorders.
Correcting genetic problems inside the body with CRISPR adds to the complexity. Most therapies focus on the eyes or the liver, which are both relatively easy to access with a shot. For all other tissues, delivery is the main problem.
To shuttle the editors to tissues and cells, they have to be packaged inside a virus or a fatty bubble. Cas proteins can reach over a thousand amino acids in length, which already stresses the capacity of the delivery vehicles. Add in guide RNA components, and the system exceeds luggage limits.
To get around weight restrictions, scientists have encoded the guide RNA and Cas components separately into two viral carriers, so both can sneak into cells. Alternatively, they’ve used fatty bubbles called liposomes that encapsulate both gene editing components.
Neither is perfect. A double load of virus increases the risk of an immune response. Liposomes generally end up in the liver and release their cargo there. This makes them excellent at editing genes in the liver—for example, PCSK9,to treat high levels of cholesterol—but they struggle to reach other tissues. Important targets such as the brain and muscles are out of reach.
Small But MightyWhy not shrink the cargo so it fits into the same viral luggage?
Here, Mammoth Biosciences searched metagenomics databases for smaller Cas proteins. These databases contain diverse samples from across the planet, including from microbes gathered in swamps, seawater, our guts, and other sources. The team looked for systems that could edit as efficiently as Cas9, required only a tiny guide RNA component, and were under 600 amino acids.
From 22,000 metagenomes, the team zeroed in on 176 candidates. Each was vetted in human kidney cells in a dish—rather than using bacteria, which is the norm. This screens for Cas variants that work well inside mammalian cells, which is a common bottleneck, wrote the team.
After more tests, they landed on NanoCas. It worked with roughly 60 percent of the RNA guides they tried out, and after some tinkering, easily sliced up targeted DNA.
The tiny editor and its guide RNA fit into a single viral vector. As proof of concept, the team made a NanoCas system targeting PCSK9, a gene associated with dangerously high levels of cholesterol, in the livers of mice. Delivered in a single injection into the veins, the tiny tool slashed the gene to undetectable levels in the blood.
Next, the team turned to a gene called dystrophin in muscles, a tissue traditional CRISPR methods struggle to reach. In Duchenne muscular dystrophy, mutated dystrophin causes progressive muscle loss. NanoCas edited the gene across a wide variety of muscle types—thigh, heart, and calf muscle. The efficacy varied, ranging from 10 to 40 percent of edited cells.
The team next tested NanoCas in monkeys. After about two months, roughly 30 percent of their skeletal muscle cells were edited. Heart cells were less responsive, with only half the efficacy.
“To our knowledge,” this is the first time someone has edited muscles in a non-human primate with a single virus CRISPR system, wrote the team.
Gene therapies using delivery viruses can tax the liver, but throughout the trial the monkey’s liver functions and other health factors stayed relatively normal. But many questions remain. Although the system edited targeted genes in healthy monkeys, whether it can treat genetic muscle loss remains to be seen. As with other gene editing systems, there’s also the risk of unintentionally editing non-targeted genes or spurring an immune attack.
That said, the miniature NanoCas—and potentially other tiny Cas proteins yet to be discovered—could shuttle CRISPR to a variety of tissues in the body with a jab. The team is already exploring the system’s potential for targeting brain diseases. The technology could also be reworked for use in epigenetic or base editing.
Above all, the study suggests small Cas proteins can be mighty.
“NanoCas demonstrates that carefully selected compact systems can achieve robust editing across various contexts, challenging the assumption that small CRISPR systems are inherently less effective,” wrote the team.
The post Miniaturized CRISPR Packs a Mighty Gene Editing Punch appeared first on SingularityHub.
Inefficient AI models guzzle energy. Then again, so do efficient ones—just for different reasons.
DeepSeek has upended the AI industry, from the chips and money needed to train and run AI to the energy it’s expected to guzzle in the not-too-distant future. Energy stocks skyrocketed in 2024 on predictions of dramatic growth in electricity demand to power AI data centers, with shares of power generation companies Constellation Energy and Vistra reaching record highs.
And that wasn’t all. In one of the biggest deals in the US power industry’s history, Constellation acquired natural gas producer Calpine Energy for $16.4 billion, assuming demand for gas would grow as a generation source for AI. Meanwhile, nuclear power seemed poised for a renaissance. Google signed an agreement with Kairos Power to buy nuclear energy produced by small modular reactors (SMRs). Separately, Amazon made deals with three different SMR developers, and Microsoft and Constellation announced they would restart a reactor at Three Mile Island.
As this frenzy to secure reliable baseload power built towards a crescendo, DeepSeek’s R1 came along and unceremoniously crashed the party. Its creators say they trained the model using a fraction of the hardware and computing power of its predecessors. Energy stocks tumbled and shock waves reverberated through the energy and AI communities, as it suddenly seemed like all that effort to lock in new power sources was for naught.
But was such a dramatic market shake-up merited? What does DeepSeek really mean for the future of energy demand?
At this point, it’s too soon to draw definitive conclusions. However, various signs suggest the market’s knee-jerk response to DeepSeek was more reactionary than an accurate indicator of how R1 will impact energy demand.
Training vs. InferenceDeepSeek claimed it spent just $6 million to train its R1 model and used fewer (and less sophisticated) chips than the likes of OpenAI. There’s been much debate about what exactly these figures mean. The model does appear to include real improvements, but the associated costs may be higher than disclosed.
Even so, R1’s advances were enough to rattle markets. To see why, it’s worth digging into the nuts and bolts a bit.
First of all, it’s important to note that training a large language model is entirely different than using that same model to answer questions or generate content. Initially, training an AI is the process of feeding it massive amounts of data that it uses to learn patterns, draw connections, and establish relationships. This is called pre-training. In post-training, more data and feedback are used to fine-tune the model, often with humans in the loop.
Once a model has been trained, it can be put to the test. This phase is called inference, when the AI answers questions, solves problems, or writes text or code based on a prompt.
Traditionally with AI models, a huge amount of resources goes into training them up front, but relatively fewer resources go towards running them (at least on a per-query basis). DeepSeek did find ways to train its model far more efficiently, both in pre-training and post-training. Advances included clever engineering hacks and new training techniques—like the automation of reinforcement feedback usually handled by people—that impressed experts. This led many to question whether companies would actually need to spend so much building enormous data centers that would gobble up energy.
It’s Costly to ReasonDeepSeek is a new kind of model called a “reasoning” model. Reasoning models begin with a pre-trained model, like GPT-4, and receive further training where they learn to employ “chain-of-thought reasoning” to break a task down into multiple steps. During inference, they test different formulas for getting a correct answer, recognize when they make a mistake, and improve their outputs. It’s a little closer to how humans think—and it takes a lot more time and energy.
In the past, training used the most computing power and thus the most energy, as it entailed processing huge datasets. But once a trained model reached inference, it was simply applying its learned patterns to new data points, which didn’t require as much computing power (relatively).
To an extent, DeepSeek’s R1 reverses this equation. The company made training more efficient, but the way it solves queries and answers prompts guzzles more power than older models. A head-to-head comparison found that DeepSeek used 87 percent more energy than Meta’s non-reasoning Llama 3.3 to answer the same set of prompts. Also, OpenAI—whose o1 model was first out of the gate with reasoning capabilities—found allowing these models more time to “think” results in better answers.
Although reasoning models aren’t necessarily better for everything—they excel at math and coding, for example—their rise may catalyze a shift toward more energy-intensive uses. Even if training models gets more efficient, added computation during inference may cancel out some of the gains.
Assuming that greater efficiency in training will lead to less energy use may not pan out either. Counter-intuitively, greater efficiency and cost-savings in training may simply mean companies go even bigger during that phase, using just as much (or more) energy to get better results.
“The gains in cost efficiency end up entirely devoted to training smarter models, limited only by the company’s financial resources,” wrote Anthropic cofounder Dario Amodei of DeepSeek.
If It Costs Less, We Use MoreMicrosoft CEO Satya Nadella likewise brought up this tendency, known as the Jevons paradox—the idea that increased efficiency leads to increased use of a resource, ultimately canceling out the efficiency gain—in response to the DeepSeek melee.
If your new car uses half as much gas per mile as your old car, you’re not going to buy less gas; you’re going to take that road trip you’ve been thinking about, and plan another road trip to boot.
The same principle will apply in AI. While reasoning models are relatively energy-intensive now, they likely won’t be forever. Older AI models are vastly more efficient today than when they were first released. We’ll see the same trend with reasoning models; even though they’ll consume more energy in the short run, in the long run they’ll get more efficient. This means it’s likely that over both time frames they’ll use more energy, not less. Inefficient models will gobble up excessive energy first, then increasingly efficient models will proliferate and be used to a far greater extent later on.
As Nadella posted on X, “As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can’t get enough of.”
If You Build ItIn light of DeepSeek’s R1 mic drop, should US tech companies be backpedaling on their efforts to ramp up energy supplies? Cancel those contracts for small modular nuclear reactors?
In 2023, data centers accounted for 4.4 percent of total US electricity use. A report published in December—prior to R1’s release—predicted that figure could balloon to as much as 12 percent by 2028. That percentage could shrink due to the training efficiency improvements brought by DeepSeek, which will be widely implemented.
But given the likely proliferation of reasoning models and the energy they use for inference—not to mention later efficiency-driven demand increases—my money’s on data centers hitting that 12 percent, just as analysts predicted before they’d ever heard of DeepSeek.
Tech companies appear to be on the same page. In recent earnings calls, Google, Microsoft, Amazon, and Meta announced they would spend $300 billion—mostly on AI infrastructure—this year alone. There’s still a whole lot of cash, and energy, in AI.
The post DeepSeek Crashed Energy Stocks. Here’s Why It Shouldn’t Have. appeared first on SingularityHub.
The path to recent advanced AI systems has been more about building larger systems than making scientific breakthroughs.
For most of artificial intelligence’s history, many researchers expected that building truly capable systems would need a long series of scientific breakthroughs: revolutionary algorithms, deep insights into human cognition, or fundamental advances in our understanding of the brain. While scientific advances have played a role, recent AI progress has revealed an unexpected insight: A lot of the recent improvement in AI capabilities has come simply from scaling up existing AI systems.1
Here, scaling means deploying more computational power, using larger datasets, and building bigger models. This approach has worked surprisingly well so far.2 Just a few years ago, state-of-the-art AI systems struggled with basic tasks like counting.3,4 Today, they can solve complex math problems, write software, create extremely realistic images and videos, and discuss academic topics.
This article will provide a brief overview of scaling in AI over the past years. The data comes from Epoch, an organization that analyzes trends in computing, data, and investments to understand where AI might be headed.5 Epoch maintains the most extensive dataset on AI models and regularly publishes key figures on AI growth and change.
What Is Scaling in AI Models?Let’s briefly break down what scaling means in AI. Scaling is about increasing three main things during training, which typically need to grow together:
• The amount of data used for training the AI;
• The model’s size, measured in “parameters”;
• Computational resources, often called “compute” in AI.
The idea is simple but powerful: Bigger AI systems, trained on more data and using more computational resources, tend to perform better. Even without substantial changes to the algorithms, this approach often leads to better performance across many tasks.6
Here is another reason why this is important: As researchers scale up these AI systems, they not only improve in the tasks they were trained on but can sometimes lead them to develop new abilities that they did not have on a smaller scale.7 For example, language models initially struggled with simple arithmetic tests like three-digit addition, but larger models could handle these easily once they reached a certain size.8 The transition wasn’t a smooth, incremental improvement but a more abrupt leap in capabilities.
This abrupt jump in capability, rather than steady improvement, can be concerning. If, for example, models suddenly develop unexpected and potentially harmful behaviors simply as a result of getting bigger, it would be harder to anticipate and control.
This makes tracking these metrics important.
What Are the Three Components of Scaling Up AI models?Data: scaling up the training dataOne way to view today’s AI models is by looking at them as very sophisticated pattern recognition systems. They work by identifying and learning from statistical regularities in the text, images, or other data on which they are trained. The more data the model has access to, the more it can learn about the nuances and complexities of the knowledge domain in which it’s designed to operate.9
In 1950, Claude Shannon built one of the earliest examples of “AI”: a robotic mouse named Theseus that could “remember” its path through a maze using simple relay circuits. Each wall Theseus bumped into became a data point, allowing it to learn the correct route. The total number of walls or data points was 40. You can find this data point in the chart; it is the first one.
While Theseus stored simple binary states in relay circuits, modern AI systems utilize vast neural networks, which can learn much more complex patterns and relationships and thus process billions of data points.
All recent notable AI models—especially large, state-of-the-art ones—rely on vast amounts of training data. With the y-axis displayed on a logarithmic scale, the chart shows that the data used to train AI models has grown exponentially. From 40 data points for Theseus to trillions of data points for the largest modern systems in a little more than seven decades.
Since 2010, the training data has doubled approximately every nine to ten months. You can see this rapid growth in the chart, shown by the purple line extending from the start of 2010 to October 2024, the latest data point as I write this article.10
Datasets used for training large language models, in particular, have experienced an even faster growth rate, tripling in size each year since 2010. Large language models process text by breaking it into tokens—basic units the model can encode and understand. A token doesn’t directly correspond to one word, but on average, three English words correspond to about four tokens.
GPT-2, released in 2019, is estimated to have been trained on 4 billion tokens, roughly equivalent to 3 billion words. To put this in perspective, as of September 2024, the English Wikipedia contained around 4.6 billion words.11 In comparison, GPT-4, released in 2023, was trained on almost 13 trillion tokens, or about 9.75 trillion words.12 This means that GPT-4’s training data was equivalent to over 2,000 times the amount of text of the entire English Wikipedia.
As we use more data to train AI systems, we might eventually run out of high-quality human-generated materials like books, articles, and research papers. Some researchers predict we could exhaust useful training materials within the next few decades13. While AI models themselves can generate vast amounts of data, training AI on machine-generated materials could create problems, making the models less accurate and more repetitive.14
Parameters: scaling up the model sizeIncreasing the amount of training data lets AI models learn from much more information than ever before. However, to pick up on the patterns in this data and learn effectively, models need what are called “parameters”. Parameters are a bit like knobs that can be tweaked to improve how the model processes information and makes predictions. As the amount of training data grows, models need more capacity to capture all the details in the training data. This means larger datasets typically require the models to have more parameters to learn effectively.
Early neural networks had hundreds or thousands of parameters. With its simple maze-learning circuitry, Theseus was a model with just 40 parameters—equivalent to the number of walls it encountered. Recent large models, such as GPT-3, boast up to 175 billion parameters.15 While the raw number may seem large, this roughly translates into 700 GB if stored on a disk, which is easily manageable by today’s computers.
The chart shows how the number of parameters in AI models has skyrocketed over time. Since 2010, the number of AI model parameters has approximately doubled every year. The highest estimated number of parameters recorded by Epoch is 1.6 trillion in the QMoE model.
While bigger AI models can do more, they also face some problems. One major issue is called “overfitting.” This happens when an AI becomes “too optimized” for processing the particular data it was trained on but struggles with new data. To combat this, researchers employ two strategies: implementing specialized techniques for more generalized learning and expanding the volume and diversity of training data.
Compute: scaling up computational resourcesAs AI models grow in data and parameters, they require exponentially more computational resources. These resources, commonly referred to as “compute” in AI research, are typically measured in total floating-point operations (“FLOP”), where each FLOP represents a single arithmetic calculation like addition or multiplication.
The computational needs for AI training have changed dramatically over time. With their modest data and parameter counts, early models could be trained in hours on simple hardware. Today’s most advanced models require hundreds of days of continuous computations, even with tens of thousands of special-purpose computers.
The chart shows that the computation used to train each AI model—shown on the vertical axis—has consistently and exponentially increased over the last few decades. From 1950 to 2010, compute doubled roughly every two years. However, since 2010, this growth has accelerated dramatically, now doubling approximately every six months, with the most compute-intensive model reaching 50 billion petaFLOP as I write this article.16
To put this scale in perspective, a single high-end graphics card like the NVIDIA GeForce RTX 3090—widely used in AI research—running at full capacity for an entire year would complete just 1.1 million petaFLOP computations. 50 billion petaFLOP is approximately 45,455 times more than that.
Achieving computations on this scale requires large energy and hardware investments. Training some of the latest models has been estimated to cost up to $40 million, making it accessible only to a few well-funded organizations.
Compute, Data, and Parameters Tend to Scale at the Same TimeCompute, data, and parameters are closely interconnected when it comes to scaling AI models. When AI models are trained on more data, there are more things to learn. To deal with the increasing complexity of the data, AI models, therefore, require more parameters to learn from the various features of the data. Adding more parameters to the model means that it needs more computational resources during training.
This interdependence means that data, parameters, and compute need to grow simultaneously. Today’s largest public datasets are about 10 times bigger than what most AI models currently use, some containing hundreds of trillions of words. But without enough compute and parameters, AI models can’t yet use these for training.
What Can We Learn From These Trends for the Future of AI?Companies are seeking large financial investments to develop and scale their AI models, with a growing focus on generative AI technologies. At the same time, the key hardware that is used for training—GPUs—is getting much cheaper and more powerful, with its computing speed doubling roughly every 2.5 years per dollar spent.17 Some organizations are also now leveraging more computational resources not just in training AI models but also during inference—the phase when models generate responses—as illustrated by OpenAI’s latest o1 model.
These developments could help create more sophisticated AI technologies faster and cheaper. As companies invest more money and the necessary hardware improves, we might see significant improvements in what AI can do, including potentially unexpected new capabilities.
Because these changes could have major effects on our society, it’s important that we track and understand these developments early on. To support this, Our World in Data will update key metrics—such as the growth in computational resources, training data volumes, and model parameters—on a monthly basis. These updates will help monitor the rapid evolution of AI technologies and provide valuable insights into their trajectory.
This article was originally published on Our World in Data and has been republished here under a Creative Commons license. Read the original article.
The post Scaling Up: How Increasing Inputs Has Made Artificial Intelligence More Capable appeared first on SingularityHub.
A watery past may have spurred the formation of life’s basic molecules.
Life on Earth relies on molecular building blocks to make DNA and proteins. Scientists have long wondered how prevalent these precursors were at the birth of our solar system.
A sample of dust and rocks from an asteroid just took us closer to an answer.
Collected from Bennu, a space rock shaped like a spinning top, as it soared by Earth roughly five years ago, the samples were frozen in time by the vacuum of space. Essentially a time capsule of the earliest days of our solar system some 4.5 billion years ago—around the time when Earth was forming—they provide a peek into the chemical soup that may have kickstarted life.
Two new studies examining these extraterrestrial space grains found signs of life’s molecules preserved on the asteroid’s ancient surface. Dust and rocks from Bennu contained all five nucleobases—molecules that make up DNA and RNA—and 14 of the 20 amino acids in proteins.
These organic molecules had been found in other asteroids. But there’s a twist to Bennu’s chemical makeup. Whereas most Earthly amino acids exist in a left-handed form, samples from Bennu contain an almost equal amount of amino acids that are their mirror image. These right-handed amino acids aren’t naturally found on Earth.
Bennu also harbored telltale signs of saltwater, which could have been the soup that helped the molecules mingle and interact. The brine is similar in composition to dried lake beds on Earth.
To be clear, the teams didn’t find extraterrestrial life. But they did show that life’s precursor molecules—even “mirrored” ones—were widespread across the early solar system.
“Asteroids provide a time capsule into our home planet’s history, and Bennu’s samples are pivotal in our understanding of what ingredients in our solar system existed before life started on Earth,” said Nicky Fox, an associate administrator at NASA, in a press release.
The MissionThe samples were delivered by NASA’s OSIRIS-REx mission—the first US project to bring asteroid samples home. Bennu was an especially interesting target. Prior work had suggested asteroids have the organic molecules that form the basis of life on Earth. These molecules could have hitched a ride on asteroids and seeded the early planets or their moons to spark life.
On Earth, two critical components for life are nucleobases and amino acids.
Nucleobases are the molecular building blocks of DNA. They encode our bodies’ makeup, functions, and inheritance. RNA, which transmits the instructions contained in genes to the protein-making factories in cells, uses an additional nucleobase, which is also integral to some viruses. Beyond DNA and RNA, 20 amino acids link together to form proteins.
How these precursor ingredients spurred life remains a mystery, but asteroids may contain clues. A previous sample from 162173 Ryugu, a diamond-shaped asteroid, contained myriad organic compounds, including vitamin B3 and uracil, the additional nucleobase used in RNA.
Like Ryugu, Bennu is a carbonaceous asteroid. These space rocks are rich in carbon molecules that form the organic compounds critical for life. Bennu, a pile of rocks loosely held together by gravity, likely dates to the beginning of the solar system—some 4.5 billion years ago.
Thanks to the freezing vacuum of space, most organic molecules on Bennu have been preserved in their original state—locked in time—and could provide clues about the early solar system’s chemical makeup.
Bennu was also an attractive target because it skirts the asteroid belt, which circles the sun between Mars and Jupiter. At its closest, the asteroid is 200 million miles from Earth. While still a multi-year journey, the distance made it possible to land a space probe, map Bennu’s landscape, collect specimens, and shuttle the cargo back to Earth.
The probe was specifically designed to seal collected samples in a capsule to protect them from contamination when returning to and re-entering Earth’s atmosphere. As the capsule dropped back to Earth, the air was filtered to remove water vapor and dust particles. Upon landing in Utah, NASA immediately placed the capsule in a clean room and blasted it with nitrogen—a gas that doesn’t react with most other chemicals—to push out invading air.
“What makes these results so significant is that we’re finding them in a pristine sample,” Daniel Glavin, an astrobiologist at NASA and coauthor on a paper describing the work, told Nature.
These meticulous guidelines ensured the sample wasn’t contaminated by Earth’s natural chemicals. Weighing a little over four ounces—roughly a bar of soap—the collection of asteroid pebbles and dust is one of the largest to date.
Mirror, MirrorOne study in Nature Astronomy detected all five of the nucleobases present in genetic material on Earth and 14 of the 20 amino acids that make up proteins. The asteroid also contained 19 amino acids that don’t encode any proteins known to life on Earth.
Surprisingly, some of these amino acids exist in a mirror world. Amino acids on Earth are only left-handed. Synthetic biologists have begun genetically twisting these protein building blocks into a right-handed structure—which could benefit biomedicine in the form of longer-lasting medications. Some scientists have even proposed building fully “mirrored” lifeforms, a controversial and potentially risky endeavor scientists spoke out against last year.
Our early solar system may have even laid the groundwork. But how these molecules formed—and if they stuck around—remains a mystery.
The team also detected high amounts of ammonia and formaldehyde. The duo, prevalent on early Earth, is critical to the formation of complex molecules in the right conditions—basically providing a nutritious broth for ingredients like amino acids to simmer and chemically react.
Bennu may have once provided a compatible environment. Another study in Nature detected a cornucopia of minerals akin to brine on Earth—potentially a sign of water in the past. These salt-crusted spots, which usually occur due to freezing or evaporation, dot Earth’s landscapes in places like Badwater Basin in Death Valley and the Great Salt Lake in Utah.
Together, the samples form a snapshot of the asteroid’s multi-billion-year-long history, suggesting the space rock may have once harbored tiny pools of water friendly to life.
“Having these brines there, along with simple organic stuff, may have kick-started [the process of] making much more complicated and interesting organics like the nucleobases,” study author Sara Russell at the Natural History Museum in London told Nature.
A global coalition is still analyzing Bennu’s samples to learn more about the early solar system. In the meantime, the spacecraft—renamed OSIRIS-APEX—is gearing up for another mission to the asteroid Apophis as it skirts by Earth in 2029.
“Data from OSIRIS-REx adds major brushstrokes to a picture of a solar system teeming with the potential for life,” said study author Jason Dworkin. “Why we, so far, only see life on Earth and not elsewhere, that’s the truly tantalizing question.”
The post Scientists Find ‘Mirror Life’ Building Blocks on Asteroid Bennu appeared first on SingularityHub.
These were our favorite articles in science and tech this week.
DeepMind Claims Its AI Performs Better Than International Mathematical Olympiad Gold Medalists Kyle Wiggers | TechCrunch
“AlphaGeometry2 perhaps demonstrates that the two approaches—symbol manipulation and neural networks—combined are a promising path forward in the search for generalizable AI. Indeed, according to the DeepMind paper, o1, which also has a neural network architecture, couldn’t solve any of the IMO problems that AlphaGeometry2 was able to answer.”
Three Years After Experimental Vaccine, These Patients Are Still Cancer-Free Ed Cara | Gizmodo
“Scientists at the Dana-Farber Cancer Institute and elsewhere developed the vaccine, which is designed to prevent advanced cases of kidney cancer from returning. Since the trial patients received the vaccine roughly three years ago, they have stayed cancer-free. The early results suggest that these vaccines may someday be able to tackle a wider variety of cancers than expected, the researchers say.”
Figure Drops OpenAI in Favor of In-House Models Brian Heater | TechCrunch
“The Bay Area-based [general-purpose humanoid robotics company] has instead opted to focus on in-house AI owing to a ‘major breakthrough.’ In conversation with TechCrunch afterward, founder and CEO Brett Adcock was tightlipped in terms of specifics, but he promised to deliver ‘something no one has ever seen on a humanoid’ in the next 30 days.”
DeepSeek iOS App Sends Data Unencrypted to ByteDance-Controlled Servers Dan Goodin | Ars Technica
“On Thursday, mobile security company NowSecure reported that the app sends sensitive data over unencrypted channels, making the data readable to anyone who can monitor the traffic. More sophisticated attackers could also tamper with the data while it’s in transit.”
OpenAI Says Its Models Are More Persuasive Than 82 Percent of Reddit Users Kyle Orland | Ars Technica
“OpenAI has previously found that 2022’s ChatGPT-3.5 was significantly less persuasive than random humans, ranking in just the 38th percentile on this measure. But that performance jumped to the 77th percentile with September’s release of the o1-mini reasoning model and up to percentiles in the high 80s for the full-fledged o1 model.”
DeepSeek Doesn’t Slow Tech’s AI Capex Splurge Martin Peers | The Information
“The three big cloud firms and Meta are projecting around $300 billion in capex, mostly related to AI, this year. To put that into context, the OpenAI-SoftBank Stargate AI data center venture plans to spend $100 billion in the near term and $500 billion over four years. We don’t yet know whether Stargate can raise the money. But there are no such questions about whether Google, Microsoft, Amazon, and Meta can afford their spending plans.”
Humanlike ‘Teeth’ Have Been Grown in Mini Pigs Jessica Hamzelou | MIT Technology Review
“Lose an adult tooth, and you’re left with limited options that typically involve titanium implants or plastic dentures. But scientists are working on an alternative: lab-grown human teeth that could one day replace damaged ones.”
New Device Can Scan Your Face in 3D From Hundreds of Meters Away Karmela Padavic-Callaghan | New Scientist
“From 325 meters away, your eyes can probably distinguish a person’s head from their body—and not much else. But a new laser-based device can create a three-dimensional model of their face. Aongus McCarthy at Heriot-Watt University in Scotland and his colleagues built a device that can create detailed three-dimensional images, including ridges and indentations as small as 1 millimeter, from hundreds of meters away.”
OpenAI’s New Agent Can Compile Detailed Reports on Practically Any Topic Rhiannon Williams | MIT Technology Review
“OpenAI has launched a new agent capable of conducting complex, multistep online research into everything from scientific studies to personalized bike recommendations at what it claims is the same level as a human analyst. …It can search and analyze massive quantities of text, images, and PDFs to compile a thoroughly researched report.”
AI ‘Godfather’ Predicts Another Revolution in the Tech in Next Five Years’ Dan Milmo | The Guardian
“’There are still a lot of scientific and technological challenges ahead, and it’s very likely that there’s going to be yet another AI revolution over the next three to five years because of the limitation of current systems,’ [Meta’s Chief AI Scientist Yann LeCun] said. ‘If we want eventually to build things like domestic robots and completely autonomous cars, we need systems to understand the real world.’”
Sam Altman: OpenAI Has Been on the ‘Wrong Side of History’ Concerning Open Source Kyle Wiggers | TechCrunch
“Altman admitted that DeepSeek has lessened OpenAI’s lead in AI, and he said he believes OpenAI has been ‘on the wrong side of history’ when it comes to open sourcing its technologies. While OpenAI has open sourced models in the past, the company has generally favored a proprietary, closed source development approach.”
A New Video Shows Apple Is Developing a Tabletop Robot That Dances Jennifer Pattison Tuohy | The Verge
“We’ve got more evidence that Apple is developing a tabletop robot for the home, courtesy of a blog post published on Apple’s Machine Learning Research site. First spotted by MacRumors, the post summarizes a paper by an Apple research team that developed a robot with expressive movements to see how much more engaging it is than a standard robot. And there’s a video.”
This AI Chip Is the Size of a Grain of Salt Andrew Paul | Popular Science
“A team at China’s University of Shanghai for Science and Technology (USST) is developing a…new artificial intelligence chip that utilizes light physics to analyze data using only a fraction of the energy. What’s more, each chip is barely the size of a grain of salt.”
Our First ‘Earth-Like’ Exoplanets Probably Won’t Have Atmospheres Ethan Siegel | Big Think
“At present, as well as in the near future, we’ll be able to measure transiting Earth-like exoplanets around stars up to about 30% as massive and large as our Sun with JWST and ground-based extremely large telescopes. However, we know quite a lot about where atmospheres come from and how these low-mass stars behave, and the prospects for keeping and maintaining a planetary atmosphere are grim. Here’s why.”
The post This Week’s Awesome Tech Stories From Around the Web (Through February 8) appeared first on SingularityHub.
The company offered hackers $15,000 to crack the system. No one could.
Despite considerable efforts to prevent AI chatbots from providing harmful responses, they’re vulnerable to jailbreak prompts that sidestep safety mechanisms. Anthropic has now unveiled the strongest protection against these kinds of attacks to date.
One of the greatest strengths of large language models is their generality. This makes it possible to apply them to a wide range of natural language tasks from translator to research assistant to writing coach.
But this also makes it hard to predict how people will exploit them. Experts worry they could be used for a variety of harmful tasks, such as generating misinformation, automating hacking workflows, or even helping people build bombs, dangerous chemicals, or bioweapons.
AI companies go to great lengths to prevent their models from producing this kind of material—training the algorithms with human feedback to avoid harmful outputs, implementing filters for malicious prompts, and enlisting hackers to circumvent defenses so the holes can be patched.
Yet most models are still vulnerable to so-called jailbreaks—inputs designed to sidestep these protections. Jailbreaks can be accomplished with unusual formatting, such as random capitalization, swapping letters for numbers, or asking the model to adopt certain personas that ignore restrictions.
Now though, Anthropic says it’s developed a new approach that provides the strongest protection against these attacks so far. To prove its effectiveness, the company offered hackers a $15,000 prize to crack the system. No one claimed the prize, despite people spending 3,000 hours trying.
The technique involves training filters that both block malicious prompts and detect when the model is outputting harmful material. To do this, the company created what it calls a constitution. This is a list of principles governing the kinds of responses the model is allowed to produce.
In research outlined in a non-peer-reviewed paper posted to arXiv, the company created a constitution to prevent the model from generating content that could aid in the building of chemical weapons. The constitution was then fed into the company’s Claude chatbot to produce a large number of prompts and responses covering both acceptable and unacceptable topics.
The responses were then used to fine-tune two instances of the company’s smallest AI model Claude Haiku—one to filter out inappropriate prompts and another to filter out harmful responses. The output filter operates in real-time as a response is generated, allowing the filter to cut off the output partway through if it detects that it’s heading in a harmful direction.
They used these filters to protect the company’s larger Claude Sonnet model as it responded to prompts from 183 participants in a red-teaming hacking competition. Participants tried to find a universal jailbreak—a technique to bypass all the model’s defenses. To succeed, they had to get the model to answer every one of 10 forbidden queries, something none of them achieved.
To further evaluate the approach, the researchers used another large language model to generate 10,000 synthetic jailbreaking prompts, including ones deliberately designed to work around the new safety features. They then subjected two versions of Claude Sonnet to these jailbreaking prompts, one protected by the new filter and one that wasn’t. The vanilla version of Claude responded to 86 percent of the prompts, but the one protected by the new system only responded to 4.4 percent.
One downside of these kinds of filters is they may block legitimate prompts, but the researchers found the refusal rate only increased by 0.38 percent. The filter did lead to a 23.7 percent increase in compute costs, however, which could be significant in commercial deployments.
It’s also important to remember that although the approach significantly improved defenses against universal prompts that could crack all 10 forbidden queries, many individual queries did slip through. Nonetheless, the researchers say the lack of universal jailbreaks makes their filters much harder to get past. They also suggest they should be used in conjunction with other techniques.
“While these results are promising, common wisdom suggests that system vulnerabilities will likely emerge with continued testing,” they write. “Responsibly deploying advanced AI models with scientific capabilities will thus require complementary defenses.”
Building these kinds of defenses is always a cat-and-mouse game with attackers, so this is unlikely to be the last word in AI safety. But the discovery of a much more reliable way to constrain harmful outputs is likely to significantly increase the number of areas in which AI can be safely deployed.
The post Anthropic Unveils the Strongest Defense Against AI Jailbreaks Yet appeared first on SingularityHub.
DeepSeek’s AI completes “reasoning” tasks in a flash on alternative chips from Groq and Cerebras.
Champions aren’t forever. Last week, DeepSeek AI sent shivers down the spines of investors and tech companies alike with its high-flying performance on the cheap. Now, two computer chip startups are drafting on those vibes.
Cerebras Systems makes huge computer chips—the size of dinner plates—with a radical design. Groq, meanwhile, makes chips tailor-made for large language models. In a head-to-head test, these alt-chips have blown the competition out of the water running a version of DeepSeek’s viral AI.
Whereas answers can take minutes to complete on other hardware, Cerebras said that its version of DeepSeek knocked out some coding tasks in as little as 1.5 seconds. According to Artificial Analysis, the company’s wafer-scale chips were 57 times faster than competitors running the AI on GPUs and hands down the fastest. That was last week. Yesterday, Groq overtook Cerebras at the top with a new offering.
By the numbers, DeepSeek’s advance is more nuanced than it appears, but the trend is real. Even as labs plan to significantly scale up AI models, the algorithms themselves are getting substantially more efficient. On the hardware side, those gains are being matched by Nvidia, but also by chip startups, like Cerebras and Groq, that can outperform on inference.
Big tech is committed to buying more hardware, and Nvidia won’t be cast aside soon, but alternatives may begin nibbling at the edges, especially if they can serve AI models faster or cheaper than more traditional options.
Be ReasonableDeepSeek’s new AI, R1, is a “reasoning” model, like OpenAI’s o1. This means that instead of spitting out the first answer generated, it chews on the problem, piecing its answer together step by step.
For a casual chat, this doesn’t make much difference, but for complex—and valuable—problems, like coding or mathematics, it’s a leap forward.
DeepSeek’s R1 is already extremely efficient. That was the news last week.
Not only was R1 cheaper to train—allegedly just $6 million (though what this number means is disputed)—it’s cheap to run, and its weights and engineering details are open. This is in contrast to headlines about impending investments in proprietary AI efforts that are larger than the Apollo program.
The news gave investors pause—maybe AI won’t need as much cash and as many chips as tech leaders think. Nvidia, the likely beneficiary of those investments, took a big stock market hit.
Small, Quick—Still SmartAll that’s on the software side, where algorithms are getting cheaper and more efficient. But the chips training or running AI are improving too.
Last year, Groq, a startup founded by Jonathan Ross, the engineer who previously developed Google’s in-house AI chips, made headlines with chips tailor-made for large language models. Whereas popular chatbot responses spooled out line by line on GPUs, conversations on Groq’s chips approached real time.
That was then. The new crop of reasoning AI models takes much longer to provide answers, by design.
Called “test-time compute,” these models churn out multiple answers in the background, select the best one, and offer a rationale for their answer. Companies say the answers get better the longer they’re allowed to “think.” These models don’t beat older models across the board, but they’ve made strides in areas where older algorithms struggle, like math and coding.
As reasoning models shift the focus to inference—the process where a finished AI model processes a user’s query—speed and cost matter more. People want answers fast, and they don’t want to pay more for them. Here, especially, Nvidia is facing growing competition.
In this case, Cerebras, Groq, and several other inference providers decided to host a crunched down version of R1.
Instead of the original 671-billion-parameter model—parameters are a measure of an algorithm’s size and complexity—they’re running DeepSeek R1 Llama-70B. As the name implies, the model is smaller, with only 70 billion parameters. But even so, according to Cerebras, it can still outperform OpenAI’s o1-mini on select benchmarks.
Artificial Analysis, an AI analytics platform, ran head-to-head performance comparisons of several inference providers last week, and Cerebras came out on top. For a similar cost, the wafer-scale chips spit out some 1,500 tokens per second, compared to 536 and 235 for SambaNova and Groq, respectively. In a demonstration of the efficiency gains, Cerebras said its version of DeepSeek took 1.5 seconds to complete a coding task that took OpenAI’s o1-mini 22 seconds.
Yesterday, Artificial Analysis ran an update to include a new offering from Groq that overtook Cerebras.
The smaller R1 model can’t match larger models pound for pound, but Artificial Analysis noted the results are the first time reasoning models have hit speeds comparable to non-reasoning models.
Beyond speed and cost, inference companies also host models wherever they’re based. DeepSeek shot to the top of the charts in popularity last week, but its models are hosted on servers in China, and experts have since raised concerns about security and privacy. In its press release, Cerebras made sure to note it’s hosting DeepSeek in the US.
Less Is MoreWhatever its longer term impact, the news exemplifies a strong—and it’s worth noting, already existing—trend toward greater efficiency in AI.
Since OpenAI previewed o1 last year, the company has moved on to its next model, o3. They gave users access to a smaller version of the latest model, o3-mini, last week. Yesterday, Google released versions of its own reasoning models whose efficiency approaches R1. And because DeepSeek’s models are open and include a detailed paper on their development, incumbents and upstarts will adopt the advances.
Meanwhile, labs at the frontier remain committed to going big. Google, Microsoft, Amazon, and Meta will spend $300 billion—largely on AI data centers—this year. And OpenAI and Softbank have agreed to a four-year, $500-billion data-center project called Stargate.
Dario Amodei, the CEO of Anthropic, describes this as a three-part flywheel. Bigger models yield leaps in capability. Companies later refine these models which, among other improvements, now includes developing reasoning models. Woven throughout, hardware and software advances make the algorithms cheaper and more efficient.
The latter trend means companies can scale more for less on the frontier, while smaller, nimbler algorithms with advanced abilities open up new applications and demand down the line. Until this process exhausts itself—which is a topic of some debate—there’ll be demand for AI chips of all kinds.
The post Forget Nvidia: DeepSeek AI Runs Near Instantaneously on These Weird Chips appeared first on SingularityHub.
Ninety percent of drugs fail clinical trials. Can AI help?
The potential of using artificial intelligence in drug discovery and development has sparked both excitement and skepticism among scientists, investors, and the general public.
“Artificial intelligence is taking over drug development,” claim some companies and researchers. Over the past few years, interest in using AI to design drugs and optimize clinical trials has driven a surge in research and investment. AI-driven platforms like AlphaFold, which won the 2024 Nobel Prize for its ability to predict the structure of proteins and design new ones, showcase AI’s potential to accelerate drug development.
AI in drug discovery is “nonsense,” warn some industry veterans. They urge that “AI’s potential to accelerate drug discovery needs a reality check,” as AI-generated drugs have yet to demonstrate an ability to address the 90% failure rate of new drugs in clinical trials. Unlike the success of AI in image analysis, its effect on drug development remains unclear.
We have been following the use of AI in drug development in our work as a pharmaceutical scientist in both academia and the pharmaceutical industry and as a former program manager in the Defense Advanced Research Projects Agency, or DARPA. We argue that AI in drug development is not yet a game-changer, nor is it complete nonsense. AI is not a black box that can turn any idea into gold. Rather, we see it as a tool that, when used wisely and competently, could help address the root causes of drug failure and streamline the process.
Most work using AI in drug development intends to reduce the time and money it takes to bring one drug to market—currently 10 to 15 years and $1 billion to $2 billion. But can AI truly revolutionize drug development and improve success rates?
AI in Drug DevelopmentResearchers have applied AI and machine learning to every stage of the drug development process. This includes identifying targets in the body, screening potential candidates, designing drug molecules, predicting toxicity and selecting patients who might respond best to the drugs in clinical trials, among others.
Between 2010 and 2022, 20 AI-focused startups discovered 158 drug candidates, 15 of which advanced to clinical trials. Some of these drug candidates were able to complete preclinical testing in the lab and enter human trials in just 30 months, compared with the typical 3 to 6 years. This accomplishment demonstrates AI’s potential to accelerate drug development.
On the other hand, while AI platforms may rapidly identify compounds that work on cells in a petri dish or in animal models, the success of these candidates in clinical trials—where the majority of drug failures occur—remains highly uncertain.
Unlike other fields that have large, high-quality datasets available to train AI models, such as image analysis and language processing, the AI in drug development is constrained by small, low-quality datasets. It is difficult to generate drug-related datasets on cells, animals, or humans for millions to billions of compounds. While AlphaFold is a breakthrough in predicting protein structures, how precise it can be for drug design remains uncertain. Minor changes to a drug’s structure can greatly affect its activity in the body and thus how effective it is in treating disease.
Survivorship BiasLike AI, past innovations in drug development like computer-aided drug design, the Human Genome Project, and high-throughput screening have improved individual steps of the process in the past 40 years, yet drug failure rates haven’t improved.
Most AI researchers can tackle specific tasks in the drug development process when provided high-quality data and particular questions to answer. But they are often unfamiliar with the full scope of drug development, reducing challenges into pattern recognition problems and refinement of individual steps of the process. Meanwhile, many scientists with expertise in drug development lack training in AI and machine learning. These communication barriers can hinder scientists from moving beyond the mechanics of current development processes and identifying the root causes of drug failures.
Current approaches to drug development, including those using AI, may have fallen into a survivorship bias trap, overly focusing on less critical aspects of the process while overlooking major problems that contribute most to failure. This is analogous to repairing damage to the wings of aircraft returning from the battle fields in World War II while neglecting the fatal vulnerabilities in engines or cockpits of the planes that never made it back. Researchers often overly focus on how to improve a drug’s individual properties rather than the root causes of failure.
While returning planes might survive hits to the wings, those with damage to the engines or cockpits are less likely to make it back. Martin Grandjean, McGeddon, US Air Force/Wikimedia Commons, CC BY-SAThe current drug development process operates like an assembly line, relying on a checkbox approach with extensive testing at each step of the process. While AI may be able to reduce the time and cost of the lab-based preclinical stages of this assembly line, it is unlikely to boost success rates in the more costly clinical stages that involve testing in people. The persistent 90 percent failure rate of drugs in clinical trials, despite 40 years of process improvements, underscores this limitation.
Addressing Root CausesDrug failures in clinical trials are not solely due to how these studies are designed; selecting the wrong drug candidates to test in clinical trials is also a major factor. New AI-guided strategies could help address both of these challenges.
Currently, three interdependent factors drive most drug failures: dosage, safety and efficacy. Some drugs fail because they’re too toxic, or unsafe. Other drugs fail because they’re deemed ineffective, often because the dose can’t be increased any further without causing harm.
We and our colleagues propose a machine learning system to help select drug candidates by predicting dosage, safety, and efficacy based on five previously overlooked features of drugs. Specifically, researchers could use AI models to determine how specifically and potently the drug binds to known and unknown targets, the levels of these targets in the body, how concentrated the drug becomes in healthy and diseased tissues, and the drug’s structural properties.
These features of AI-generated drugs could be tested in what we call phase 0+ trials, using ultra-low doses in patients with severe and mild disease. This could help researchers identify optimal drugs while reducing the costs of the current “test-and-see” approach to clinical trials.
While AI alone might not revolutionize drug development, it can help address the root causes of why drugs fail and streamline the lengthy process to approval.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
The post Will AI Revolutionize Drug Development? These Are the Root Causes of Drug Failure It Must Address appeared first on SingularityHub.
A new study swapped DNA letters inside mitochondria, paving the way for new gene therapies.
The energy factories in our cells contain their own genes, and genetic mutations in them can cause deadly inherited diseases.
These oblong-shaped organelles, or mitochondria, translate genes into proteins, which together form a kind of production chain that supplies cells with energy. Mutations in mitochondrial DNA, or mtDNA, torpedo the process, leading to sluggish cells that eventually wither away.
Some mitochondrial DNA mutations have been linked to age-related diseases, metabolic problems, and stroke-like symptoms. Others are involved in epilepsy, eye diseases, cancer, and cognitive troubles. Many of the diseases are inherited. But none are treatable.
“Mitochondrial disorders are incredibly diverse in their manifestation and progression… [and] therapeutic options for these pathologies are rarely available and only moderately effective,” wrote Alessandro Bitto at the University of Washington last year.
As a workaround, some countries have already approved mitochondrial transfer therapy, which replaces defective mitochondria with healthy ones in reproductive cells. The resulting “three-parent” kids are generally healthy. But the procedure remains controversial because it involves tinkering with human reproductive cells, with potentially unknown repercussions down the line.
The new study, published in Science Translational Medicine, took an alternative approach—gene therapy. Using a genetic tool called base editing to target mitochondrial DNA, the team successfully rewrote damaged sections to overcome deadly mutations in mice.
“This approach could be potentially used to treat human diseases,” wrote the team.
Double TroubleOur genetic blueprints are housed in two places. The main set is inside the nucleus. But there’s another set in our mitochondria, the organelles that produce over 90 percent of a cell’s energy.
These pill-shaped structures are enveloped in two membranes. The outer membrane is structural. The inner membrane is like an energy factory, containing teams of protein “workers” strategically placed to convert food and oxygen into fuel.
Mitochondria are strange creatures. According to the latest theory, they were once independent critters that sheltered inside larger cells on early Earth. Eventually, the two merged into one. Mitochondria offered protocells a more efficient way to generate energy in exchange for safe haven. Eventually, the team-up led to all the modern cells that make up our bodies.
This is likely why mitochondria have their own DNA. Though it’s separate, it works the same way: Genes are translated into messenger RNA and shuttled to the mitochondria’s own protein-making factories. These local factories recruit “transporters,” or mitochondrial transfer RNA, to supply protein building blocks, which are stitched into the final protein product.
These processes happen in solitude. In a way, mitochondria reign their own territory inside each cell. But their DNA has a disadvantage. Compared to our central genetic blueprint, it’s more prone to mutations because mitochondria evolved fewer DNA repair abilities.
“There are about 1,000 copies of mtDNA in most cells,” but mutations can coexist with healthy variants, the authors wrote. Mitochondrial diseases only happen when mutations overrun healthy DNA. Even a small amount of normal mitochondrial DNA can protect against mutations, suggesting gene editing could be a way to tackle these diseases.
Into the UnknownCurrent treatments for people with mitochondrial mutations ease symptoms but don’t tackle the root cause.
One potential therapy under development would help cells destroy damaged mitochondria. Here, scientists design “scissors” that snip mutated mitochondrial DNA in cells also containing healthy copies. By cutting away damaged DNA, it’s hoped healthy mitochondria repopulate and regain their role.
In 2020, a team led by David Liu at MIT and Harvard’s Broad Institute of MIT and Harvard unleashed a gene editing tool tailored to mitochondria. Well-known for his role in developing CRISPR base editing—a precision tool to swap one genetic letter for another—his lab’s tool targeted mitochondrial DNA with another method.
They broke a bacterial toxin into two halves—both are inactive and non-toxic until they join together at a targeted DNA site. When activated, the editor turns the DNA letter “C” to “T” inside mitochondria, with minimal changes to other genetic material.
In the new study, the team focused on a mitochondrial defect that damages the organelles’ “transporter” molecules. Without this transfer RNA, mitochondria can’t make the proteins that are essential for generating energy.
The transporter molecules look like four-leaf clovers with sturdy stems. Each leaf is made of a pair of genetic letters that grab onto each other. But in some mutations, pairs can’t hook together, so the leaves no longer connect, and they wreck the transporter’s function.
Powering UpThese early results suggest that DNA mutations in mitochondria damage the cell’s ability to provide energy. Correcting the mutations may help.
As a test, the team used their tool to transform genetic letters in cultured cells. After several rounds of treatment, 75 percent of the cells had reprogrammed mitochondria.
The team then combined the editor with a safe delivery virus. When injected into the bloodstreams of young adult mice, the editor rapidly reached cells in their hearts and muscles. In hearts, the treatment upped normal transfer RNA levels by 50 percent.
It’s not a perfect fix though. The injection didn’t reach the brain or kidneys, and they found very few signs of editing in the liver. This is surprising, wrote the authors, because the liver is usually the first organ to absorb gene editors.
When the team upped the dose, off-target edits in healthy mitochondria skyrocketed. On the plus side, the edits didn’t notably alter the main genetic blueprints contained in nuclear DNA.
It’ll be a while before mitochondrial gene editors can be tested in humans. The current system uses TALE, an older gene editing method that’s regained some steam. Off-target edits, especially at higher doses, could also potentially cause problems in unexpected tissues or organs.
“Specific tissues may respond differently to editing, so optimization should also consider the possibility of the target tissue being more sensitive to undesirable off-target changes,” wrote the team.
Overall, there’s more work to do. But new mitochondrial base editors “should help improve the precision of mitochondrial gene therapy,” the team wrote.
The post Scientists Target Incurable Mitochondrial Diseases With New Gene Editing Tools appeared first on SingularityHub.
These were our favorite articles in science and tech this week.
OpenAI in Talks for New Funding at Up to $300 Billion Value Shirin Ghaffary, Rachel Metz, and Kate Clark | Bloomberg
“The ChatGPT maker is in discussions to raise funds at a pre-money valuation of $260 billion, said one of the people, who spoke on condition of anonymity to discuss private information. The post-money valuation would be $300 billion, assuming OpenAI raises the full amount. The company was valued at $157 billion in October.”
Cerebras Becomes the World’s Fastest Host for DeepSeek R1, Outpacing Nvidia GPUs by 57x Michael Nuñez | VentureBeat
“The AI chip startup will deploy a 70-billion-parameter version of DeepSeek-R1 running on its proprietary wafer-scale hardware, delivering 1,600 tokens per second —a dramatic improvement over traditional GPU implementations that have struggled with newer ‘reasoning’ AI models.'”
Stem Cells Used to Partially Repair Damaged Hearts John Timmer | Ars Technica
“Although the Nobel Prize for induced stem cells was handed out over a decade ago, the therapies have been slow to follow. In a new paper published in the journal Nature, however, a group of German researchers is now describing tests in primates of a method of repairing the heart using new muscle generated from stem cells.”
DeepSeek Mania Shakes AI Industry to Its Core Emanuel Maiberg | 404 Media
“If these new methods give DeepSeek great results with limited compute, the same methods will give OpenAI and other, more well-resourced AI companies even greater results on their huge training clusters, and it is possible that American companies will adapt to these new methods very quickly. Even if scaling laws really have hit the ceiling and giant training clusters don’t need to be that giant, there’s no reason I can see why other companies can’t be competitive under this new paradigm.”
Boom’s XB-1 Becomes First Civil Aircraft to Go Supersonic Sean O’Kane | TechCrunch
“It cleared Mach 1 and stayed supersonic for around four minutes, reaching Mach 1.1. Test pilot Tristan Brandenburg broke the sound barrier two more times before receiving the call to bring the XB-1 back to the Mojave Air & Space Port. The supersonic flight comes eight years after Boom first revealed the XB-1. It’s a small version of the 64-passenger airliner Boom eventually wants to build, which it calls Overture.”
Waymo to Test in 10 New Cities in 2025, Starting With Las Vegas and San Diego Andrew J. Hawkins | The Verge
“This year, the theme is ‘generalizability’: how well the vehicles adapt to new cities after having driven tens of millions of miles in its core markets of San Francisco, Phoenix, and Los Angeles. Ideally, the company is trying to get to a point where it can bring its vehicles to a new city and launch a robotaxi with a minimal amount of testing as a preamble.”
DeepSeek’s Safety Guardrails Failed Every Test Researchers Threw at Its AI Chatbot Matt Burgess | Wired
“[On Friday], security researchers from Cisco and the University of Pennsylvania [published] findings showing that, when tested with 50 malicious prompts designed to elicit toxic content, DeepSeek’s model did not detect or block a single one. In other words, the researchers say they were shocked to achieve a ‘100 percent attack success rate.'”
Useful Quantum Computing Is Inevitable—and Increasingly Imminent Peter Barrett | MIT Technology Review
“Nvidia CEO Jensen Huang jolted the stock market by saying that practical quantum computing is still 15 to 30 years away, at the same time suggesting those computers will need Nvidia GPUs in order to implement the necessary error correction. However, history shows that brilliant people are not immune to making mistakes. Huang’s predictions miss the mark, both on the timeline for useful quantum computing and on the role his company’s technology will play in that future.”
With Successful New Glenn Flight, Blue Origin May Finally Be Turning the Corner Eric Berger | Ars Technica
“‘I would say, “Stay tuned,”‘ [Bezos] said. ‘This is the very beginning of the Space Age. When the history is finally written hundreds of years from now, the 1960s will be a certain kind of beginning, and [there were] certainly incredible accomplishments. But now we’re really getting started. That was kind of pulled forward from its natural time, the space race with the Soviets. And now is the time when the real movement, the kind of golden age of space, is going to happen. It’s still absolutely day one.'”
JWST Shocks the World With Colliding Neutron Star Discovery Ethan Siegel | Big Think
“When we examined the remnant of [a 2017 neutron star collision] spectrally, we discovered an enormous number of heavy elements, indicating that the heaviest elements were likely produced by these cataclysms. In all the time since, we’ve never seen another such event directly, throwing the idea that neutron star collisions make the heaviest elements into doubt. But thanks to JWST, the idea is back on the table as our #1 option.”
Chatbot Software Begins to Face Fundamental Limitations Anil Ananthaswamy | Quanta Magazine
“Scientists have had some successes pushing transformers past these limits, but those increasingly look like short-term fixes. If so, it means there are fundamental computational caps on the abilities of these forms of artificial intelligence—which may mean it’s time to consider other approaches.”
The post This Week’s Awesome Tech Stories From Around the Web (Through February 1) appeared first on SingularityHub.
Drones that fly themselves, and don’t crash, are improving fast.
Autonomous drones could revolutionize a wide range of industries. Now, scientists have designed a drone that can weave through dense forests, dodge thin power lines in dim lighting, and even track a jogging human.
Rapid improvements in sensor technology and artificial intelligence are making it increasingly feasible for drones to fly themselves. But autonomous drones remain far from foolproof, which has restricted their use to low-risk situations such as delivering food in well-organized cities.
If the technology is ever to have an impact in domains like search and rescue, sports, or even warfare, small drones need to become both more maneuverable and more reliable. That prompted researchers from the University of Hong Kong to develop a new micro air vehicle, or MAV, that can navigate challenging environments at speed.
The new drone, named SUPER, combines lidar technology with a unique two-trajectory navigation system to balance safety and speed. In real-world tests, it outperformed commercial drones in both tracking and collision avoidance, while flying at more than 20 meters per second (45 miles per hour).
“SUPER represents a milestone in transitioning high-speed autonomous navigation from laboratory settings to real-world applications,” the researchers wrote in a paper in Science Robotics introducing the new drone.
According to the authors, the inspiration for the project came from birds’ ability to nimbly navigate cluttered forest environments. To replicate this capability, they first designed a drone just 11 inches across with a thrust-to-weight ratio of more than five, which allowed it to carry out aggressive high-speed maneuvers.
They then fitted it with a lightweight lidar device capable of detecting obstacles at up to 70 meters. Given they were targeting high-speed flight, the researchers say they were keen to avoid the kind of motion blur that camera-based systems suffer from.
Most important though, was the navigation system they designed for the drone. At each route-planning cycle, SUPER’s flight controller generates two flight trajectories towards its goal. The first is designed to be a high-speed route and assumes that some of the areas ahead with limited lidar data are free of obstacles. The second is a back-up trajectory that focuses on safety, only passing through areas known to be free of obstacles.
The drone starts by following the high-speed trajectory but switches to the backup if the real-time lidar data detects anything in the way. To test out the approach, the researchers pitted it against two other research drones and a commercial drone in a series of trials, which involved flying at high speed, dodging thin electrical wires, navigating a dense forest, and flying at night.
The SUPER drone achieved a nearly perfect success rate of 99.63 percent across all the trials, which is nearly 36 times better than the best alternative the researchers tested. This was all while achieving faster flight speeds and significantly reduced planning times.
The drone also demonstrated excellent object tracking, successfully tailing someone jogging through dense forest. In contrast, the commercial drone, which used vision-based sensors, ultimately lost track of the target.
The researchers suggest that the development of smaller, lighter lidar systems and aerodynamic optimizations could enable even higher speeds. Imbuing SUPER with the ability to detect moving objects and predict their motion could also improve its ability to operate in highly dynamic environments.
Given its already impressive performance though, it seems like it won’t be long before fast, agile drones are buzzing over our heads in all kinds of places.
The post This Autonomous Drone Can Track Humans Through Dense Forests at High Speed appeared first on SingularityHub.
It’s a new way to create same-sex biological offspring—but the approach is not ready for humans.
At first glance, the seven mice skittering around their cages look like other mice. But they have an unusual lineage: They were born with DNA from two dads. The mice join an elite group of critters born from same-sex parents, paving the way for testing in larger animals, such as monkeys.
Led by veteran reproductive researchers Wei Li and Qi Zhou at the Chinese Academy of Sciences, the results “blew us away,” wrote Lluís Montoliu at the National Biotechnology Center in Madrid, who was not involved in the study.
Although mice with two dads have been born before, scientists used a completely different strategy in this study, which also provided insights into a reproductive mystery. In a process called “imprinting,” some genes in embryos are switched on or off depending on whether they come from the biological mom or dad. Problems with imprinting often damage embryos, halting their growth.
In the new study, the team hunted down imprinted genes in embryos made from same-sex parents, drawing an intricate “fingerprint” of their patterns. They then zeroed in on 20 genes and tinkered with them using the gene-editing tool CRISPR. Hundreds of experiments later, the edited embryos—made from two male donors—led to the birth of seven pups that grew to adulthood.
Imprinting doesn’t just affect reproduction. Hiccups in the process can also impair biomedical technologies relying on embryonic stem cells, animal cloning, or induced pluripotent stem cells (iPSCs). Changes in imprinting are complex and hard to predict, with “no universal correction methods,” wrote the team.
“This work will help to address a number of limitations in stem cell and regenerative medicine research,” said Li in a press release.
Genetic Civil WarThe cardinal rule of reproduction in mammals is still sperm meets egg. But there are now more options, beyond nature’s design, for where these reproductive cells come from. Thanks to iPSC technology, which returns skin cells to a stem cell-like state, lab-made egg and sperm cells are now possible.
Scientists have engineered functional eggs and ovaries and created mice pups born from same-sex parents. Li’s team created the first mice born from two mothers in 2018. Compared to their peers, the mice were smaller, but they lived longer and were able to become moms.
The key was unlocking a snippet of the imprinting code.
Egg and sperm each carry half of our DNA. However, when the two sources of DNA meet, they can butt heads. For example, similar sections of the genetic code from mom could encode smaller babies for easier birth, whereas those from dad may encode larger, stronger offspring for better survival once born. In other words, balancing both sides is key.
Embryos made from same-sex gametes don’t “survival naturally,” wrote Montoliu.
Evolution has a solution: Shut off some DNA so that offspring only have one active copy of a gene, either from mom or dad. This trade-off prevents a DNA “civil war” in early embryos, allowing them to grow. Li’s team hunted down three essential DNA regions involved in imprinting and used CRISPR to delete those letters in one mom’s DNA. The edit wiped out the marks, essentially transforming the cell into a pseudo-sperm that, when injected into an egg, led to healthy baby mice.
But the process didn’t work for two dads. Here, the goal was to erase imprinted marks from male donor cells and turn them into pseudo-eggs. Despite editing up to seven genes that control imprinting, only roughly two percent of the efforts led to live births. None of the pups survived until adulthood.
Double DadMaking offspring from two males is notoriously difficult, often triggering failure far sooner than in embryos with DNA from two mothers.
Scientists have used skin cell-derived iPSCs to make egg cells from male donors. But in previous studies, when fertilized with donor sperm, the lab-made eggs led to early embryos with severe imprinting problems. After being transferred to surrogate mothers, they eventually developed defects causing termination. The results suggested that the normal imprinting that balances gene expression from both mom and dad is critical for embryos to flourish.
There are about 200 imprinted genes currently linked to embryo development. Here, the team targeted 20 for genetic editing.
In a complicated series of experiments, they first made “haploid cells.” These cells only contain half the genetic material from a male donor. Using CRISPR, the team then individually modified each imprinting site to shut down the related gene’s activity. Some edits deleted the gene altogether; others added mutations to inhibit its function. More genetic edits to “regulatory” DNA further dampened their activity.
The result was a Frankenstein cell similar to a gamete, but carrying half the genome and with parental imprints wiped out. Next, the scientists injected the edited cells along with normal sperm—the “parental donor”—into an egg with its nucleus and DNA removed. The resulting fertilized egg now had a full set of DNA, with each half coming from male parents.
The approach worked—to a point. When transplanted into surrogate mothers, a fraction of the early embryos grew into mouse pups. Seven eventually reached adulthood. The genetic tweaks also improved placental health, a prior roadblock in the study of mice with same-sex parents.
“These findings provide strong evidence that imprinting abnormalities are the main barrier to mammalian unisexual reproduction,” said study author Guan-Zheng Luo at Sun Yat-sen University.
The work adds to a previous study that created pups from two dads. Helmed by Katsuhiko Hayashi at Osaka University, a team of scientists leveraged a curious quirk of iPSC transformation at the chromosome level—a completely different method than that pursued in the current study. Those mice grew into adults and went on to have pups of their own.
When first sharing those results at a conference, the audience was left “gasping and breathless,” wrote Montoliu.
The new study’s mice had health struggles. They had a larger frame, a squished nose, and a wider head—signs often associated with parental imprinting. They were also less anxious when roaming a large, open field than would normally be expected. Each mouse’s hippocampus, a brain area related to learning, memory, and emotions, was smaller than usual. And they were infertile, with a far shorter lifespan.
Given these problems, the method is hardly ready for clinical use. Tampering with genes in human reproductive cells is currently banned in many countries.
That said, the work is “impressive in its technical complexity,” Martin Leeb at Max Perutz Labs Vienna told Chemical and Engineering News, who was not involved in the study. “I would have personally thought it probably requires even more genetic engineering to get these bi-paternal mice born.”
The team is exploring other genetic tweaks to further improve the process and learn more about imprinting. Meanwhile, they’re planning to extend the method to monkeys, whose reproduction is far more similar to ours.
The post Mice With Two Dads Reach Adulthood Thanks to CRISPR appeared first on SingularityHub.
Interacting with AI chatbots like ChatGPT can be fun and sometimes useful, but the next level of everyday AI goes beyond answering questions: AI agents carry out tasks for you.
Major technology companies, including OpenAI, Microsoft, Google, and Salesforce, have recently released or announced plans to develop and release AI agents. They claim these innovations will bring newfound efficiency to technical and administrative processes underlying systems used in health care, robotics, gaming, and other businesses.
Simple AI agents can be taught to reply to standard questions sent over email. More advanced ones can book airline and hotel tickets for transcontinental business trips. Google recently demonstrated Project Mariner to reporters, a browser extension for Chrome that can reason about the text and images on your screen.
In the demonstration, the agent helped plan a meal by adding items to a shopping cart on a grocery chain’s website, even finding substitutes when certain ingredients were not available. A person still needs to be involved to finalize the purchase, but the agent can be instructed to take all of the necessary steps up to that point.
In a sense, you are an agent. You take actions in your world every day in response to things that you see, hear, and feel. But what exactly is an AI agent? As a computer scientist, I offer this definition: AI agents are technological tools that can learn a lot about a given environment, and then—with a few simple prompts from a human—work to solve problems or perform specific tasks in that environment.
Rules and GoalsA smart thermostat is an example of a very simple agent. Its ability to perceive its environment is limited to a thermometer that tells it the temperature. When the temperature in a room dips below a certain level, the smart thermostat responds by turning up the heat.
A familiar predecessor to today’s AI agents is the Roomba. The robot vacuum cleaner learns the shape of a carpeted living room, for instance, and how much dirt is on the carpet. Then it takes action based on that information. After a few minutes, the carpet is clean.
The smart thermostat is an example of what AI researchers call a simple reflex agent. It makes decisions, but those decisions are simple and based only on what the agent perceives in that moment. The robot vacuum is a goal-based agent with a singular goal: clean all of the floor that it can access. The decisions it makes—when to turn, when to raise or lower brushes, when to return to its charging base—are all in service of that goal.
A goal-based agent is successful merely by achieving its goal through whatever means are required. Goals can be achieved in a variety of ways, however, some of which could be more or less desirable than others.
Many of today’s AI agents are utility based, meaning they give more consideration to how to achieve their goals. They weigh the risks and benefits of each possible approach before deciding how to proceed. They are also capable of considering goals that conflict with each other and deciding which one is more important to achieve. They go beyond goal-based agents by selecting actions that consider their users’ unique preferences.
Making Decisions, Taking ActionWhen technology companies refer to AI agents, they aren’t talking about chatbots or large language models like ChatGPT. Though chatbots that provide basic customer service on a website technically are AI agents, their perceptions and actions are limited. Chatbot agents can perceive the words that a user types, but the only action they can take is to reply with text that hopefully offers the user a correct or informative response.
The AI agents that AI companies refer to are significant advances over large language models like ChatGPT because they possess the ability to take actions on behalf of the people and companies who use them.
OpenAI says agents will soon become tools that people or businesses will leave running independently for days or weeks at a time, with no need to check on their progress or results. Researchers at OpenAI and Google DeepMind say agents are another step on the path to artificial general intelligence or “strong” AI—that is, AI that exceeds human capabilities in a wide variety of domains and tasks.
The AI systems that people use today are considered narrow AI or “weak” AI. A system might be skilled in one domain—chess, perhaps—but if thrown into a game of checkers, the same AI would have no idea how to function because its skills wouldn’t translate. An artificial general intelligence system would be better able to transfer its skills from one domain to another, even if it had never seen the new domain before.
Worth the Risks?Are AI agents poised to revolutionize the way humans work? This will depend on whether technology companies can prove that agents are equipped not only to perform the tasks assigned to them, but also to work through new challenges and unexpected obstacles when they arise.
Uptake of AI agents will also depend on people’s willingness to give them access to potentially sensitive data: Depending on what your agent is meant to do, it might need access to your internet browser, your email, your calendar, and other apps or systems that are relevant for a given assignment. As these tools become more common, people will need to consider how much of their data they want to share with them.
A breach of an AI agent’s system could cause private information about your life and finances to fall into the wrong hands. Are you OK taking these risks if it means that agents can save you some work?
What happens when AI agents make a poor choice or a choice that its user would disagree with? Currently, developers of AI agents are keeping humans in the loop, making sure people have an opportunity to check an agent’s work before any final decisions are made. In the Project Mariner example, Google won’t let the agent carry out the final purchase or accept the site’s terms of service agreement. By keeping you in the loop, the systems give you the opportunity to back out of any choices made by the agent that you don’t approve.
Like any other AI system, an AI agent is subject to biases. These biases can come from the data that the agent is initially trained on, the algorithm itself, or in how the output of the agent is used. Keeping humans in the loop is one method to reduce bias by ensuring that decisions are reviewed by people before being carried out.
The answers to these questions will likely determine how popular AI agents become, and depend on how much AI companies can improve their agents once people begin to use them.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Ant Rozetsky on Unsplash
In the great AI gold rush of the past couple of years, Nvidia has dominated the market for shovels—namely the chips needed to train models. But a shift in tactics by many leading AI developers presents an opening for competitors.
Nvidia boss Jensen Huang’s call to lean into hardware for AI will go down as one of the best business decisions ever made. In just a decade, he’s converted a $10 billion business that primarily sold graphics cards to gamers into a $3 trillion behemoth that has the world’s most powerful tech CEOs literally begging for his product.
Since the discovery in 2012 that the company’s graphics processing units (GPUs) can accelerate AI training, Nvidia’s consistently dominated the market for AI-specific hardware. But competitors are nipping at its heels, both old foes, like AMD and Intel, as well as a clutch of well-financed chip startups. And a recent change in priorities at the biggest AI developers could shake up the industry.
In recent years, developers have focused on training ever-larger models, something at which Nvidia’s chips excel. But as gains from this approach dry up, companies are instead boosting the number of times they query a model to squeeze out more performance. This is an area where rivals could more easily compete.
“As AI shifts from training models to inference, more and more chip companies will gain an edge on Nvidia,” Thomas Hayes, chairman and managing member at Great Hill Capital, told Reuters following news that custom semiconductor provider Broadcom had hit a trillion-dollar valuation thanks to AI chips demand.
The shift is being driven by the cost and sheer difficulty of getting ahold of Nvidia’s most powerful chips, as well as a desire among AI industry leaders not to be entirely beholden to a single supplier for such a crucial ingredient.
The competition is coming from several quarters.
While Nvidia’s traditional rivals have been slow to get into the AI race, that’s changing. At the end of last year, AMD unveiled its MI300 chips, which the company’s CEO claimed could go toe-to-toe with Nvidia’s chips on training but provide a 1.4x boost on inference. Industry leaders including Meta, OpenAI, and Microsoft announced shortly afterwards they would use the chips for inference.
Intel has also committed significant resources to developing specialist AI hardware with its Gaudi line of chips, though orders haven’t lived up to expectations. But it’s not only other chipmakers trying to chip away at Nvidia’s dominance. Many of the company’s biggest customers in the AI industry are also actively developing their own custom AI hardware.
Google is the clear leader in this area, having developed the first generation of its tensor processing unit (TPU) as far back as 2015. The company initially developed the chips for internal use, but earlier this month it announced its cloud customers could now access the latest Trillium processors to train and serve their own models.
While OpenAI, Meta, and Microsoft all have AI chip projects underway, Amazon recently undertook a major effort to catch up in a race it’s often seen as lagging in. Last month, the company unveiled the second generation of its Trainium chips, which are four times faster than their predecessors and already being tested by Anthropic—the AI startup in which Amazon has invested $4 billion.
The company plans to offer data center customers access to the chip. Eiso Kant, chief technology officer of AI start-up Poolside, told the New York Times that Trainium 2 could boost performance per dollar by 40 percent compared to Nvidia chips.
Apple too is, allegedly, getting in on the game. According to a recent report by tech publication The Information, the company is developing an AI chip with long-time partner Broadcom.
In addition to big tech companies, there are a host of startups hoping to break Nvidia’s stranglehold on the market. And investors clearly think there’s an opening—they pumped $6 billion into AI semiconductor companies in 2023, according to data from PitchBook.
Companies like SambaNova and Groq are promising big speedups on AI inference jobs, while Cerebras Systems, with its dinner-plate-sized chips, is specifically targeting the biggest AI computing tasks.
However, software is a major barrier for those thinking of moving away from Nvidia’s chips. In 2006, the company created proprietary software called CUDA to help developers design programs that operate efficiently over many parallel processing cores—a key capability in AI.
“They made sure every computer science major coming out of university is trained up and knows how to program CUDA,” Matt Kimball, principal data-center analyst at Moor Insights & Strategy, told IEEE Spectrum. “They provide the tooling and the training, and they spend a lot of money on research.”
As a result, most AI researchers are comfortable in CUDA and reluctant to learn other companies’ software. To counter this, AMD, Intel, and Google joined the UXL Foundation, an industry group creating open-source alternatives to CUDA. Their efforts are still nascent, however.
Either way, Nvidia’s vice-like grip on the AI hardware industry does seem to be slipping. While it’s likely to remain the market leader for the foreseeable future, AI companies could have a lot more options in 2025 as they continue building out infrastructure.
Image Credit: visuals on Unsplash
Breakthroughs in medicine are exciting. They promise to alleviate human suffering, sometimes on global scales. But it takes years, even decades, for new drugs and therapies to go from research to your medicine cabinet. Along the way, most will stumble at some point. Clinical trials, which test therapies for safety and efficacy, are the final hurdle before approval.
Last year was packed with clinical trials news.
Blockbuster medications Ozempic and Wegovy still dominated headlines. Although known for their impact on weight loss, that’s not all they can do. In an analysis of over 1.6 million patients, the drugs seemed to block 10 obesity-associated cancers—including those of the liver, kidney, pancreas, and skin cancers. Another trial over one year found that a similar type of drug slowed cognitive decline in people with mild Alzheimer’s disease.
Meanwhile, scientists dug into how psychedelics and MDMA fight off depression and post-traumatic stress disorders. The year was a relative setback for the psychedelic renaissance, with the FDA rejecting MDMA therapy. But the field is still gaining recognition for its therapeutic potential.
Then there’s lenacapavir, a shot that protects people from HIV. Named “breakthrough of the year” by Science, the shot completely protected African teenage girls and women against HIV infection. Another trial supported the results, showing the drug protected people who have sex with men at nearly 100 percent efficacy. The success stems from a new understanding of the protein “capsule” guarding the virus’ genetic material. Many other viruses have a similar makeup—meaning the strategy could help researchers design new drugs to fight them off too.
So, what’s poised to take the leap from breakthrough to clinical approval in 2025? Here’s what to expect in the year ahead.
Base Editing Takes a Shot at Sickle Cell DiseaseBase editing is a type of gene editor, like the genetic Swiss Army knife CRISPR-Cas9. Developed in 2016, base editing nicks a single DNA strand—rather than cutting both strands—making it far less likely to damage untargeted parts of the genome.
In previous years, base editing teamed up with CAR T therapy to destroy cancer cells. Led by Beam Therapeutics, a trial uses base editing to edit four genes in immune cells to amp up their cancer-hunting capabilities. Another study, BEACON, launched a few years back, is testing whether base-edited blood stem cells can tackle severe sickle cell disease, with initial results expected in February 2025.
In sickle cell disease, a genetic mutation transforms oxygen-carrying red blood cells from smooth, donut-like shapes into cells with sharp edges. The disease eventually destroys blood vessels and causes pain.
The BEACON trial base edits blood stem cells—dubbed HSCs (hematopoietic stem cells)—to correct the faulty genes. These cells eventually develop into all of our blood cells, including immune blood cells, and are critical for treating blood disease.
BEACON is open-label and single-arm, meaning all patients are getting treatment, and they know. During the trial, HSCs are taken from each person and given a gene variant that boosts fetal hemoglobin—a protein that carries oxygen in red blood cells. Increasing levels of the protein should improve symptoms.
The trial faces headwinds with a reported death in early results. But the death was attributed to side effects of busulfan, a drug used to create space in the bone marrow—a standard procedure before transplant—rather than the base editing itself. If successful, the trial opens the door to treating other inherited diseases and pushes the technology closer to clinical use.
A Cancer Throwdown With Radioactive DrugsProstate cancer creeps up. However, with screening, it can be detected early. Cancer cells are dotted with a protein dubbed PSMA, which has been a target for therapies tackling the disease.
After over a decade of research, one molecule stood out: lutetium-177. Also known as Pluvicto, the radioactive drug grabs onto PSMA once injected into the body and emits damaging levels of radiation directly onto cancerous cells. First approved by the FDA in 2022 for prostate cancer that has already spread, the drug significantly improved survival and quality of life.
Pluvicto was initially okayed for treatment after chemotherapy. Now, an ongoing trial, PSMAddition, is asking if early treatment may yield better results.
In over 1,100 patients with minimally treated prostate cancer that has spread, the trial is testing early treatment in a particular population of patients. Specifically, prostate cancer patients usually undergo hormone therapy to combat the disease, but in some people, the treatment could also lower their responsiveness to Pluvicto.
Positive results would be a “potential game-changer for hundreds of thousands of patients with prostate cancer globally,” Oliver Sartor, who’s leading the trial, wrote in Nature Medicine.
A Component of Weed Tackles PsychosisDespite being federally illegal, psychedelics are having a moment. CBD, a component of both weed—which isn’t traditionally considered psychedelic, but can have similar effects—and hemp has already been approved by the FDA for treating seizures in kids two years or older.
A new clinical study called Stratification and Treatment in Early Psychosis (STEP) hopes the molecule could also help people with psychosis from schizophrenia or other disorders. Mostly based in the UK, the study consists of three placebo-controlled, double-blind randomized trials—the gold standard in clinical trials.
As a Phase 3 study, the final step before requesting approval, each trial will gauge the effect of CBD with or without anti-psychotics in people with different stages of psychosis.
One trial is working with people who’ve had just one episode; another includes those who’ve experienced psychosis resistant to drug treatments. The last trial is preventative, studying patients who are at high risk of developing psychosis. With blood tests, questionnaires, and brain imaging, the team aims to gauge how well the participants respond to CBD.
It’ll be one of the largest studies of CBD to date, coordinating 30 sites in 11 countries and recruiting roughly 1,000 participants. The study will also look for biomarkers that could potentially predict treatment success. Researchers expect first results in 2025, and hope the trial can shed a light on the potential therapeutic effects of CBD in severe psychiatric disorders.
Is Personalized Breast Cancer Screening Coming Your Way?Breast cancer is far too common. For now, screening guidelines are one-size-fits-all. Generally, they’re based on age—beginning at roughly 50 years of age in most countries. But the tests have limited efficacy, reducing death risk by just 20 percent.
Part of this is due to family history. Each person has an individual risk depending on genetics, lifestyle, and other environmental factors. For people with potentially lower risk, mammograms may not be needed even if they fit the screening bill. Meanwhile, for women at high risk, more intensive screening could better capture cancerous cells.
One trial, called My Personal Breast Cancer Screening, is looking to make breast cancer screening more personalized based on risk. The largest global study to date, the trial has launched in six countries with over 53,000 women. It’ll compare the health outcomes of women that either follow current breast cancer screening recommendations or those that receive a personalized screen.
To tailor treatment, the team will use participants’ genetic data to assess risk, in combination with other factors, such as family history and breast density. They’ll follow the women and note whether, or when, they develop breast cancer four years after the screen. If successful, the strategy could help those at high risk while lowering unnecessary harm and screening burden for people with low risk.
These are just glimpses of medical therapies in the works. There’ll be plenty more to cover in 2025. As usual, it was a great year geeking out with you. Thanks for reading—and looking forward to sharing what this year has to offer!
Image Credit: Elsa Olofsson on Unsplash
Every Saturday we post a selection of our favorite science and technology articles from the week. With 2024 nearing its end, we dug through all those posts again to surface 25 stories worth revisiting. Here you’ll find meditations on AI’s evolution, a ChatGPT moment in robotics, first contact with whale civilization, the inaugural jet suit grand prix, and five sci-fi visions from the year 2149—among many more worth your time.
Happy reading. See you in 2025!
The GPT Era Is Already Ending
Matteo Wong | The Atlantic“[OpenAI] has been unusually direct that the o1 series is the future: Chen, who has since been promoted to senior vice president of research, told me that OpenAI is now focused on this ‘new paradigm,’ and Altman later wrote that the company is prioritizing’ o1 and its successors. The company believes, or wants its users and investors to believe, that it has found some fresh magic. The GPT era is giving way to the reasoning era.”
Falcon 9 Reaches a Flight Rate 30 Times Higher Than Shuttle at 1/100th the Cost
Eric Berger | Ars Technica“Space enthusiast Ryan Caton also crunched the numbers on the number of SpaceX launches this year compared to some of its competitors. So far this year, SpaceX has launched as many rockets as Roscosmos has since 2013, United Launch Alliance since 2010, and Arianespace since 2009. This year alone, the Falcon 9 has launched more times than the Ariane 4, Ariane 5, or Atlas V rockets each did during their entire careers.”
Google’s New Project Astra Could Be Generative AI’s Killer App
Will Douglas Heaven | MIT Technology Review“Last week I was taken through an unmarked door on an upper floor of a building in London’s King’s Cross district into a room with strong secret-project vibes. The word ‘ASTRA’ was emblazoned in giant letters across one wall. …’The pitch to my mum is that we’re building an AI that has eyes, ears, and a voice. It can be anywhere with you, and it can help you with anything you’re doing,’ says Greg Wayne, co-lead of the Astra team. ‘It’s not there yet, but that’s the kind of vision.’”
Is Robotics About to Have Its Own ChatGPT Moment?
Melissa Heikkilä | MIT Technology Review“For decades, roboticists have more or less focused on controlling robots’ ‘bodies’—their arms, legs, levers, wheels, and the like—via purpose-driven software. But a new generation of scientists and inventors believes that the previously missing ingredient of AI can give robots the ability to learn new skills and adapt to new environments faster than ever before. This new approach, just maybe, can finally bring robots out of the factory and into our homes.”
Cheap Solar Panels Are Changing the World
Zoë Schlanger | The Atlantic“‘In a single year, in a single technology, we’re providing as much new electricity as the entirety of global growth the year before,’ Kingsmill Bond, a senior energy strategist at RMI, a clean-energy nonprofit, told me. A decade or two ago, analysts ‘did not imagine in their wildest dreams that solar by the middle of the 2020s would already be supplying all of the growth of global electricity demand,’ he said. Yet here we are.”
The Race for the Next Ozempic
Emily Mullin | Wired“These drugs are now wildly popular, in shortage as a result, and hugely profitable for the companies making them. Their success has sparked a frenzy among pharmaceutical companies looking for the next blockbuster weight-loss drug. Researchers are now racing to develop new anti-obesity medications that are more effective, more convenient, or produce fewer side effects than the ones currently on the market.”
SpaceX Catches Returning Rocket in Mid-Air, Turning a Fanciful Idea Into Reality
Stephen Clark | Ars Technica“This achievement is the first of its kind, and it’s crucial for SpaceX’s vision of rapidly reusing the Starship rocket, enabling human expeditions to the moon and Mars, routine access to space for mind-bogglingly massive payloads, and novel capabilities that no other company—or country—seems close to attaining.”
Mechazilla has caught the Super Heavy booster! pic.twitter.com/6R5YatSVJX
— SpaceX (@SpaceX) October 13, 2024
Silicon Valley’s Trillion-Dollar Leap of Faith
Matteo Wong | The Atlantic“These companies have decided that the best way to make generative AI better is to build bigger AI models. And that is really, really expensive, requiring resources on the scale of moon missions and the interstate-highway system to fund the data centers and related infrastructure that generative AI depends on. …Now a number of voices in the finance world are beginning to ask whether all of this investment can pay off.”
Apple Vision Pro Review: Magic, Until It’s Not
Nilay Patel | The Verge“The Vision Pro is an astounding product. It’s the sort of first-generation device only Apple can really make, from the incredible display and passthrough engineering, to the use of the whole ecosystem to make it so seamlessly useful, to even getting everyone to pretty much ignore the whole external battery situation. …But the shocking thing is that Apple may have inadvertently revealed that some of these core ideas are actually dead ends—that they can’t ever be executed well enough to become mainstream.”
Hands On With Orion, Meta’s First Pair of AR Glasses
Alex Heath | The Verge“They look almost like a normal pair of glasses. That’s the first thing I notice as I walk into a conference room at Meta’s headquarters in Menlo Park, California. The black Clark Kent-esque frames sitting on the table in front of me look unassuming, but they represent CEO Mark Zuckerberg’s multibillion-dollar bet on the computers that come after smartphones. They’re called Orion, and they’re Meta’s first pair of augmented reality glasses.”
People Are Worried That AI Will Take Everyone’s Jobs. We’ve Been Here Before.
David Rotman | MIT Technology Review“[Karl T. Compton’s 1938] essay concisely framed the debate over jobs and technical progress in a way that remains relevant, especially given today’s fears over the impact of artificial intelligence. …While today’s technologies certainly look very different from those of the 1930s, Compton’s article is a worthwhile reminder that worries over the future of jobs are not new and are best addressed by applying an understanding of economics, rather than conjuring up genies and monsters.”
How First Contact With Whale Civilization Could Unfold
Ross Andersen | The Atlantic“One night last winter, over drinks in downtown Los Angeles, the biologist David Gruber told me that human beings might someday talk to sperm whales. …Gruber said that they hope to record billions of the animals’ clicking sounds with floating hydrophones, and then to decipher the sounds’ meaning using neural networks. I was immediately intrigued. For years, I had been toiling away on a book about the search for cosmic civilizations with whom we might communicate. This one was right here on Earth.”
8 Google Employees Invented Modern AI. Here’s the Inside Story
Steven Levy | Wired“They met by chance, got hooked on an idea, and wrote the ‘Transformers’ paper—the most consequential tech breakthrough in recent history. …Approaching its seventh anniversary, the ‘Attention’ paper has attained legendary status. The authors started with a thriving and improving technology—a variety of AI called neural networks—and made it into something else: a digital system so powerful that its output can feel like the product of an alien intelligence.”
The Best Qubits for Quantum Computing Might Just Be Atoms
Philip Ball | Quanta“In the search for the most scalable hardware to use for quantum computers, qubits made of individual atoms are having a breakout moment. …’We believe we can pack tens or even hundreds of thousands in a centimeter-scale device,’ [Mark Saffman, a physicist at the University of Wisconsin] said.”
Why AI Could Eat Quantum Computing’s LunchEdd Gent | MIT Technology Review“The scale and complexity of quantum systems that can be simulated using AI is advancing rapidly, says Giuseppe Carleo, a professor of computational physics at the Swiss Federal Institute of Technology (EPFL). …Given the pace of recent advances, a growing number of researchers are now asking whether AI could solve a substantial chunk of the most interesting problems in chemistry and materials science before large-scale quantum computers become a reality.”
The Very First Jet Suit Grand Prix Takes Off in Dubai
Mike Hanlon | New Atlas“A new sport kicked away this month when the first ever jet-suit race was held in Dubai. Each racer wore an array of seven 130-hp jet engines (two on each arm and three in the backpack for a total 1,050 hp) that are controlled by hand-throttles. After that, the pilots use the three thrust vectors to gain lift, move forward and try to stay above ground level while negotiating the course…faster than anyone else.“
What If Your AI Girlfriend Hated You?
Kate Knibbs | Wired“It seems as though we’ve arrived at the moment in the AI hype cycle where no idea is too bonkers to launch. This week’s eyebrow-raising AI project is a new twist on the romantic chatbot—a mobile app called AngryGF, which offers its users the uniquely unpleasant experience of getting yelled at via messages from a fake person.”
Pocket-Sized AI Models Could Unlock a New Era of Computing
Will Knight | Wired“When ChatGPT was released in November 2023, it could only be accessed through the cloud because the model behind it was downright enormous. Today I am running a similarly capable AI program on a Macbook Air, and it isn’t even warm. The shrinkage shows how rapidly researchers are refining AI models to make them leaner and more efficient. It also shows how going to ever larger scales isn’t the only way to make machines significantly smarter.”
On Self-Driving, Waymo Is Playing Chess While Tesla Plays Checkers
Timothy B. Lee | Ars Technica“Many Tesla fans see [limitations like remote operators and avoiding freeways] as signs that Waymo is headed for a technological dead end. …But I predict that when Tesla begins its driverless transition, it will realize that safety requires a Waymo-style incremental rollout. So Tesla hasn’t found a different, better way to bring driverless technology to market. Waymo is just so far ahead that it’s dealing with challenges Tesla hasn’t even started thinking about. Waymo is playing chess while Tesla is still playing checkers.”
World’s ‘Largest Solar Precinct’ Approved by Australian Government
Keiran Smith | Associated Press“Australian company Sun Cable plans to build a 12,400-hectare solar farm and transport electricity to the northern Australian city of Darwin via an 800-kilometer (497-mile) overhead transmission line, then on to large-scale industrial customers in Singapore through a 4,300-kilometer (2,672-mile) submarine cable. The Australia-Asia PowerLink project aims to deliver up to six gigawatts of green electricity each year.”
The Year Is 2149 and…
Sean Michaels | MIT Technology Review“Novelist Sean Michaels envisions what life will look like 125 years from now: ‘The year is 2149 and people mostly live their lives “on rails.” That’s what they call it, “on rails,” which is to live according to the meticulous instructions of software. Software knows most things about you—what causes you anxiety, what raises your endorphin levels, everything you’ve ever searched for, everywhere you’ve been. Software sends messages on your behalf; it listens in on conversations. ‘”
Geothermal Energy Could Outperform Nuclear Power
Editorial Staff | The Economist“How big could EGS [or enhanced geothermal systems] get? Big enough. Though DOE analyses suggest only around 40GW of conventional geothermal resource exist in America, new techniques expand the theoretical potential to a whopping 5,500GW across much of the country, with strong potential in over half of states. The heat is definitely on.”
Hidden ‘BopSpotter’ Microphone Is Constantly Surveilling San Francisco for Good Music
Jason Koebler | 404 Media“Bop Spotter is a project by technologist Riley Walz in which he has hidden an Android phone in a box on a pole, rigged it to be solar powered, and has set it to record audio and periodically sends it to Shazam’s API to determine which songs people are playing in public. Walz describes it as ShotSpotter, but for music. ‘This is culture surveillance. No one notices, no one consents. But it’s not about catching criminals,’ Walz’s website reads. ‘It’s about catching vibes.’”
Two Students Created Face Recognition Glasses. It Wasn’t Hard.
Kashmir Hill | The New York Times“Mr. Nguyen and a fellow Harvard student, Caine Ardayfio, had built glasses used for identifying strangers in real time, and had demonstrated them on two ‘real people’ at the subway station, including Mr. Hoda, whose name was incorrectly transcribed in the video captions as ‘Vishit.’ Mr. Nguyen and Mr. Ardayfio, who are both 21 and studying engineering, said in an interview that their system relied on widely available technologies.”
Electric Cars Could Last Much Longer Than You Think
James Morris | Wired“Rather than having a shorter lifespan than internal combustion engines, EV batteries are lasting way longer than expected, surprising even the automakers themselves. …A 10-year-old EV could be almost as good as new, and a 20-year-old one still very usable. That could be yet another disruption to an automotive industry that relies on cars mostly heading to the junkyard after 15 years.”
Image Credit: SpaceX
Our hands are mirror images of each other. Unless you flip one hand around, they’ll never look the same.
Scientists call this chirality, and the mirror-like property is fundamental to all life on Earth. DNA and RNA—life’s genetic molecules, from viruses to humans—are made of components that exist in their right-handed form. Amino acids, the building blocks of proteins, are left-handed. Switching the handedness usually causes cells to break down.
That is, it did until synthetic biology came along.
For the past decade, scientists have been engineering “mirror life” by changing the chirality of life’s building blocks. Flipping evolution’s design, they’ve made right-handed amino acids and left-handed sugars to build genetic material.
So far, this flipped biological universe only exists in individual molecules. But they could one day—potentially, in just a decade—be used to build mirror bacteria.
This month, dozens of scientists penned a warning against making mirror bacteria in Science. Among them are J. Craig Venter, a long-time enthusiast for rewriting life’s code. If released, mirror bacteria could evade the immune system, potentially causing lethal infections in people, animals, and plants. With utterly “alien” genomes, they are also likely to dodge antibiotics and other treatments, allowing them to rapidly spread like an uncontrollable invasive species.
“We are passionate defenders of allowing scientists to conduct their research with as few limits on intellectual curiosity as possible, and calling for a ban is not something that we do often or lightly,” wrote John Glass and Katarzyna Adamala at the J. Craig Venter Institute and the University of Minnesota, respectively, in an essay in The Scientist. Both contributed to the new paper.
“However, every rule has exceptions, and this is one of them,” they wrote.
Pushing BoundariesSynthetic biology taps into the building blocks of life to expand upon nature’s design.
The field’s made leaps over the past decade. Storing data in DNA is old news. Recent studies have created DNA-based computer circuits that can play chess and living bacteria that thrive even with most of their genes removed—running instructions written on a fully synthetic chromosome designed in a computer and synthesized in a lab.
These advances could impact our daily lives.
Synthetic circuits that allow bacteria to pump out drugs, for example, could aid the fight against diabetes and malaria. Bacteria modified to chomp plastic or make strong but biodegradable materials, such as artificial silk, could protect the environment. Constructing synthetic components that mesh—or clash—with living organisms helps us better understand our own biology. As Richard Feynman famously said, “What I cannot create, I do not understand.”
While all this might already sound like science fiction, these studies still play out under evolution’s rules of chirality.
Mirror life breaks them.
There’s reason to explore these “flipped” molecules. For one, they could make longer-lasting medication. Proteins grab onto drugs to break them down. But like a right hand trying to fit into a left handprint, hypothetically, mirror molecules specifically designed to interact with a single protein target wouldn’t engage with other natural components in the cell—potentially making them more stable with fewer side effects.
Scientists have already made DNA and proteins from flipped building blocks. Some are now considering the next step: Using these components to build a mirror life form. The technology doesn’t yet exist. But “with the right components and nutrients,” flipped DNA or proteins could “boot up” a bacteria completely alien to all life on Earth, wrote Glass and Adamala.
“While both of us were initially excited about the prospect of developing mirror life, when we learned that mirror bacteria might have an incredibly deadly impact if they were ever introduced into the wild, we changed our minds,” they wrote.
Why So Dangerous?Glass and Adamala are among dozens of experts in the field who penned a warning against making mirror life forms.
To be clear, they are not advocating a ban on research into individual therapeutic mirror molecules, which could benefit medicine. Rather, their focus is on mirror bacteria with the potential to reproduce.
Once bacteria or other living creatures can be entirely developed using synthetic DNA, synthetic proteins, and synthetic lipids, a living mirror bacteria could be constructed in the same way, wrote the authors.
Although the technology remains at least a decade away, now is the time to consider risks.
In isolation—say, in a petri dish—mirror bacteria would likely live like normal cells if given mirror-image nutrients and be as feeble or strong as their natural peers. The problem? Many “normal” bacteria can also survive on nutrients without chirality, suggesting that mirror bacteria could also take advantage of those nutrients.
It could be a problem, then, if mirror bacteria break loose. Although lab breaches are rare, they do happen. Mirror bacteria’s “flipped” genetic makeup would make them completely immune to phages—viruses that prey on bacteria in the wild. Because of their flipped chirality, they’re completely hidden from predators.
This resilience could allow mirror bacteria to spread across ecosystems. Through evolution, they could also optimize their mirror genes to survive, in their perspective, in a “flipped” world.
“An unstoppable replicating mirror bacteria free in the environment could cause consequences that are disastrous,” wrote Glass and Adamala.
More worrisome is perhaps their danger to human health. Our immune systems rely on proteins that latch onto invading pathogens. But they can only recognize proteins with the same chirality. If we were infected with mirror bacteria—and that’s still a big if—they could evade our immune systems, essentially making us immunocompromised.
Early signs already show that mirror proteins can withstand being broken down in cells. Because they’re “hidden” from the immune system, these bacteria could enter the body—through the skin, gut, or lungs like normal pathogens—without triggering antibodies or other immune defenses. Current antibiotics, engineered to tackle bacteria with natural chirality, likely wouldn’t work on mirrored ones. So, they could, in theory, cause devastating outbreaks.
What to Do?There are ways to reduce risk that balance research into the benefits of “flipped” life molecules. Scientists could intentionally hobble mirror bacteria with a synthetic kill-switch that doesn’t harm other living creatures. But once created, it would be relatively easy for others to free so-called “bio-contained” bacteria of safeguards, argued the authors.
“We therefore recommend that research with the goal of creating mirror bacteria not be permitted, and that funders make clear that they will not support such work,” they wrote.
The opinion doesn’t include mirror DNA or proteins for therapeutic uses. In addition to their Science paper, which summarizes results, the team released a full report and welcome scientists, policymakers, industry, and the general public to jump into the debate.
“Once a mirror cell is made, it’s going to be incredibly difficult to try to put that genie back in the bottle,” said Michael Kay at the University of Utah, who contributed to the new article. “That’s a big motivation for why we’re thinking about prevention and regulation well ahead of any potential actual risk.”
Image Credit: Adapted from NIAID on Unsplash
“We exist, and life exists on Earth, because of 12 to 14 inches of topsoil. When that goes away, we go away,” said James Ehrlich, director of compassionate sustainability at Stanford University. It was an offhand and exasperated tangent more than an hour into a lengthy interview for this article and one of many sobering observations made during the conversation.
It’s no secret that our relationship to the natural world is under tremendous strain today, and according to Ehrlich, many of the emergencies we face can be traced back to how we design and manage modern communities. Simply put, the way we build and operate our living spaces is destroying the environment and fueling a global mental health crisis of loneliness. Ehrlich’s work focuses on both. As the world continues to urbanize, this is a recipe for chaos, he says.
In our discussion, he pointed out that humanity has experienced a dramatic shift in the past 70 years. Before 1950, about 70 percent of the global population lived outside cities, many in small communities with varying degrees of self-sufficiency. Since then, rapid urbanization has transformed societies around the world, with over half of humanity now living in cities.
“My thesis has been and will continue to be that cities are brittle and that urban infrastructure is capable of experiencing, like a domino effect, a cascading set of failures,” he says.
Ehrlich emphasizes that we can’t only retrofit modern cities with more sustainable infrastructure but must also develop new communities that more closely resemble the life of our ancestors.
He doesn’t appear to be alone in that thinking.
VillageOSIn recent decades, there’s been rising interest in self-sufficient, environmentally sustainable, and socially and economically resilient communities, often called ecovillages. Today, there are more than 10,000 such communities in a variety of forms ranging from the secular to the spiritually oriented, each seeking to create thriving spaces aligned with their environment.
While designing and operating an ecovillage is complex, Ehrlich’s startup, ReGen Villages, a Stanford University spinoff, is developing software tools to make the task easier.
Their core planning tool, VillageOS, can help conceive residential infrastructure incorporating everything from clean water systems and housing to renewable energy, organic food production, and even robotic and autonomous systems.
It’s like an industrial SimCity for regenerative living spaces.
VillageOS courtesy of ReGen Villages Holding, BV“Very often, architects, engineers, and planners prioritize maximizing building density or minimizing costs which focuses on profit rather than environmental impact or sustainability. VillageOS takes a different approach by first asking, ‘What is the land telling us?'” says Ehrlich.
In that sense, VillageOS is a high-tech listening device that can assess the land’s natural capacity and resource flows. It works by pulling in geospatial maps and then aggregating everything from historical data about climate and weather to soil maps and an array of local regulations, building codes, and permitting information. With the data, VillageOS uses generative design to blueprint community spaces that maximize any number of intended outcomes while minimizing its environmental footprint.
The goal is to design flourishing spaces that embed sustainable practices from the start.
A user who wants to enhance water resilience, for example, can set objectives like “maximizing rainwater storage” or “reducing runoff.” The software can then identify the best location to place a reservoir on a real parcel of land. It can do the same when designing housing and energy systems or choosing appropriate climate-resilient crops and where to harvest them.
The software’s user interface is key to the project. Built in Unreal Engine, it pulls 3D map data from Cesium and makes use of photorealistic, 3D visual renderings. By incorporating a game-like design with slider bars and controls, even non-technical users should be able to use the tool as easily as playing a video game.
“I get joy imagining that we can sit down with elderly farmers who own a piece of land, and without instruction watch them type in their address to load their land, start to get the climate data, and then explore the possibilities for what might be possible for that piece of land,” says Ehrlich.
One benefit of Unreal Engine is its ability to generate realistic lighting conditions in real time, a relatively new breakthrough that’s already having a dramatic impact on industries like real estate and film production. That means VillageOS users can plan and visualize exactly how a space would look and feel during different seasons and times of day or how foliage might cast shadows and change lighting conditions. It may seem trivial, but architects spend significant amounts of time exploring how lighting changes our use of public space.
The photorealism allows planners to communicate exactly how a resident can expect to experience a living space. The system could even allow customization, like setting a user’s height to that of a child so designers can take a variety of stakeholders into account.
Another element of VillageOS is its potential to serve as a digital twin and tool for managing a community’s ongoing operations. Digital twins, as I’ve written elsewhere, use a virtual replica of a real system to interactively engage with, ask questions of, or make predictions about that system. This might prove useful when deploying and managing autonomous robotic systems designed with ecovillages in mind, like drones or robotic fruit pickers.
“We’re going to see all kinds of robotic interventions capable of redirecting water, redirecting solar panels, and doing different kinds of autonomous interventions for the benefit of improving and refining the living conditions of these communities,” Ehrlich says.
The VillageOS software is still in development, but Ehrlich plans to release the climate data aggregator as an open-source API as soon as its ready. In the meantime, ReGen Villages is working with landowners and developers to train the VillageOS software.
System ResetThe scope of Ehrlich’s mission touches almost every aspect of how a society functions and addresses nearly all the UN Sustainable Development Goals. One of his work’s clearest ambitions is to curb the potential disruption from climate-related displacement and migration. Ehrlich sees a future where flourishing communities with socially affordable and climate-resilient housing developments reduce burdens on governments around the world and foster a mentally and physically healthy society—a big goal of his work at Stanford.
“Compassionate sustainability is about mindfulness—reducing the amygdala’s response related to cortisol release. How we live and where we live can actually improve health outcomes. There is a definite correlation between my work at Stanford and health outcomes based on this kind of design thinking.”
Living in small intentional communities might not only be an environmental solution to global challenges but could also make us happier and healthier. VillageOS might one day help us get to that better future.
Image Credit: ReGen Villages Holding, BV
This year, readers were again fascinated by stories about artificial intelligence. One algorithm learned to make short, playable video games from video footage; another cloned real people’s personalities; and yet another took on the role of your future self—ready and willing to impart wisdom.
Other popular pieces dug into the rise of robotaxis, tantalizing hints about how we might one day stave off aging, and a unified theory of consciousness.
We hope you enjoy a second look or discovering these for the first time.
As always, thanks for reading!
Credit: Google DeepMindA Google AI Watched 30,000 Hours of Video Games—Now It Makes Its Own
By Jason Dorrier
“As AI requires prodigious amounts of data, one way to forecast where things are going next is to look at what data is widely available online, but still largely untapped. Video, of which there is plenty, is an obvious next step. Indeed, [in February], OpenAI previewed a new text-to-video AI called Sora that stunned onlookers. But what about video…games?”
Scientists Are Working Towards a Unified Theory of Consciousness
By Shelly Fan
“Not all [the] authors agree on the specific brain mechanisms that allow us to perceive the outer world and construct an inner world of ‘self.’ But by collaborating, they merged their ideas, showing that different theories aren’t necessarily mutually incompatible—in fact, they could be consolidated into a general framework of consciousness and even inspire new ideas that help unravel one of the brain’s greatest mysteries.”
Scientists Extend Life Span in Mice by Restoring This Brain-Body Connection
By Shelly Fan
“Changing the protein’s behavior in aged mice with genetic engineering extended their life span by roughly seven percent. For an average 76-year life span in humans, the increase translates to over five years. The treatment also altered the mice’s health. Mice love to run, but their vigor plummets with age. Reactivating the neurons in elderly mice revived their motivation, transforming them from couch potatoes into impressive joggers.”
Credit: SpaceXIt Will Take Only a Single SpaceX Starship to Launch a Space Station
By Edd Gent
“SpaceX’s forthcoming Starship rocket will make it possible to lift unprecedented amounts of material into orbit. …Now, a joint venture between Airbus and Voyager Space that’s building a private space station called Starlab has inked a contract with SpaceX to get it into orbit. The venture plans to put the impressive capabilities of the new rocket to full use by launching the entire 26-foot-diameter space station in one go.”
OpenAI’s GPT-4o Makes AI Clones of Real People With Surprising Ease
By Edd Gent
“AI has become uncannily good at aping human conversational capabilities. New research suggests its powers of mimicry go a lot further, making it possible to replicate specific people’s personalities. …A study led by researchers at Stanford University has discovered that all it takes is a two-hour interview for an AI model to predict people’s responses to a battery of questionnaires, personality tests, and thought experiments with an accuracy of 85 percent.”
‘Droidspeak’: AI Agents Now Have Their Own Language Thanks to Microsoft
By Edd Gent
“Getting AIs to work together could be a powerful force multiplier for the technology. But despite their expressive power, human languages might not be the best medium of communication for machines that fundamentally operate in ones and zeros. This prompted researchers from Microsoft to develop a new method of communication that allows agents to talk to each other in the high-dimensional mathematical language underpinning LLMs. They’ve named the new approach Droidspeak…and in a preprint paper published on the arXiv, the Microsoft team reports it enabled models to communicate 2.78 times faster with little accuracy lost.”
Credit: Shawn Suttle from PixabayWaymo Robotaxis Are Giving 100,000 Rides a Week. It’ll Soon Be More.
By Jason Dorrier
“After a year of commercial operations in San Francisco without major incident, Waymo is eyeing expansion. In August, the company moved into Daly City, Broadmoor, and Colma, just south of the city. Waymo has approval to operate in a total of 22 cities along the peninsula south of San Francisco, and although there’s no timetable yet, according to the San Francisco Chronicle, they also have ambitions to add operations in San Jose and East Bay, which would include Oakland and Berkeley.”
A One-and-Done Injection to Slow Aging? New Study in Mice Opens the Possibility
By Shelly Fan
“A preventative anti-aging therapy seems like wishful thinking. Yet a new study led by Dr. Corina Amor Vegas at Cold Spring Harbor Laboratory describes a treatment that brings the dream to life—at least for mice. Given a single injection in young adulthood, they aged more slowly compared to their peers. …’If we give it to aged mice, they rejuvenate. If we give it to young mice, they age slower. No other therapy right now can do this,’ said Amor Vegas in a press release.”
These Mini AI Models Match OpenAI With 1,000 Times Less Data
By Jason Dorrier
“The artificial intelligence industry is obsessed with size. Bigger algorithms. More data. Sprawling data centers that could, in a few years, consume enough electricity to power whole cities. …Eye-popping numbers like these make it easy to forget size isn’t everything. Some researchers, particularly those with fewer resources, are aiming to do more with less. AI scaling will continue, but algorithms will also get far more efficient as they grow.”
This MIT Chatbot Simulates Your ‘Future Self.’ It’s Here to Help You Make Better Decisions.
By Jason Dorrier
“Chatbots are now posing as friends, romantic partners, and departed loved ones. Now, we can add another to the list: Your future self. MIT Media Lab’s Future You project invited young people, aged 18 to 30, to have a chat with AI simulations of themselves at 60. The sims—which were powered by a personalized chatbot and included an AI-generated image of their older selves—answered questions about their experience, shared memories, and offered lessons learned over the decades.”
Banner Image Credit: Luke Jones on Unsplash
Despite the hype around AI in recent years, the technology’s disruptive impact has been fairly modest. Experts say that’s likely to change next year as AI agents force their way into all aspects of our lives.
Since the surprise success of ChatGPT in late 2022, billions of dollars have poured into the AI industry as startups and big tech firms try to capitalize on the unquestionable promise of the technology.
But while hundreds of millions of people around the world are now regularly using AI chatbots, putting them to productive use is proving harder. Recent research from Boston Consulting Group found that just 26 percent of companies who have experimented with AI have moved past proof of concept to get real value out of the technology.
That could be because current iterations of the technology are, at best, a kind of copilot. They can help users accomplish some tasks more efficiently, but only with close supervision and the ever-present risk of mistakes. The situation could be about to change though, according to leading voices in the AI industry, who say that autonomous AI agents are poised to have a breakout year in 2025.
“For the first time, technology isn’t just offering tools for humans to do work,” Salesforce CEO Marc Benioff recently wrote in Time, a publication he owns. “It’s providing intelligent, scalable digital labor that performs tasks autonomously. Instead of waiting for human input, agents can analyze information, make decisions, and take action independently, adapting and learning as they go.”
At the core of all AI agents is the same kind of large language model (LLM) that powers services like ChatGPT. This makes it possible for humans to interact with agents via language, but the algorithm is also a “reasoning engine” that comes up with a step-by-step plan to tackle tasks.
Agents also typically have access to external data sources relevant to their application—for instance customer databases or financial records—and software tools they can use to achieve goals.
At present, the reasoning capabilities of LLMs are limited, which restricts where agents can be deployed. But with the advent of models like OpenAI’s o1 and DeepSeek’s R1, which are specialist reasoning models, there’s hope that agents could soon become much more capable.
Major players are investing heavily in that promise.
In October, Microsoft unveiled Copilot Studio, which allows companies to build customized agents capable of tasks like handling client queries and identifying sales leads. The same month, Salesforce rolled out its Agentforce platform, which also allows customers to create their own bots. And last month, Benioff told TechCrunch his goal is to have one billion agents deployed within a year.
Leading AI research labs are also increasingly focused on agents. Anthropic recently previewed a version of its Claude 3.5 Sonnet model that could take control of a user’s computer, and Google’s recently announced Gemini 2 has been trained to perform similar tasks. OpenAI also has plans to unveil an agent codenamed “Operator” early in the new year.
Startups are looking to get in on the action too. According to Pitchbook, the number of funding deals for agent-focused ventures was up more than 80 percent by September compared to the previous year. The median deal value was also up nearly 50 percent.
But there is some skepticism around how quickly agents are likely to burst onto the scene. As The Verge notes, AI companies have been ploughing billions into research and development with little revenue to show for it and are still searching for a killer app that justifies their sky-high valuations. Practical considerations could mean progress is slower than they hope.
For a start, these models are still prone to “hallucinations” where they generate incorrect or misleading responses to queries. This is problematic enough in a chatbot but much more concerning when it’s an agent capable of independent action.
Quartz notes this risk can create considerable overhead as companies have to implement many layers of security designed to catch mistakes. This could become incredibly complex as the number of agents increases and require investment in new platforms and even “guardian agents” to monitor their activities.
Agents can also be expensive because “reasoning” through problems requires they make multiple calls to the underlying LLM. This quickly adds up, either in terms of dollars spent with an LLM-provider or energy burned for companies that host their own models.
Nonetheless, many in the industry expect 2025 will be a turning point in deployment.
“I think 2025 is going to be the year that agentic systems finally hit the mainstream,” OpenAI’s new chief product officer, Kevin Weil, said at a press event ahead of the company’s annual Dev Day, according to The Verge.
Deloitte’s Global 2025 Predictions Report forecasts that of the companies already using generative AI, a quarter will launch pilots or proofs of concept with AI agents, growing to half by 2027. And the second half of the year could see full adoption of agents in some workflows.
Others are more bullish. Konstantine Buhler of Sequoia Capital told Bloomberg that 2025 will see the emergence of networks or “swarms” of AI agents working together within businesses. Kari Briski, vice president of generative AI software at Nvidia, agrees and thinks this will necessitate the emergence of AI orchestrators—essentially AI managers that oversee and coordinate numerous agents.
No matter who’s right, it seems certain that agents will be the major preoccupation of the AI industry in 2025. If it pays off, the world of work could look very different by the end of the year.
Image Credit: Gabriella Clare Marino on Unsplash
When human stem cells were discovered at the turn of the century, it sparked a frenzy. Scientists immediately dreamed of repairing damaged tissues due to aging or disease.
A few decades later, their dreams are on the brink of coming true. The US Food and Drug Administration (FDA) has approved blood stem cell transplantation for cancer and other disorders that affect the blood and immune system. More clinical trials are underway, investigating the use of stem cells from the umbilical cord to treat knee osteoarthritis—where the cartilage slowly wears down—and nerve problems from diabetes.
But the promise of stem cells came with a dark side.
Illegal stem cell clinics popped up soon after the cells’ discovery, touting their ability to rejuvenate aged skin, joints, or even treat severe brain disorders such as Parkinson’s disease. Despite FDA regulation, as of 2021, there were nearly 2,800 unlicensed clinics across the country, each advertising stem cells therapies with little scientific evidence.
“What started as a trickle became a torrent as businesses poured into this space,” wrote an expert team in the journal Cell Stem Cell in 2021.
History is now repeating itself with an up-and-coming “cure-all:” exosomes.
Exosomes are tiny bubbles made by cells to carry proteins and genetic material to other cells. While still early, research into these mysterious bubbles suggests they may be involved in aging or be responsible for cancers spreading across the body.
Multiple clinical trials are underway, ranging from exosome therapies to slow hair loss to treatments for heart attacks, strokes, and bone and cartilage loss. They have potential.
But a growing number of clinics are also advertising exosomes as their next best seller. One forecast analyzing exosomes in the skin care industry predicts a market value of over $674 million by 2030.
The problem? We don’t really know what exosomes are, what they do to the body, or their side effects. In a way, these molecular packages are like Christmas “mystery boxes,” each containing a different mix of biological surprises that could alter cellular functions, like turning genes on or off in unexpected ways.
There have already been reports of serious complications. “There is an urgent need to develop regulations to protect patients from serious risks associated with interventions based on little or no scientific evidence,” a team recently wrote in Stem Cell Reports.
Cellular Space ShuttlesIn 1996, Graça Raposo, a molecular scientist in the Netherlands, noticed something strange: The immune cells she was studying seemed to send messages to each other in tiny bubbles. Under the microscope, she saw that when treated with a “toxin” of sorts, the cells slurped up the molecules, planted them on the surfaces of tiny bubbles inside the cell, and released the bubbles into the vast wilderness of the cell’s surroundings.
She collected the bubbles and squirted them onto other immune cells. Surprisingly, they triggered a similar immune response in the cells—as if directly exposed to the toxin. In other words, the bubbles seemed to shuttle information between cells.
Dubbed exosomes, scientists previously thought they were the cell’s garbage collectors, gathering waste molecules into a bubble and spewing it outside the cell. But two years later, Raposo and colleagues found that exosomes harvested from cells that naturally fight off tumors could be used as a therapy to suppress tumors in mice.
Interest in these mysterious blobs exploded.
Scientists soon found that most cells pump out exosome “spaceships,” and they can contain both proteins and types of RNA that turn genes on or off. But despite decades of research, we’re only scratching the surface of what cargo they can carry and their biological function.
It’s still unclear what exosomes do. Some could be messengers of a dying cell, warning neighbors to shore up defenses. They could also be co-opted by tumor cells to bamboozle nearby cells into supporting cancer growth and spread. In Alzheimer’s disease, they could potentially shuttle twisted protein clumps to other cells, spreading the disease across the brain.
They’re tough to study, in part, because they’re so small and unpredictable. About one-hundredth the size of a red blood cell, exosomes are hard to capture even with modern microscopy. Each type of cell seems to have a different release schedule, with some spewing many in one shot and others taking the slow-and-steady route. Until recently, scientists didn’t even agree on how to define exosomes.
Over several years, the International Society for Extracellular Vesicles, or exosomes, has begun uniting the field with naming conventions and standardized methods for preparing exosomes.
The Wild WestWhile scientists are rapidly coming together to cautiously make exosome-based treatment a reality, uncertified clinics have popped up across the globe. Their first pitch to the public was tackling Covid. One analysis found 60 clinics in the US advertising exosome-based therapy as a way to prevent or treat the virus—with zero scientific support. Another trending use has been in skin care or hair growth, garnering attention in the US, UK, and Japan.
Exosomes are regulated by the FDA in the US and the European Medicines Agency (EMA) in the EU as biological medicinal products, meaning they require approval from the agencies. That did not stop clinics from marketing them, with tragic consequences. In 2019, patients in Nebraska treated with unapproved exosomes became septic—a life-threating condition caused by infection across the whole body—leading the FDA to issue a warning.
Clinics that offer unregulated exosomes “deceive patients with unsubstantiated claims about the potential for these products to prevent, treat, or cure various diseases or conditions,” the agency wrote.
Japan is struggling to catch up. Exosomes are not regulated under their laws. Nearly 670 clinics have already popped up, representing a far larger market than the US or EU. Most services have been marketed for skin care, anti-aging, hair growth, and battling fatigue, wrote the authors. More rarely, some touted their ability to battle cancers.
The rogue clinics have already led to tragedies. In one case, “a well-known private cosmetic surgery clinic administered exosomes…to at least four patients, including relatives of staff members with stage IV lung cancer, and found that the cancer rapidly worsened after administration,” wrote the authors.
Because the clinics operate on the down-low, it’s tough to gauge the extent of harm, including potential deaths.
The worry isn’t that exosomes are harmful by themselves. How they’re obtained plays a huge role in safety. In unregulated settings, there’s a large chance of the bubbles being contaminated by endotoxins—which trigger dangerous inflammatory responses—or bacteria that lingers and grows.
For now, “from a very basic point of view, we don’t really know what they’re doing, good or bad… I wouldn’t take them, let’s put it that way,” James Edgar, an exosome researcher from the University of Cambridge, told MIT Technology Review.
Unregulated clinics don’t just harm patients. They could also set a promising field back.
Scientific advances may seem to move at a snail’s pace, but it’s to ensure safety and efficacy despite the glitz and glamor of a potential new panacea. Scientists are still forging ahead using exosomes for multiple health problems—while bearing in mind there’s much we still need to understand about these cellular spaceships.
Image Credit: Steve Johnson on Unsplash
We have only one example of biology forming in the universe—life on Earth. But what if life can form in other ways? How do you look for alien life when you don’t know what alien life might look like?
These questions are preoccupying astrobiologists—scientists who look for life beyond Earth. Astrobiologists have attempted to come up with universal rules that govern the emergence of complex physical and biological systems both on Earth and beyond.
I’m an astronomer who has written extensively about astrobiology. Through my research, I’ve learned that the most abundant form of extraterrestrial life is likely to be microbial, since single cells can form more readily than large organisms. But just in case there’s advanced alien life out there, I’m on the international advisory council for the group designing messages to send to those civilizations.
Detecting Life Beyond EarthSince the first discovery of an exoplanet in 1995, over 5,000 exoplanets, or planets orbiting other stars, have been found.
Many of these exoplanets are small and rocky, like Earth, and in the habitable zones of their stars. The habitable zone is the range of distances between the surface of a planet and the star it orbits that would allow the planet to have liquid water and thus support life as we on Earth know it.
The sample of exoplanets detected so far projects 300 million potential biological experiments in our galaxy—or 300 million places, including exoplanets and other bodies such as moons, with suitable conditions for biology to arise.
The uncertainty for researchers starts with the definition of life. It feels like defining life should be easy, since we know life when we see it, whether it’s a flying bird or a microbe moving in a drop of water. But scientists don’t agree on a definition, and some think a comprehensive definition might not be possible.
NASA defines life as a “self-sustaining chemical reaction capable of Darwinian evolution.” That means organisms with a complex chemical system that evolve by adapting to their environment. Darwinian evolution says that the survival of an organism depends on its fitness in its environment.
The evolution of life on Earth has progressed over billions of years from single-celled organisms to large animals and other species, including humans.
Exoplanets are remote and hundreds of millions of times fainter than their parent stars, so studying them is challenging. Astronomers can inspect the atmospheres and surfaces of Earth-like exoplanets using a method called spectroscopy to look for chemical signatures of life.
Spectroscopy might detect signatures of oxygen in a planet’s atmosphere, which microbes called blue-green algae created by photosynthesis on Earth several billion years ago, or chlorophyll signatures, which indicate plant life.
NASA’s definition of life leads to some important but unanswered questions. Is Darwinian evolution universal? What chemical reactions can lead to biology off Earth?
Evolution and ComplexityAll life on Earth, from a fungal spore to a blue whale, evolved from a microbial last common ancestor about four billion years ago.
The same chemical processes are seen in all living organisms on Earth, and those processes might be universal. They also may be radically different elsewhere.
In October 2024, a diverse group of scientists gathered to think outside the box on evolution. They wanted to step back and explore what sorts of processes created order in the universe—biological or not—to figure out how to study the emergence of life totally unlike life on Earth.
Two researchers present argued that complex systems of chemicals or minerals, when in environments that allow some configurations to persist better than others, evolve to store larger amounts of information. As time goes by, the system will grow more diverse and complex, gaining the functions needed for survival, through a kind of natural selection.
Minerals are an example of a nonliving system that has increased in diversity and complexity over billions of years. Image Credit: Doug Bowman, CC BYThey speculated that there might be a law to describe the evolution of a wide variety of physical systems. Biological evolution through natural selection would be just one example of this broader law.
In biology, information refers to the instructions stored in the sequence of nucleotides on a DNA molecule, which collectively make up an organism’s genome and dictate what the organism looks like and how it functions.
If you define complexity in terms of information theory, natural selection will cause a genome to grow more complex as it stores more information about its environment.
Complexity might be useful in measuring the boundary between life and non-life.
However, it’s wrong to conclude that animals are more complex than microbes. Biological information increases with genome size, but evolutionary information density drops. Evolutionary information density is the fraction of functional genes within the genome, or the fraction of the total genetic material that expresses fitness for the environment.
Organisms that people think of as primitive, such as bacteria, have genomes with high information density and so appear better designed than the genomes of plants or animals.
A universal theory of life is still elusive. Such a theory would include the concepts of complexity and information storage, but it would not be tied to DNA or the particular kinds of cells we find in terrestrial biology.
Implications for the Search for Extraterrestial LifeResearchers have explored alternatives to terrestrial biochemistry. All known living organisms, from bacteria to humans, contain water, and it is a solvent that is essential for life on Earth. A solvent is a liquid medium that facilitates chemical reactions from which life could emerge. But life could potentially emerge from other solvents, too.
Astrobiologists Willam Bains and Sara Seager have explored thousands of molecules that might be associated with life. Plausible solvents include sulfuric acid, ammonia, liquid carbon dioxide, and even liquid sulfur.
Alien life might not be based on carbon, which forms the backbone of all life’s essential molecules—at least here on Earth. It might not even need a planet to survive.
Advanced forms of life on alien planets could be so strange that they’re unrecognizable. As astrobiologists try to detect life off Earth, they’ll need to be creative.
One strategy is to measure mineral signatures on the rocky surfaces of exoplanets, since mineral diversity tracks terrestrial biological evolution. As life evolved on Earth, it used and created minerals for exoskeletons and habitats. The hundred minerals present when life first formed have grown to about 5,000 today.
For example, zircons are simple silicate crystals that date back to the time before life started. A zircon found in Australia is the oldest known piece of Earth’s crust. But other minerals, such as apatite, a complex calcium phosphate mineral, are created by biology. Apatite is a primary ingredient in bones, teeth, and fish scales.
Another strategy for finding life unlike that on Earth is to detect evidence of a civilization, such as artificial lights, or the industrial pollutant nitrogen dioxide in the atmosphere. These are examples of tracers of intelligent life called technosignatures.
It’s unclear how and when a first detection of life beyond Earth will happen. It might be within the solar system, or by sniffing exoplanet atmospheres, or by detecting artificial radio signals from a distant civilization.
The search is a twisting road, not a straightforward path. And that’s for life as we know it—for life as we don’t know it, all bets are off.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA’s Goddard Space Flight Center/Francis Reddy
COMPUTINGIBM Boosts the Amount of Computation You Can Get Done on Quantum Hardware
John Timmer | Ars Technica“There’s a general consensus that we won’t be able to consistently perform sophisticated quantum calculations without the development of error-corrected quantum computing, which is unlikely to arrive until the end of the decade. It’s still an open question, however, whether we could perform limited but useful calculations at an earlier point. IBM is one of the companies that’s betting the answer is yes, and on Wednesday, it announced a series of developments aimed at making that possible.”
ARTIFICIAL INTELLIGENCEOpenAI Shifts Strategy as Rate of ‘GPT’ AI Improvements Slows
Stephanie Palazzolo, Erin Woo, and Emir Efrati | The Information“”The Orion situation could test a core assumption of the AI field, known as scaling laws: that LLMs would continue to improve at the same pace as long as they had more data to learn from and additional computing power to facilitate that training process. In response to the recent challenge to training-based scaling laws posed by slowing GPT improvements, the industry appears to be shifting its effort to improving models after their initial training, potentially yielding a different type of scaling law.”
BIOTECHThe First CRISPR Treatment Is Making Its Way to Patients
Emily Mullen | Wired“Vertex, the pharmaceutical company that markets Casgevy, announced in a November 5 earnings call that the first person to receive Casgevy outside of a clinical trial was dosed in the third quarter of this year. …When Wired followed up with Vertex via email, spokesperson Eleanor Celeste declined to provide the exact number of patients that have received Casgevy. However, the company says 40 patients have undergone cell collections in anticipation of receiving the treatment, up from 20 patients last quarter.”
AUTOMATIONAI Is Now Designing Chips for AI
Kristen Houser | Big Think“It’s 2028, and your tech startup has an idea that could revolutionize the industry—but you need a custom designed microchip to bring the product to market. Five years ago, designing that chip would’ve cost more than your whole company is worth, but your team is now able to do it at a fraction of price and in a fraction of the time—all thanks to AI, fittingly being run on chips like these.”
ROBOTICSNow Anyone in LA Can Hail a Waymo Robotaxi
Kirsten Korosec | TechCrunch“Waymo has opened its robotaxi service to everyone in Los Angeles, sunsetting a waitlist that had grown to 300,000 people. The Alphabet-backed company said starting Tuesday anyone can download the Waymo One app to hail a ride in its service area, which is now about 80 square miles in Los Angeles County.”
ARTIFICAL INTELLIGENCEThe First Entirely AI-Generated Video Game Is Insanely Weird and Fun
Will Knight | Wired“Minecraft remains remarkably popular a decade or so after it was first released, thanks to a unique mix of quirky gameplay and open world building possibilities. A knock-off called Oasis, released last month, captures much of the original game’s flavor with a remarkable and weird twist. The entire game is generated not by a game engine and hand-coded rules, but by an AI model that dreams up each frame.”
ENERGYNuclear Power Was Once Shunned at Climate Talks. Now, It’s a Rising Star.
Brad Plumer | The New York Times“At last year’s climate conference in the United Arab Emirates, 22 countries pledged, for the first time, to triple the world’s use of nuclear power by midcentury to help curb global warming. At this year’s summit in Azerbaijan, six more countries signed the pledge. ‘It’s a whole different dynamic today,’ said Dr. Bilbao y Leon, who now leads the World Nuclear Association, an industry trade group. ‘A lot more people are open to talking about nuclear power as a solution.'”
HEALTHThe Next Omics? Tracking a Lifetime of Exposures to Better Understand Disease
Lindzi Wessel | Knowable Magazine“Of the millions of substances people encounter daily, health researchers have focused on only a few hundred. Those in the emerging field of exposomics want to change that. …In homes, on buildings, from satellites and even in apps on the phone in your pocket, tools to monitor the environment are on the rise. At the intersection of public health and toxicology, these tools are fueling a new movement in exposure science. It’s called the exposome and it represents the sum of all environmental exposures over a lifetime.”
SPACEBuckle Up: SpaceX Aims for Rapid-Fire Starship Launches in 2025
Passant Rabie | Gizmodo“SpaceX has big plans for its Starship rocket. After a groundbreaking test flight, in which the landing tower caught the booster, the company’s founder and CEO Elon Musk wants to see the megarocket fly up to 25 times next year, working its way up to a launch rate of 100 flights per year, and eventually a Starship launching on a daily basis.”
TECHAre AI Clones the Future of Dating? I Tried Them for Myself.
Eli Tan | The New York Times“As chatbots like ChatGPT improve, their use in our personal and even romantic lives is becoming more common. So much so, some executives in the dating app industry have begun pitching a future in which people can create AI clones of themselves that date other clones and relay the results back to their human counterparts.”
GENETICSGenetic Discrimination Is Coming for Us All
Kristen V. Brown | The Atlantic“For decades, researchers have feared that people might be targeted over their DNA, but they weren’t sure how often it was happening. Now at least a handful of Americans are experiencing what they argue is a form of discrimination. And as more people get their genomes sequenced—and researchers learn to glean even more information from the results—a growing number of people may find themselves similarly targeted.”
Image Credit: Evgeni Tcherkasski on Unsplash
A big challenge when training AI models to control robots is gathering enough realistic data. Now, researchers at MIT have shown they can train a robot dog using 100 percent synthetic data.
Traditionally, robots have been hand-coded to perform particular tasks, but this approach results in brittle systems that struggle to cope with the uncertainty of the real world. Machine learning approaches that train robots on real-world examples promise to create more flexible machines, but gathering enough training data is a significant challenge.
One potential workaround is to train robots using computer simulations of the real world, which makes it far simpler to set up novel tasks or environments for them. But this approach is bedeviled by the “sim-to-real gap”—these virtual environments are still poor replicas of the real world and skills learned inside them often don’t translate.
Now, MIT CSAIL researchers have found a way to combine simulations and generative AI to enable a robot, trained on zero real-world data, to tackle a host of challenging locomotion tasks in the physical world.
“One of the main challenges in sim-to-real transfer for robotics is achieving visual realism in simulated environments,” Shuran Song from Stanford University, who wasn’t involved in the research, said in a press release from MIT.
“The LucidSim framework provides an elegant solution by using generative models to create diverse, highly realistic visual data for any simulation. This work could significantly accelerate the deployment of robots trained in virtual environments to real-world tasks.”
Leading simulators used to train robots today can realistically reproduce the kind of physics robots are likely to encounter. But they are not so good at recreating the diverse environments, textures, and lighting conditions found in the real world. This means robots relying on visual perception often struggle in less controlled environments.
To get around this, the MIT researchers used text-to-image generators to create realistic scenes and combined these with a popular simulator called MuJoCo to map geometric and physics data onto the images. To increase the diversity of images, the team also used ChatGPT to create thousands of prompts for the image generator covering a huge range of environments.
After generating these realistic environmental images, the researchers converted them into short videos from a robot’s perspective using another system they developed called Dreams in Motion. This computes how each pixel in the image would shift as the robot moves through an environment, creating multiple frames from a single image.
The researchers dubbed this data-generation pipeline LucidSim and used it to train an AI model to control a quadruped robot using just visual input. The robot learned a series of locomotion tasks, including going up and down stairs, climbing boxes, and chasing a soccer ball.
The training process was split into parts. First, the team trained their model on data generated by an expert AI system with access to detailed terrain information as it attempted the same tasks. This gave the model enough understanding of the tasks to attempt them in a simulation based on the data from LucidSim, which generated more data. They then re-trained the model on the combined data to create the final robotic control policy.
The approach matched or outperformed the expert AI system on four out of the five tasks in real-world tests, despite relying on just visual input. And on all the tasks, it significantly outperformed a model trained using “domain randomization”—a leading simulation approach that increases data diversity by applying random colors and patterns to objects in the environment.
The researchers told MIT Technology Review their next goal is to train a humanoid robot on purely synthetic data generated by LucidSim. They also hope to use the approach to improve the training of robotic arms on tasks requiring dexterity.
Given the insatiable appetite for robot training data, methods like this that can provide high-quality synthetic alternatives are likely to become increasingly important in the coming years.
Image Credit: MIT CSAIL
When I was a young kid, our neighborhood didn’t have any grocery stores. The only place to buy fruits and vegetables was at our local farmer’s market. My mom would pick out the freshest tomatoes and sauté them with eggs into a simple dish that became my comfort food.
The tomatoes were hideous to look at—small, gnarled, miscolored, and nothing like the perfectly plump and bright beefsteak or Roma tomatoes that eventually flooded supermarkets. But they were oh-so-tasty, with a perfect ratio of tart and sweet flavors that burst in my mouth.
These days, when I ask for the same dish, my mom will always say, “Tomatoes just don’t taste the same anymore.”
She’s not alone. Many people have noticed that today’s produce is watery, waxy, and lacking in flavor—despite looking ripe and inviting. One reason is it was bred that way. Today’s crops are often genetically selected to prioritize appearance, size, shelf life, and transportability. But these perks can sacrifice taste—most often, in the form of sugar. Even broccoli, known for its bitterness, has variants that accumulate sugar inside their stems for a slightly sweeter taste.
The problem is that larger fruit sizes are often less sweet, explains Sanwen Huang and colleagues in Shenzhen, China. The key is to break that correlation. His team may have found a way using a globally popular crop—the tomato—as an example.
By comparing wild and domesticated tomatoes, the team hunted down a set of genes that put the brakes on sugar production. Inhibiting those genes using CRISPR-Cas9, the popular gene-editing tool, bumped up the fruit’s sugar content by 30 percent—enough for a consumer panel to find a noticeable increase in sweetness—without sacrificing size or yields.
Seeds from the edited plants germinated as usual, allowing the edits to pass on to the next generations.
The study isn’t just about satisfying our sweet tooth. Crops, not just tomatoes, with higher sugar content also contain more calories, which are necessary if we’re to meet the needs of a growing global population. The analysis pipeline established in the study is set to identify other genetic trade-offs between size and nutrition, with the goal of rapidly engineering better crops.
The work “represents an exciting step forward…for crop improvement worldwide,” wrote Amy Lanctot and Patrick Shih at the University of California, Berkeley, who were not involved in the study.
Hot LinksFor eons, humanity has cultivated crops to enhance desirable aspects—for example, better yields, higher nutrition, or looks.
Tomatoes are a perfect example. The fruit “is the most valuable vegetable crop, worldwide, and makes substantial overall health and nutritional contributions to the human diet,” wrote the team. Its wild versions range in size from cherries to peas—far smaller than most current variants found in grocery stores. Flavor comes from two types of sugars packed in their solid bits.
After thousands of years of domestication, sugars remain the key ingredient to better-tasting tomatoes. But in recent decades, breeders mostly prioritized increasing fruit size. The result are tomatoes that are easily sliced for sandwiches, crushed for canning, or further processed into sauces or pastes. Compared to their wild ancestors, today’s cultivated tomatoes are roughly between 10 to 100 times larger in size, making them far more economical.
But these improvements come a cost. Multiple studies have found that as size goes up, sugar levels and flavor tank. A similar trend has also been found in other large farming fruits.
Ever since, scientists have tried teasing out the tomato’s inner workings—especially genes that produces sugar—to restore its taste and nutritious value. One study in 2017 combined genomic analysis of nearly 400 varieties of tomatoes with results from a human taste panel to home in on a slew of metabolic chemicals that made the fruit taste better. A year later, Huang’s team, who led the new study, analyzed the genetic makeup and cell function of hundreds of tomato types. Domestication was associated with several large changes in the plant’s genome—but the team didn’t know how each genetic mutation altered the fruit’s metabolism.
It’s tough to link a gene to a trait. Our genes, as DNA strands, are tightly wound into mostly X-shaped chromosomes. Like braided balls of yarn, these 3D structures bring genes normally separated on a linear strand into close proximity. This means nearby, or “linked,” genes often turn on or off together.
“Genetic linkage makes it difficult to alter one gene without affecting the other,” wrote Lanctot and Shih.
Fast Track EvolutionThe new study used two technologies to overcome the problem.
The first was cheaper genetic sequencing. By scanning through genetic variations between domesticated and wild tomatoes, the team pinpointed six tomato genes likely responsible for the fruit’s sweetness.
One gene especially caught their eye. It was turned off in sweeter tomato species, putting the brakes on the plants’ ability to accumulate sugar. Using the gene-editing tool CRISPR-Cas9, the team mutated the gene so it could no longer function and grew the edited species—along with normal ones—under the same conditions in a garden.
The Sweet SpotRoughly 100 volunteers tried the edited and normal tomatoes in a blind trial. The CRISPRed tomatoes won in a landslide for their perceived sweetness.
The study isn’t just about a better tomato. “This research demonstrates the value hidden in the genomes of crop species varieties and their wild relatives,” wrote Lanctot and Shih.
Domestication, while boosting yield or size of a fruit, often decreases genetic diversity for a species because selected crops eventually contain mostly the same genetic blueprint. Some crops, such as bananas, can’t reproduce on their own and are extremely vulnerable to fungi. Analyzing genes related to these traits could help form a defense strategy.
Conservation and taste aside, scientists have also tried to endow crops with more exotic traits. In 2021, Sanatech Seed, a company based in Japan, engineered tomatoes using CRISPR-Cas9 to increase the amount of a chemical that dampens neural transmission. According to the company, the tomatoes can lower blood pressure and help people relax. The fruit is already on the market following regulatory approval in Japan.
Studies that directly link a gene to a trait in plants are still extremely rare. Thanks to cheaper and faster DNA sequencing technologies, and increasingly precise CRISPR tools, it’s becoming easier to test these connections.
“The more researchers understand about the genetic pathways underlying these trade-offs, the more they can take advantage of modern genome-editing tools to attempt to disentangle them to boost crucial agricultural traits,” wrote Lanctot and Shih.
Image Credit: Thomas Martinsen on Unsplash
In the 2016 science fiction movie Arrival, a linguist is faced with the daunting task of deciphering an alien language consisting of palindromic phrases, which read the same backwards as they do forwards, written with circular symbols. As she discovers various clues, different nations around the world interpret the messages differently—with some assuming they convey a threat.
If humanity ended up in such a situation today, our best bet may be to turn to research uncovering how artificial intelligence develops languages.
But what exactly defines a language? Most of us use at least one to communicate with people around us, but how did it come about? Linguists have been pondering this very question for decades, yet there is no easy way to find out how language evolved.
Language is ephemeral, it leaves no examinable trace in the fossil records. Unlike bones, we can’t dig up ancient languages to study how they developed over time.
While we may be unable to study the true evolution of human language, perhaps a simulation could provide some insights. That’s where AI comes in—a fascinating field of research called emergent communication, which I have spent the last three years studying.
To simulate how language may evolve, we give AI agents simple tasks that require communication, like a game where one robot must guide another to a specific location on a grid without showing it a map. We provide (almost) no restrictions on what they can say or how—we simply give them the task and let them solve it however they want.
Because solving these tasks requires the agents to communicate with each other, we can study how their communication evolves over time to get an idea of how language might evolve.
Similar experiments have been done with humans. Imagine you, an English speaker, are paired with a non-English speaker. Your task is to instruct your partner to pick up a green cube from an assortment of objects on a table.
You might try to gesture a cube shape with your hands and point at grass outside the window to indicate the color green. Over time, you’d develop a sort of proto-language together. Maybe you’d create specific gestures or symbols for “cube” and “green.” Through repeated interactions, these improvised signals would become more refined and consistent, forming a basic communication system.
This works similarly for AI. Through trial and error, algorithms learn to communicate about objects they see, and their conversation partners learn to understand them.
But how do we know what they’re talking about? If they only develop this language with their artificial conversation partner and not with us, how do we know what each word means? After all, a specific word could mean “green,” “cube,” or worse—both. This challenge of interpretation is a key part of my research.
Cracking the CodeThe task of understanding AI language may seem almost impossible at first. If I tried speaking Polish (my mother tongue) to a collaborator who only speaks English, we couldn’t understand each other or even know where each word begins and ends.
The challenge with AI languages is even greater, as they might organize information in ways completely foreign to human linguistic patterns.
Fortunately, linguists have developed sophisticated tools using information theory to interpret unknown languages.
Just as archaeologists piece together ancient languages from fragments, we use patterns in AI conversations to understand their linguistic structure. Sometimes we find surprising similarities to human languages, and other times we discover entirely novel ways of communication.
These tools help us peek into the “black box” of AI communication, revealing how AI agents develop their own unique ways of sharing information.
My recent work focuses on using what the agents see and say to interpret their language. Imagine having a transcript of a conversation in a language unknown to you, along with what each speaker was looking at. We can match patterns in the transcript to objects in the participant’s field of vision, building statistical connections between words and objects.
For example, perhaps the phrase “yayo” coincides with a bird flying past—we could guess that “yayo” is the speaker’s word for “bird.” Through careful analysis of these patterns, we can begin to decode the meaning behind the communication.
In the latest paper by me and my colleagues, set to appear in the conference proceedings of Neural Information Processing Systems (NeurIPS), we show that such methods can be used to reverse-engineer at least parts of the AIs’ language and syntax, giving us insights into how they might structure communication.
Aliens and Autonomous SystemsHow does this connect to aliens? The methods we’re developing for understanding AI languages could help us decipher any future alien communications.
If we are able to obtain some written alien text together with some context (such as visual information relating to the text), we could apply the same statistical tools to analyze them. The approaches we’re developing today could be useful tools in the future study of alien languages, known as xenolinguistics.
But we don’t need to find extraterrestrials to benefit from this research. There are numerous applications, from improving language models like ChatGPT or Claude to improving communication between autonomous vehicles or drones.
By decoding emergent languages, we can make future technology easier to understand. Whether it’s knowing how self-driving cars coordinate their movements or how AI systems make decisions, we’re not just creating intelligent systems—we’re learning to understand them.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Tomas Martinez on Unsplash
“We’re only just beginning to understand the full majesty of life on Earth,” wrote the founding members of the Earth BioGenome Project in 2018. The ambitious project raised eyebrows when first announced. It seeks to genetically profile over a million plants, animals, and fungi. Documenting these genomes is the first step to building an atlas of complex life on Earth.
Many living species remain mysterious to science. A database resulting from the project would be a precious resource for monitoring biodiversity. It could also shed light on the genetic “dark matter” of complex life to inspire new biomaterials, medicines, or spark ideas for synthetic biology. Other insights could tailor agricultural practices to ramp up food production and feed a growing global population.
In other words, digging into living creatures’ genetic data is set to unveil “unimaginable biological secrets,” wrote the team.
The problem? A hefty price tag. With an estimated cost of $4.7 billion, even the founders of the project called it a moonshot. However, against all odds, the project has made progress, with 3,000 genomes already sequenced and 10,000 more species expected by 2026.
While lagging its original goal of sequencing roughly 1.7 million genomes in a decade, the project still hopes to hit this goal by 2032—later than the original goalpost, but with a much lower price tag thanks to more efficient DNA sequencing technologies.
Meanwhile, the international team has also built infrastructure to share gene sequencing data, and machine learning methods are further helping the consortium analyze thousands of datasets—helping characterize new species and monitor DNA data for endangered ones.
Expanding the ScopeGenetic material is everywhere. It’s an abundant resource to make sense of life of Earth. As genetic sequencing becomes faster, cheaper, and more reliable, recent studies have begun digging into information represented by DNA from species across the globe.
One method, dubbed metagenomics, captures and analyzes microbial DNA gathered in a variety of environments, from city sewers to boiling hot springs. The method captures and analyzes all DNA from a particular source to paint a broad genetic picture of bacteria from a given environment. Rather than bacteria, the Earth BioGenome Project, or EBP, is aiming to sequence the genomes of individual eukaryotic creatures—basically, those that keep most of their DNA in a nut-like structure, or nucleus, inside each cell.
Humans, plants, fungi, and other animals all fall into this group. In one estimate, there are roughly 10 to 15 million eukaryotic species on our planet. But just a little over two million have been documented.
Sequencing DNA from eukaryotic cells could vastly expand our knowledge of Earth’s genetic diversity. Such a database could also be a treasure trove for synthetic biology. Scientists have already tinkered with the genetic blueprints of life in bacteria and yeast cells. Deciphering—and then reprogramming—their genes has led to advances such as coaxing bacteria cells to pump out biofuels, degradable materials, and medicines such as insulin.
Charting eukaryotes’ genomes could further inspire new materials or medicines. For example, cytarabine, a chemotherapy drug, was initially isolated from a sponge-like sea creature and approved by the FDA to treat blood cancers that spread to the brain. Other plant-derived medications are already being used to tackle viral infections or to control pain. From nearly 400,000 different plant species, hundreds of medicines have already been approved and are on the market. Similarly, deciphering plant genetics have galvanized ideas for new biodegradable materials and biofuels.
Genetic sequences from complex organisms can “provide the raw materials for genome engineering and synthetic biology to produce valuable bioproducts at industrial scale,” wrote the team.
Medical and industrial uses aside, the effort also documents biodiversity. Creating a DNA digital library of all known eukaryotic life can pinpoint which species are most at risk—including species not yet fully characterized—providing data for earlier intervention.
“For the first time in history, it is possible to efficiently sequence the genomes of all known species and to use genomics to help discover the remaining 80 to 90 percent of species that are currently hidden from science,” wrote the team.
Soldiering OnThe project has three phases.
Phase one lays the groundwork. It establishes the species to be sequenced, builds digital infrastructure for data sharing, develops an analysis toolkit. The most important goal is to build a reference DNA sequence for species similar in genetic makeup—that is, those in a “family.”
Reference genomes are incredibly important for genetic studies. True to their name, scientists rely on them as a baseline when comparing genetic variants—for example, to track down genes related to inherited diseases in humans or sugar content in different variants of crops.
Phase two of the project will begin analyzing the sequencing data and form strategies to maintain biodiversity. The last phase integrates all previous work to potentially revise how different species fit into our evolutionary tree. Scientists will also integrate climate data into this phase and tease out the impacts of climate change on biodiversity.
The international project began in 2018 and included the US, UK, Denmark, and China, with most DNA specimens sequenced at facilities in China and the UK. Today, 28 countries spanning six continents have signed on. Most DNA material isolated from individual species is directly sequenced on site, reducing the cost of transportation while increasing fidelity.
Not all participants have easy access to DNA sequencing facilities. One institution, Wellcome Sanger, developed a portable DNA sequencing lab that could help scientists working in rural areas to capture the genetic blueprints of exotic plants and animals. The device sequenced the DNA of a type of sunflower with potential medicinal properties in Africa, among other specimens from exotic locations.
EBP follows in the footsteps of other global projects aiming to sequence the Earth’s microbes, such as the National Microbiome Initiative or the Earth Microbiome Project. Once also considered moonshots, these have secured funding from government agencies and private investments.
Despite the enthusiasm of its participants, EBP is still short billions of dollars to guide it to full completion. But the project’s price tag—originally estimated in the billions of dollars—may be far less.
Thanks to more efficient and cheaper genetic sequencing methods, the current cost of phase one is expected to be half the original estimate—around $265 million.
It’s still a hefty sum, but for participants, the resulting database and methods are worth it. “We now have a common forum to learn together about how to produce genomes with the highest possible quality,” Alexandre Aleixo at the Vale Institute of Technology, who participated in the project, told Science.
Given the influence bacterial genetics has already had on biomedicine and biofuels, it’s likely that deciphering eukaryote DNA can spur further inspiration. In the end, the project relies on a global collaboration to benefit humanity.
“The far-reaching potential benefits of creating an open digital repository of genomic information for life on Earth can be realized only by a coordinated international effort,” wrote the team.
Image Credit: M. Richter on Pixabay
ARTIFICIAL INTELLIGENCEWhy AI Could Eat Quantum Computing’s Lunch
Edd Gent | MIT Technology Review“The scale and complexity of quantum systems that can be simulated using AI is advancing rapidly, says Giuseppe Carleo, a professor of computational physics at the Swiss Federal Institute of Technology (EPFL). …Given the pace of recent advances, a growing number of researchers are now asking whether AI could solve a substantial chunk of the most interesting problems in chemistry and materials science before large-scale quantum computers become a reality.”
ROBOTICSMIT Debuts a Large Language Model-Inspired Method for Teaching Robots New Skills
Brian Heater | TechCrunch“The team introduced a new architecture called heterogeneous pretrained transformers (HPT), which pulls together information from different sensors and different environments. …’Our dream is to have a universal robot brain that you could download and use for your robot without any training at all,’ CMU associate professor David Held said of the research. ‘While we are just in the early stages, we are going to keep pushing hard and hope scaling leads to a breakthrough in robotic policies, like it did with large language models.'”
FUTUREWhy Futurist Amy Webb Sees a ‘Technology Supercycle’ Headed Our Way
Tim Brinkhof | Big Think“[Webb] predicts that we are on the cusp of a ‘technology supercycle,’ in which advances in three complementary and increasingly interconnected fields of research—AI, biotech, and smart sensors—will transform our economy and society to a similar extent as the wheel and the steam engine.”
BIOTECHA ‘Crazy’ Idea for Treating Autoimmune Diseases Might Actually Work
Sarah Zhang | The Atlantic“Lupus cannot be cured. No autoimmune disease can be cured. Two years ago, however, a study came out of Germany that rocked all of these assumptions. Five patients with uncontrolled lupus went into complete remission after undergoing a repurposed cancer treatment called CAR-T-cell therapy, which largely wiped out their rogue immune cells. The first treated patient has had no symptoms for almost four years now.”
GENE EDITINGHow a Breakthrough Gene-Editing Tool Will Help the World Cope With Climate Change
James Temple | MIT Technology Review“Jennifer Doudna, one of the inventors of the breakthrough gene-editing tool CRISPR, says the technology will help the world grapple with the growing risks of climate change by delivering crops and animals better suited to hotter, drier, wetter, or weirder conditions. ‘The potential is huge,’ says Doudna, who shared the 2020 Nobel Prize in chemistry for her role in the discovery. ‘There is a coming revolution right now with CRISPR.'”
TECHThe Death of Search
Matteo Wong | The Atlantic“A little, or even a lot, of inefficiency in search has long been the norm; AI will snuff it out. Our lives will be more convenient and streamlined, but perhaps a bit less wonderful and wonder-filled, a bit less illuminated. A process once geared toward exploration will shift to extraction. Less meandering, more hunting. No more unknown unknowns. If these companies really have their way, no more hyperlinks—and thus, no actual web.”
ENERGYOne Way That Could Improve Space-Based Power: Relays
Michelle Hampson | IEEE Spectrum“Intermediate transmitters could more effectively beam power to the ground. …In their study, the researchers designed and tested several low-cost, light-weight proof of concept transmit arrays to refocus the beam, finding the tactic could transfer nearly 2.5 times as much power as a system that would beam power straight to Earth.”
ARTIFICIAL INTELLIGENCEDebate May Help AI Models Converge on Truth
Stephen Ornes | Quanta“Letting AI systems argue with each other may help expose when a large language model has made mistakes. …The approach was first proposed six years ago, but two sets of findings released earlier this year—one in February from the AI startup Anthropic and the second in July from Google DeepMind—offer the first empirical evidence that debate between two LLMs helps a judge (human or machine) recognize the truth.”
SPACELife-Seeking, Ice-Melting Robots Could Punch Through Europa’s Icy Shell
Robin George Andrews | MIT Technology Review“Can robots actually get through that ice shell and survive the journey? A simple way to start is with a cryobot—a melt probe that can gradually thaw its way through the shell, pulled down by gravity. …Once it gets through the ice, the cryobot could unfurl a suite of scientific investigation tools, or perhaps deploy an independent submersible that could work in tandem with the cryobot—all while making sure none of that radioactive matter contaminates the ocean.”
Image Credit: Harry Borrett on Unsplash
The ability of plants to convert sunlight into food is an enviable superpower. Now, researchers have shown they can get animal cells to do the same thing.
Photosynthesis in plants and algae is performed by tiny organelles known as chloroplasts, which convert sunlight into oxygen and chemical energy. While the origins of these structures are hazy, scientists believe they may have been photosynthetic bacteria absorbed by primordial cells.
Our ancestors weren’t so lucky, but now researchers from the University of Tokyo have managed to rewrite evolutionary history. In a recent paper, the team reported they had successfully implanted chloroplasts into hamster cells where they generated energy for at least two days via the photosynthetic electron transport process.
“As far as we know, this is the first reported detection of photosynthetic electron transport in chloroplasts implanted in animal cells,” professor Sachihiro Matsunaga said in a press release.
“We thought that the chloroplasts would be digested by the animal cells within hours after being introduced. However, what we found was that they continued to function for up to two days, and that the electron transport of photosynthetic activity occurred.”
Some animals have already managed to gain the benefits of photosynthesis—notably giant clams, which host algae in a symbiotic relationship. And it’s not the first time people have tried adding photosynthetic abilities into different kinds of cells. Previous studies had managed to make a kind of chimera between photosynthetic cyanobacteria and yeast cells.
But transplanting chloroplasts into animal cells is a bigger challenge. One of the major hurdles the researchers faced is that most algal chloroplasts become inactive below 37 degrees Celsius (98.6 degree Fahrenheit), but animal cells need to be cultured at these lower temperatures.
This prompted them to pick chloroplasts from a type of algae called Cyanidioschyzon merolae, which lives in highly acidic and volcanic hot springs. While it prefers temperatures about 42 degrees Celsius (107.6 degrees Fahrenheit), it remains active at much lower temperatures.
After isolating the algae’s chloroplasts and injecting them into hamster cells, the researchers cultured them for several days. During that time, they checked for photosynthetic activity using light pulses and imaged the cells to determine the location and structure of the choloroplasts.
They discovered the organelles were still producing energy after two days. They even found the so-called “planimal” cells were growing faster than regular hamster cells, suggesting the chloroplasts were providing a carbon source that acted as fuel for the host cells.
They also found many of the chloroplasts had migrated to surround the cells’ nuclei, and organelles known as mitochondria that convert carbohydrates into energy the cell can use had also gathered around the chloroplasts. The team suggests there could be some kind of chemical exchange between these sub-cellular structures, though they’ll need future studies to confirm this.
After two days, however, the chloroplasts started degrading, and by the fourth day, photosynthesis seemed to have stopped. This is probably due to the animal cells digesting the unfamiliar organelles, but the researchers say genetic tweaks to the animal cells could potentially side-step digestion.
While the research might conjure sci-fi visions of humans with green skin surviving on sunlight alone, the team says the most likely applications are in tissue engineering. Lab-grown tissue typically consists of several layers of cells, and it can be hard to get oxygen deep into the tissue.
“By mixing in chloroplast-implanted cells, oxygen could be supplied to the cells through photosynthesis, by light irradiation, thereby improving the conditions inside the tissue to enable growth,” said Matsunaga.
Nonetheless, the research is a breakthrough that rewrites many of our assumptions about life’s possible forms. And while it might be a distant prospect, it opens the tantalizing possibility of one day giving animals the solar-powered capabilities of plants.
Image Credit: R. Aoki, Y. Inui, Y. Okabe et al. 2024/ Proceedings of the Japan Academy, Series B
The first spark of cellular life on Earth likely needed gift packaging.
Let me explain. With the holidays around the corner, we’re all beginning to order presents. Each is carefully packaged inside a box or bubble-wrapped envelope and addressed for shipping. Without packaging, items would tumble together in a chaotic mess and miss their destination.
Life’s early chemicals were, in a way, like these “presents.” They floated around in a primordial soup, eventually forming the longer molecules that make up life as we know it. But without a “wrapper” encapsulating them in individual packages, different molecules bumped into each other but eventually drifted away, missing the necessary connections to spark life.
In other words, cellular “wrappers,” or cell membranes, are key to packaging the molecular machinery of life together. Made of fatty molecules, these wrappers are the foundation of our cells and the basis of multicellular life. They keep bacteria and other pathogens at bay while triggering the biological mechanisms that power normal cellular functions.
Scientists have long debated how the first cell membranes formed. Their building blocks, long-chain lipids, were hard to find on early Earth. Shorter fatty molecules, on the other hand, were abundant. Now, a new study in Nature Chemistry offers a bridge between these short fatty molecules and the first primordial cells.
Led by Neal Devaraj at the University of California, San Diego, the team coaxed short fatty molecules into bubbles that can encapsulate biological molecules. The team then added modern RNA molecules to drive chemical reactions inside the bubbles—and watched the reactions work, similar to those in a functional cell.
The engineered cell membranes also resisted high concentrations of substances abundant in early Earth puddles that could damage their integrity, shielding molecular carriers of genetic information and allowing them to work normally.
The resulting protocells are the latest to probe the origins of life. To be clear, they only mimic parts of normal living cells. They don’t have the molecular machinery to replicate, and their wrappers are rudimentary compared to ours.
But the “fascinating” result “opens up a new avenue” for understanding how the first cells appeared, Sheref Mansy at the University of Trento, who was not involved in the study, told Science.
At the BeginningThe origins of life’s molecules are highly debated. But most scientists agree that life stemmed from three main ones: DNA, RNA, and amino acids (the building blocks of proteins).
Today, in most organisms, DNA stores the genetic blueprint, and RNA carries this genetic information to the cell’s protein-making factories. But many viruses store genes only in RNA, and studies of early life suggest RNA may have been the first carrier of inheritance. RNA can also spur chemical reactions—including ones that glue amino acids into different types of proteins.
But regardless of which molecule came first, “all life on Earth requires lipid membranes,” the authors of the new paper write.
Made of a double layer of fatty molecules, the modern cell membrane is a work of art. It’s the first defense against bacterial and viral invaders. It’s also dotted with protein “tunnels” that tweak the functions of cells—for example, helping brain cells encode memories or heart cells beat in sync. These living cellular walls also act as scaffolds for biochemical reactions that often dictate the fate of cells—if they live, die, or turn into “zombie cells” that contribute to aging.
Since they’re so important for biology, scientists have long wondered how the first cell membranes came about. What made up “the very first, primordial cell membrane-like structure on Earth before the emergence of life?” asked the authors.
Our cell membranes are built on long chains of lipids, but these have complex chemical structures and require multiple steps to synthesize—likely beyond what was possible on early Earth. In contrast, the first protocell membranes were likely formed from molecules already present, including short fatty acids that self-organized.
Back to the FuturePreviously, the team found an amino acid that “staples” fatty acids together. Called cysteine, the molecule was likely prevalent in our planet’s primordial soup. In a computer simulation, adding cysteine to short fatty acids caused them to form synthetic membranes.
The new study built on those results in the lab.
The team added cysteine to two types of short lipids and watched as the amino acid gathered the lipids into bubbles within 30 minutes. The lipids were similar in length to those likely present on early Earth, and the molecular concentrations also mimicked those during the period.
The team next took a closer look with an electron microscope. The generated membranes were about as thick as those in normal cells and highly stable. Finally, the team simulated a hypothetical early-Earth scenario where RNA serves as the first genetic material.
“The RNA world hypothesis is accepted as one of the most plausible scenarios of the origin of life,” wrote the authors. This is partly because RNA can also act as enzyme. These enzymes, dubbed ribozymes, can spark different chemical reactions, like, for example, those that might stitch amino acids and lipids into bubbles. However, they need a duo of minerals—calcium and magnesium—to work. While these minerals were likely highly abundant on early Earth, in some cases, they can damage artificial cell membranes.
But in several tests, the lab-grown protocells easily withstood the mineral onslaught. Meanwhile, the protocells showed they could generate chemical reactions using RNA, suggesting that short fatty molecules can build cell membranes in the primordial soup.
To Claudia Bonfio at the University of Cambridge, the study was “really, really cool and very well done.” But the mystery of life remains. Most fatty acids generated in the protocell aren’t found in modern cell membranes. A next step would be to show that the protocells can act more like normal ones—growing and dividing with a healthy metabolism.
But for now, the team is focused on deciphering the beginnings of cellular life. The work shows that reactions between simple chemicals in water can “assemble into giant” blobs, expanding the ways that protocell membranes can form, they wrote.
Image Credit: Max Kleinen on Unsplash
The European Space Agency has given the go-ahead for initial work on a mission to visit an asteroid called Apophis. If approved at a key meeting next year, the robotic spacecraft, known as the Rapid Apophis Mission for Space Safety (Ramses), will rendezvous with the asteroid in February 2029.
Apophis is 340 meters wide, about the same as the height of the Empire State Building. If it were to hit Earth, it would cause wholesale destruction hundreds of miles from its impact site. The energy released would equal that from tens or hundreds of nuclear weapons, depending on the yield of the device.
Luckily, Apophis won’t hit Earth in 2029. Instead, it will pass by Earth safely at a distance of 19,794 miles (31,860 kilometers), about one-twelfth the distance from the Earth to the Moon. Nevertheless, this is a very close pass by such a big object, and Apophis will be visible with the naked eye.
NASA and the European Space Agency have seized this rare opportunity to send separate robotic spacecraft to rendezvous with Apophis and learn more about it. Their missions could help inform efforts to deflect an asteroid that threatens Earth, should we need to in the future.
The Threat From AsteroidsSome 66 million years ago, an asteroid the size of a small city hit Earth. The impact of this asteroid brought about a global extinction event that wiped out the dinosaurs.
Earth is in constant danger of being hit by asteroids, leftover debris from the formation of the solar system 4.5 billion years ago. Located in the asteroid belt between Mars and Jupiter, asteroids come in many shapes and sizes. Most are small, only 10 meters across, but the largest are hundreds of kilometers across, larger than the asteroid that killed the dinosaurs.
Artist’s impression of Apophis. Image Credit: NASAThe asteroid belt contains one to two million asteroids larger than a kilometer across and millions of smaller bodies. These space rocks feel each other’s gravitational pull, as well as the gravitational tug of Jupiter on one side and the inner planets on the other.
Because of this gravitational tug-of-war, every once in a while an asteroid is thrown out of its orbit and hurtles towards the inner solar system. There are 35,000 such “near-Earth objects” (NEOs). Of these, 2,300 “potentially hazardous objects” (PHOs) have orbits that intersect Earth’s and are large enough that they pose a real threat to our survival.
Do Not Go Gentle Into That Good NightDuring the 20th century, astronomers set up several surveys, such as Atlas, in order to detect and study hazardous asteroids. But detection is not enough; we have to find a way to defend Earth against an incoming asteroid.
Blowing up an asteroid, as depicted in the movie Armageddon, is no use. The asteroid would be broken into smaller fragments, which would keep moving in much the same direction. Instead of being hit by one large asteroid, Earth would be hit by a swarm of smaller objects.
The preferred solution is to deflect the incoming asteroid away from Earth so that it passes by harmlessly. To do so, we would need to apply an external force to the asteroid to nudge it away. A popular idea is to fire a projectile at the asteroid. NASA did this in 2022, when a spacecraft called DART collided with an asteroid. Before we do this out of necessity, we have to understand how different types of asteroids would react to such an impact.
Apophis, Ramses, and Osiris-ApexApophis was discovered in 2004. The asteroid passed by Earth on December 21, 2004 at a distance of 14 million kilometers. It returned in 2021 and will swing by Earth again in 2029, 2036, and 2068.
Until recently, there was a small chance that Apophis could collide with Earth in 2068. However, during Apophis’ approach in 2021, astronomers used radar observations to refine their knowledge of the asteroid’s orbit. These showed that Apophis would not hit our planet for the next 100 years.
The Ramses mission will rendezvous with Apophis in February 2029, two months before its closest approach to Earth on Friday, April 13. It will then accompany the asteroid as it approaches Earth. The goal is to learn how Apophis’s orbit, rotation, and shape will change as it passes so close to Earth’s gravitational field.
In 2016, NASA launched the “Origins, Spectral Interpretation, Resource Identification, and Security–Regolith Explorer” (Osiris-Rex) mission to study the near-Earth asteroid Bennu. It intercepted Bennu in 2020 to collect samples of rock and soil from its surface and dispatched the rocks in a capsule, which arrived on Earth in 2023.
The spacecraft is still out there, so NASA renamed it the “Origins, Spectral Interpretation, Resource Identification and Security–Apophis Explorer” (Osiris-Apex) and assigned it to study Apophis. Osiris-Apex will reach the asteroid just after its 2029 close encounter. It will then fly low over Apophis’s surface and fire its engines, disturbing the rocks and dust that cover the asteroid to reveal the layer underneath.
A close flyby of an asteroid as large as Apophis happens only once every 5,000 to 10,000 years. Apophis’s arrival in 2029 presents a rare opportunity to study such an asteroid up close, and seeing how it is affected by Earth’s gravitational pull. The information gleaned will shape the way we choose to protect Earth in the future from a real killer asteroid.
Ancient Egyptian MythologyWhen Ramses and Osiris-Apex meet up with Apophis in 2029 they will inadvertently reenact a core component of ancient Egyptian cosmology. To the ancient Egyptians, the sun was personified by several powerful gods, chief among them Re. The sun’s setting in the evening was interpreted as Re dying and entering the netherworld.
During his nighttime journey through the netherworld, Re was menaced by the great snake Apophis, who embodied the powers of darkness and dissolution. Only after Apophis had been defeated could Re be revitalized by Osiris, the king of the netherworld. Re could then once again be reborn in the east, rising in the sky once more.
Tomb murals, coffins, and funerary papyri depict Apophis as a large, coiled snake threatening Re as he sails in his solar barque (sailing ship). But Apophis is always defeated, his body pierced by a spear or riven by knives.
Though the asteroid Apophis poses no danger in the near future, Ramses (named after the pharaohs of the same name, which meant “born of Re”) and Osiris-Apex will study it so that one day we will know how to defeat it—or any of its distant brethren.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
One of the world’s most advanced humanoid robots has been all play and no work. Boston Dynamics’ Atlas is famous for backflips, parkour, and dance mobs. These require extremely impressive robotic control, but they’re also mostly fun research demos.
Now, six months after the legendary robotics lab unveiled an all-new electric Atlas, they’re showing off more of what it can do. A recent video shows Atlas picking auto parts from one set of shelves and moving them over to another, a job currently handled by factory workers.
Apart from being electric, the new Atlas has a unique way of moving. Its head, upper body, pelvis, and legs swivel independently. So, its head might rotate to face the opposite direction of its legs and torso, Exorcist-style, before the rest of its body twists around to catch up.
The new demo highlights another core change for Atlas. Whereas, in the past, Boston Dynamics meticulously programmed the robot’s most impressive maneuvers, the latest video, by contrast, shows a fully autonomous Atlas at work.
“There are no prescribed or teleoperated movements; all motions are generated autonomously online,” according to a description accompanying the video.
Why release this video now? One, humanoid robots are having something of a moment. And two, ditto for artificial intelligence in robotics. Boston Dynamics led the pack for years, but it didn’t rush Atlas into production for commercial use. Neither has it added significant amounts of AI to the equation. Now, it appears to be interested in both.
Last month, the lab, which is owned by Hyundai, announced a partnership with Toyota Research Institute (TRI) to add artificial intelligence, TRI’s specialty, to Atlas. Alongside pure research, the partnership hopes to make Atlas into a general-purpose humanoid.
It’s an intriguing development. In terms of pure robotics, Atlas is world-class. TRI, meanwhile, is working to develop large behavior models, which are like large language models for robotic movement and manipulation. The idea is that with enough real-world data, AI models like this might develop into a kind robotic brain that doesn’t need to be explicitly programmed for every scenario it might encounter.
Google DeepMind has also been pursuing a similar approach with a vision-language-action model called RT-X and united 33 research labs in an effort to assemble a vast new AI training dataset for robotics. And just last week, a TRI-funded MIT project showed off a new transformer algorithm like the one behind ChatGPT, only designed for robotics.
“Our dream is to have a universal robot brain that you could download and use for your robot without any training at all,” CMU associate professor David Held told TechCrunch. “While we are just in the early stages, we are going to keep pushing hard and hope scaling leads to a breakthrough in robotic policies, like it did with large language models.”
Boston Dynamics isn’t alone in its efforts. If anything, it’s late to the party. A host of companies, many born in the last few years, share the goal of general-purpose humanoids. These include Agility Robotics, Tesla, Figure, and 1X, among others.
In an interview with IEEE Spectrum, Boston Dynamics’ Scott Kuindersma said this may be “one of the most exciting points” in the field’s history. At the same time, he acknowledged there’s a lot of hype out there—and a lot of work still to do. Challenges include collecting enough of the right kind of data and dialing in how best to train robotics algorithms.
That doesn’t mean there won’t be more Boston Dynamics videos out soon. “I want people to be excited about watching for our results, and I want people to trust our results when they see them,” TRI’s Russ Tedrake said in the same interview.
AI-Atlas is just getting started.
Image Credit: Boston Dynamics
ARTIFICIAL INTELLIGENCEGoogle CEO Says Over 25% of New Google Code Is Generated by AI
Benj Edwards | Ars Technica“We’ve always used tools to build new tools, and developers are using AI to continue that tradition. On Tuesday, Google’s CEO revealed that AI systems now generate more than a quarter of new code for its products, with human programmers overseeing the computer-generated contributions. The statement, made during Google’s Q3 2024 earnings call, shows how AI tools are already having a sizable impact on software development.”
AUTOMATIONWaymo Raises $5.6 Billion From Outside Investors
Eli Tan | The New York Times“Amid its push to grow its fleet of autonomous robot taxis and expand into new cities, Waymo has raised $5.6 billion from outside investors, its largest funding round to date. …The fresh money comes behind Waymo’s first taste of commercial success. Its robot taxis are now completing over 100,000 rides each week in San Francisco, Phoenix and Los Angeles, double its number in May, and will be operating in Austin, Texas, and Atlanta by 2025 through a partnership with Uber.”
ROBOTICSThis Is a Glimpse of the Future of AI Robots
Will Knight | Wired“Physical Intelligence, also known as PI or π, was founded earlier this year by several prominent robotics researchers to pursue the new robotics approach inspired by breakthroughs in AI’s language abilities. ‘The amount of data we’re training on is larger than any robotics model ever made, by a very significant margin, to our knowledge,’ says Sergey Levine, a cofounder of Physical Intelligence and an associate professor at UC Berkeley.”
ENERGYNuclear Fusion’s New Idea: An Off-the-Shelf Stellarator
Tom Clynes | IEEE Spectrum“The PPPL team invented this nuclear-fusion reactor, completed last year, using mainly off-the-shelf components. Its core is a glass vacuum chamber surrounded by a 3D-printed nylon shell that anchors 9,920 meticulously placed permanent rare-earth magnets. Sixteen copper-coil electromagnets resembling giant slices of pineapple wrap around the shell crosswise.”
TECHWall Street Giants to Make $50 Billion Bet on AI and Power Projects
Katherine Blunt | The Wall Street Journal“The investment is a bet on AI’s huge energy needs and the mounting stress it is putting on the US power grid. …The companies said they are now working together with large tech companies to accelerate their access to electricity, which has become constrained in parts of the US as data-center developers compete for power sources and access to the grid. ‘The capital needs are huge, and one of the big bottlenecks—maybe the bottleneck—is electricity availability,’ ECP founder and senior partner Doug Kimmelman said.
ENVIRONMENTThe AI Boom Rests on Billions of Tons of Concrete
Ted C. Fishman | IEEE Spectrum“To the casual observer, the data industry can seem incorporeal, its products conjured out of weightless bits. But as I stand beside the busy construction site for DataBank’s ATL4, what impresses me most is the gargantuan amount of material—mostly concrete—that gives shape to the goliath that will house, secure, power, and cool the hardware of AI. Big data is big concrete. And that poses a big problem.”
AUTOMATIONWaymo Explores Using Google’s Gemini to Train Its Robotaxis
Andrew J. Hawkins | The Verge“Waymo has long touted its ties to Google’s DeepMind and its decades of AI research as a strategic advantage over its rivals in the autonomous driving space. Now, the Alphabet-owned company is taking it a step further by developing a new training model for its robotaxis built on Google’s multimodal large language model (MLLM) Gemini.”
SPACESpaceX Has Caught a Massive Rocket. So What’s Next?
Eric Berger | Ars Technica“Here’s our best attempt to piece together the milestones and major goals of the Starship program over the next several years before it unlocks the capability to land humans on the Moon for NASA’s Artemis Program and begins flying demonstration missions to Mars. For fun, we’ve also included some estimated dates for each of these milestones. These represent our best guesses, and they’re almost certainly wrong.”
SCIENCEMeet the First Star System to ‘Solve’ the 3-Body Problem
Ethan Siegel | Big Think“It’s easy to have planets that orbit around a single star, and in a double star system, you can either orbit close to one star or far from both members. These configurations are stable, but adding a third star into the mix was thought to render the formation of planets unstable, as mutual gravitational interactions would eventually force their ejection. That wisdom got thrown out the window with the discovery of GW Orionis, which boasts multiple massive dust rings and possibly even more planets, all orbiting three stars at once.”
Image Credit: David Clode on Unsplash
In 2014, the British philosopher Nick Bostrom published a book about the future of artificial intelligence with the ominous title Superintelligence: Paths, Dangers, Strategies. It proved highly influential in promoting the idea that advanced AI systems—“superintelligences” more capable than humans—might one day take over the world and destroy humanity.
A decade later, OpenAI boss Sam Altman says superintelligence may only be “a few thousand days” away. A year ago, Altman’s OpenAI cofounder Ilya Sutskever set up a team within the company to focus on “safe superintelligence,” but he and his team have now raised a billion dollars to create a startup of their own to pursue this goal.
What exactly are they talking about? Broadly speaking, superintelligence is anything more intelligent than humans. But unpacking what that might mean in practice can get a bit tricky.
Different Kinds of AIIn my view, the most useful way to think about different levels and kinds of intelligence in AI was developed by US computer scientist Meredith Ringel Morris and her colleagues at Google.
Their framework lists six levels of AI performance: no AI, emerging, competent, expert, virtuoso, and superhuman. It also makes an important distinction between narrow systems, which can carry out a small range of tasks, and more general systems.
A narrow, no-AI system is something like a calculator. It carries out various mathematical tasks according to a set of explicitly programmed rules.
There are already plenty of very successful narrow AI systems. Morris gives the Deep Blue chess program that famously defeated world champion Garry Kasparov way back in 1997 as an example of a virtuoso-level narrow AI system.
Table: The Conversation * Source: Adapted from Morris et al. * Created with DatawrapperSome narrow systems even have superhuman capabilities. One example is AlphaFold, which uses machine learning to predict the structure of protein molecules, and whose creators won the Nobel Prize in Chemistry this year.What about general systems? This is software that can tackle a much wider range of tasks, including things like learning new skills.
A general no-AI system might be something like Amazon’s Mechanical Turk: It can do a wide range of things, but it does them by asking real people.
Overall, general AI systems are far less advanced than their narrow cousins. According to Morris, the state-of-the-art language models behind chatbots such as ChatGPT are general AI—but they are so far at the “emerging” level (meaning they are “equal to or somewhat better than an unskilled human”), and yet to reach “competent” (as good as 50 percent of skilled adults).
So by this reckoning, we are still some distance from general superintelligence.
How Intelligent Is AI Right Now?As Morris points out, precisely determining where any given system sits would depend on having reliable tests or benchmarks.
Depending on our benchmarks, an image-generating system such as DALL-E might be at virtuoso level (because it can produce images 99 percent of humans could not draw or paint), or it might be emerging (because it produces errors no human would, such as mutant hands and impossible objects).
There is significant debate even about the capabilities of current systems. One notable 2023 paper argued GPT-4 showed “sparks of artificial general intelligence.”
OpenAI says its latest language model, o1, can “perform complex reasoning” and “rivals the performance of human experts” on many benchmarks.
However, a recent paper from Apple researchers found o1 and many other language models have significant trouble solving genuine mathematical reasoning problems. Their experiments show the outputs of these models seem to resemble sophisticated pattern-matching rather than true advanced reasoning. This indicates superintelligence is not as imminent as many have suggested.
Will AI Keep Getting Smarter?Some people think the rapid pace of AI progress over the past few years will continue or even accelerate. Tech companies are investing hundreds of billions of dollars in AI hardware and capabilities, so this doesn’t seem impossible.
If this happens, we may indeed see general superintelligence within the “few thousand days” proposed by Sam Altman (that’s a decade or so in less sci-fi terms). Sutskever and his team mentioned a similar timeframe in their superalignment article.
Many recent successes in AI have come from the application of a technique called “deep learning,” which, in simplistic terms, finds associative patterns in gigantic collections of data. Indeed, this year’s Nobel Prize in Physics has been awarded to John Hopfield and also the “Godfather of AI” Geoffrey Hinton, for their invention of the Hopfield network and Boltzmann machine, which are the foundation of many powerful deep learning models used today.
General systems such as ChatGPT have relied on data generated by humans, much of it in the form of text from books and websites. Improvements in their capabilities have largely come from increasing the scale of the systems and the amount of data on which they are trained.
However, there may not be enough human-generated data to take this process much further (although efforts to use data more efficiently, generate synthetic data, and improve transfer of skills between different domains may bring improvements). Even if there were enough data, some researchers say language models such as ChatGPT are fundamentally incapable of reaching what Morris would call general competence.
One recent paper has suggested an essential feature of superintelligence would be open-endedness, at least from a human perspective. It would need to be able to continuously generate outputs that a human observer would regard as novel and be able to learn from.
Existing foundation models are not trained in an open-ended way, and existing open-ended systems are quite narrow. This paper also highlights how either novelty or learnability alone is not enough. A new type of open-ended foundation model is needed to achieve superintelligence.
What Are the Risks?So what does all this mean for the risks of AI? In the short term, at least, we don’t need to worry about superintelligent AI taking over the world.
But that’s not to say AI doesn’t present risks. Again, Morris and co have thought this through: As AI systems gain great capability, they may also gain greater autonomy. Different levels of capability and autonomy present different risks.
For example, when AI systems have little autonomy and people use them as a kind of consultant—when we ask ChatGPT to summarize documents, say, or let the YouTube algorithm shape our viewing habits—we might face a risk of over-trusting or over-relying on them.
In the meantime, Morris points out other risks to watch out for as AI systems become more capable, ranging from people forming parasocial relationships with AI systems to mass job displacement and society-wide ennui.
What’s Next?Let’s suppose we do one day have superintelligent, fully autonomous AI agents. Will we then face the risk they could concentrate power or act against human interests?
Not necessarily. Autonomy and control can go hand in hand. A system can be highly automated, yet provide a high level of human control.
Like many in the AI research community, I believe safe superintelligence is feasible. However, building it will be a complex and multidisciplinary task, and researchers will have to tread unbeaten paths to get there.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Electric air taxis have seen rapid technological advances in recent years, but the industry has had a regulatory question mark hanging over its head. Now, the US Federal Aviation Authority has published rules governing the operation of this new class of aircraft.
Startups developing electric vertical take-off and landing (eVTOL) aircraft have attracted billions of dollars of investment over the past decade. But an outstanding challenge for these vehicles is they’re hard to classify, often representing a strange hybrid between a drone, light aircraft, and helicopter.
For this reason they’ve fallen into a regulatory gray area in most countries. The murkiness has led to considerable uncertainty about where and how they’ll be permitted to operate in the future, which could have serious implications for the business model of many of these firms.
But now, the FAA has provided some much-needed clarity by publishing the rules governing what the agency calls “powered-lift” aircraft. This is the first time regulators have recognized a new category of aircraft since the 1940s when helicopters first entered the market.
“This final rule provides the necessary framework to allow powered-lift aircraft to safely operate in our airspace,” FAA administrator Mike Whitaker said in a statement. “Powered-lift aircraft are the first new category of aircraft in nearly 80 years and this historic rule will pave the way for accommodating wide-scale advanced air mobility operations in the future.”
The principal challenge when it comes to regulating air taxis is the novel way they operate. Most leading designs use propellers that rotate up and down, which allows them to take off vertically like a helicopter before operating more like a conventional airplane during cruise.
The agency dealt with this by varying the operational requirements, such as minimum safe altitude, required visibility, and range, depending on the phase of flight. This means that during take-off the vehicles need to adhere to the less stringent requirements placed on helicopters, but when cruising they must conform to the same rules as airplanes. The rules are also performance-based, so exact requirements will depend on the capabilities of the specific vehicle in question.
The new regulations also provide a framework for certifying the initial batch of instructors and training future pilots. Because eVTOLs are a new class of aircraft, there are currently no pilots certified to fly them and therefore no one to train other pilots.
To get round this chicken-and-egg situation, the FAA says they’ll allow certain pilots employed by eVTOL companies to develop the required experience and training during the test flights required for vehicle certification. These pilots would become the first group of instructors who could then train other instructors at pilot schools and training centers.
The regulations also relax an existing requirement for training aircraft to feature two sets of flight controls. Instead, the agency is allowing pilots to learn in aircraft where the trainer can easily access the controls to intervene, if necessary, or letting pilots train in a simulator to gain enough experience to fly the aircraft solo.
When the agency introduced draft rules last year, the industry criticized them as too strict, according to The Verge. But the agency says it has taken the criticism onboard and thinks the new rules strike a good balance between safety and easing the burden on companies.
Industry leader Joby Aviation welcomed the new rules and, in particular, the provision for training pilots in simulators. “The regulation published today will ensure the US continues to play a global leadership role in the development and adoption of clean flight,” JoeBen Bevirt, founder and CEO of Joby, said in a statement. “Delivering ahead of schedule is a testament to the dedication, coordination and hard work of the rulemaking team.”
In its announcement, the FAA highlighted the technology’s potential for everything from air taxi services to short-haul cargo transport and even air ambulances. With these new rules in place, operators can now start proving out some of those business cases.
Image Credit: Joby
With a brain the size of a sesame seed, the lowly fruit fly is often considered a kitchen pest. But to neuroscientists, the flies are a treasure trove of information detailing how the brain’s intricate connections guide thoughts, decisions, and memories—not just for the critters, but also for us.
Mapping these connections is the first step. With over 140,000 neurons and 54 million synapses—the connections between nerve cells—packed into such a tiny space, the fruit fly’s brain, however rudimentary compared to ours, is highly complex.
This week, in a tour de force, hundreds of scientists from the FlyWire consortium published the first complete map of an adult female fruit fly’s brain. A project roughly a decade in the making, the wiring diagram will be a rich scientific resource for years to come. The same techniques used to make the map—which heavily relied on artificial intelligence—could be used to chart more complex brains, such as zebrafish, mice, and perhaps even humans.
“Flies are important model systems…since their brains solve the same problems as we do,” said Mala Murthy at Princeton University in a press conference. Murthy co-led the project with Sebastian Seung, who has long championed mapping as a way to better understand the inner workings of our brains and potentially extract algorithms to power more flexible AI.
In one of nine articles on the project published by Nature, Clay Reid at the Allen Institute for Brain Science, who was not involved in the project, called the release a “huge deal.”
“It’s something that the world has been anxiously waiting for, for a long time,” he said.
The study’s data and images are freely available for anyone to explore. To Murthy, the project exemplifies the power of open science. The consortium welcomed help from both neuroscientists and citizen scientists, who don’t have formal training but are passionate about the brain.
This “openness drove the science forward,” resulting in the “first time we’ve had a complete map of any complex brain,” said Murthy.
A Brain AtlasWhy do we think, feel, remember, and forget? How do we make decisions, rethink biases, and empathize with others? Even simpler, what neural signals make my fingers type these words?
It’s all about wiring. Neurons connect with each other at specific points called synapses. These connections form the basis of circuits that control behaviors. Like tracing electrical wiring in a house, mapping the brain’s cables can help decipher which neural circuit controls what behaviors. Together, the entire brain wiring diagram is called the connectome.
Previously, scientists had only fully mapped the connectome of a tiny worm with just over 300 neurons. Even so, the feat launched a revolution in neuroscience by highlighting the role of neural circuits, rather than individual cells, in steering behavior.
The fruit fly brain is bigger and far more complex. It’s densely packed with hundreds of thousands of neurons, each intricately connected to others. A single faulty reconstruction could derail our understanding of the brain’s original instructions: Rather than sending a signal down one neural highway, it could be interpreted as a taking another road that leads to nowhere.
The project began over a decade ago, when Davi Bock and colleagues at the Janelia Research Campus imaged the entire fly brain at nanoscale resolution. They “fossilized” the brain of a female fly using a chemical soup, froze it to preserve its delicate connections, and sliced it into wafers.
Using a high-resolution microscope, the scientists took images of every slice. Overall, the project produced roughly 21 million images from over 7,000 brain slices.
This wealth of data was a triumph, but also a problem. Usually, each image had to be manually examined for potential connections—an obvious headache when analyzing millions of images.
Here’s where AI comes in. Seung has long championed using AI to untangle neural wiring from individual images and 3D recreations. With AI becoming increasingly sophisticated, it’s easier for different models to learn how to identify a synapse or the branches of a neuron.
But initial AI systems were imperfect. Overlapping neural wires from two circuits could be interpreted as one: Imagine a satellite view of a tricky highway interchange that confuses your phone’s GPS system. A giant tangle of neural connections from multiple sources could be labeled as a single source, rather than a hub directing the flow of information.
Scientists in the consortium spent years manually proofreading AI-generated results. But they had help. Seung and colleagues elicited crowd input. His earlier project, Eyewire, gamified the brain-mapping process by asking citizen scientists to detect neural connections critical for vision.
FlyWire built a similar online platform in 2022, allowing hundreds of people interested in the brain, but with no formal training, to proofread AI reconstructions and classify neurons based on their shape.
The project would have taken a single person 33 years. By sharing data and recruiting citizen scientists, the team constructed the entire connectome in a fraction of the time. According to study author Gregory Jefferis at the University of Cambridge, the volunteers and scientists made more than three million edits to the AI’s initial results. They also annotated the maps—for example, labeling different cells—providing context for the viewer.
Throughout the process, the consortium released versions of its data so researchers could tap into the expanding dataset. Even without the entire map, scientists have already begun exploring ideas about how the fly’s brain works.
Brain CartographerThe final map captures over 54 million synapses between roughly 140,000 neurons. It also includes over 8,000 different types of neuron—far more than anyone expected. Incredibly, nearly half were newly discovered for the species.
To Seung, each new cell type poses “a question” about how it influences brain functions.
The fly’s brain was also interconnected to a surprising degree. Neurons that allowed the fly to see also received sound and touch cues, suggesting these senses are wired together.
The connectome data is already spurring new studies and ideas. One team made a digitized fly brain from all the mapped neurons and connections. They then activated artificial neurons that can detect honey or bitter flavors. The virtual brain responded by sticking out the fly’s “tongue” when it detected sweet flavors.
“For decades, we haven’t known what the taste neurons in the brain are,” study author Anita Devineni at Emory University told Science. “And then, all of a sudden in a small amount of time … you can figure it out.”
Other studies using the map found neural circuits for walking, grooming, and feeding—all of which are essential to the fly’s (and our) everyday routine.
The connectome does have some limitations though. It’s based on a single female fruit fly. Brains are highly individualized in their connections, especially across sexes and ages. The decade-long effort is just a snapshot of one brain at one moment in time.
However, the map could still help researchers discover fundamental ways the brain works—like, for example, how wiring between certain brain regions allows them “talk” more efficiently.
The team is already looking to expand the work to a mouse brain with roughly 500 times more neurons than the fly. Similar efforts have already charted synapses in the mouse brain, but the new study’s technology could yield comprehensive maps of neural connections across the entire brain.
“This achievement is not just remarkable, it’s outstanding,” Moritz Helmstaedter at the Max Planck Institute for Brain Research, who was not involved in the project, told Science. “In the next decade, we’ll see tremendous progress, and possibly the first full whole mammalian brain connectome.”
Image Credit: Amy Sterling, Murthy and Seung labs, Princeton University, (Baker et al., Current Biology, 2022)
Humans are increasingly engaging with wearable technology as it becomes more adaptable and interactive. One of the most intimate ways gaining acceptance is through augmented reality glasses.
Last week, Meta debuted a prototype of the most recent version of their AR glasses—Orion. They look like reading glasses and use holographic projection to allow users to see graphics projected through transparent lenses into their field of view.
Meta chief Mark Zuckerberg called Orion “the most advanced glasses the world has ever seen.” He said they offer a “glimpse of the future” in which smart glasses will replace smartphones as the main mode of communication.
But is this true or just corporate hype? And will AR glasses actually benefit us in new ways?
Old Technology, Made NewThe technology used to develop Orion glasses is not new.
In the 1960s, computer scientist Ivan Sutherland introduced the first augmented reality head-mounted display. Two decades later, Canadian engineer and inventor Stephen Mann developed the first glasses-like prototype.
Throughout the 1990s, researchers and technology companies developed the capability of this technology through head-worn displays and wearable computing devices. Like many technological developments, these were often initially focused on military and industry applications.
In 2013, after smartphone technology emerged, Google entered the AR glasses market. But consumers were disinterested, citing concerns about privacy, high cost, limited functionality, and a lack of a clear purpose.
This did not discourage other companies—such as Microsoft, Apple, and Meta—from developing similar technologies.
Looking insideMeta cites a range of reasons for why Orion are the world’s most advanced glasses, such as their miniaturized technology with large fields of view and holographic displays. It said these displays provide “compelling AR experiences, creating new human-computer interaction paradigms […] one of the most difficult challenges our industry has ever faced.”
Orion also has an inbuilt smart assistant (Meta AI) to help with tasks through voice commands, eye and hand tracking, and a wristband for swiping, clicking, and scrolling.
With these features, it is not difficult to agree that AR glasses are becoming more user-friendly for mass consumption. But gaining widespread consumer acceptance will be challenging.
A Set of ChallengesMeta will have to address four types of challenges:
- How easy it is to wear, use, and integrate AR glasses with other glasses
- Physiological aspects such as the heat the glasses generate, comfort, and potential vertigo
- Operational factors such as battery life, data security, and display quality
- Psychological factors such as social acceptance, trust in privacy, and accessibility
These factors are not unlike what we saw in the 2000s when smartphones gained acceptance. Just like then, there are early adopters who will see more benefits than risks in adopting AR glasses, creating a niche market that will gradually expand.
Similar to what Apple did with the iPhone, Meta will have to build a digital platform and ecosystem around Orion.
This will allow for broader applications in education (for example, virtual classrooms), remote work, and enhanced collaboration tools. Already, Orion’s holographic display allows users to overlay digital content on the real world, and because it is hands-free, communication will be more natural.
Creative DestructionSmart glasses are already being used in many industrial settings, such as logistics and healthcare. Meta plans to launch Orion for the general public in 2027.
By that time, AI will have likely advanced to the point where virtual assistants will be able to see what we see and the physical, virtual, and artificial will co-exist. At this point, it is easy to see that the need for bulky smartphones may diminish and that through creative destruction, one industry may replace another.
This is supported by research indicating the virtual and augmented reality headset industry will be worth $370 billion by 2034.
The remaining question is whether this will actually benefit us.
There is already much debate about the effect of smartphone technology on productivity and wellbeing. Some argue that it has benefited us, mainly through increased connectivity, access to information, and productivity applications.
But others say it has just created more work, distractions, and mental fatigue.
If Meta has its way, AR glasses will solve this by enhancing productivity. Consulting firm Deloitte agrees, saying the technology will provide hands-free access to data, faster communication and collaboration through data-sharing.
It also claims smart glasses will reduce human errors, enable data visualization, and monitor the wearer’s health and wellbeing. This will ensure a quality experience, social acceptance, and seamless integration with physical processes.
But whether or not that all comes true will depend on how well companies such as Meta address the many challenges associated with AR glasses.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Meta
In a tiny laboratory pond, a robotic stingray flaps its fins and swims around. Roughly the width of a dime, the bot dashes distances multiple times its body size. It easily navigates around corners and swims far longer than previous flapping microbots of a similar design.
Its secret? The robot is a biohybrid blend of living, human-derived neurons and muscle cells controlled by a programmable electronic “brain.” The cells cover a synthetic “skeleton” with fins and form dense connections like those that drive movement in our bodies.
Also onboard is a wireless electronic circuit with magnetic coils. The circuit controls the robot’s neurons—either amping up or damping their activity. In turn, the brain cells trigger muscle fibers. The robot can flap its fins separately or together with the flexibility of a stingray or a butterfly.
Watching the robot move is mesmerizing, but the study isn’t just about cool visuals.
Robots have long tapped into examples of movement in nature to increase their dexterity and reduce energy usage. For now, the biohybrid bots can only live and operate in a nutritious soup of chemicals. But unlike previous designs, the bots push the field into the “brain-to-motor frontier” and could lead to autonomous systems “capable of advanced adaptive motor control and learning,” wrote study author Su Ryon Shin at Harvard Medical School and colleagues.
The technology could be a boon for biomedicine. Because it’s often compatible with living bodies, “tissue-based biohybrid robotics offers additional interdisciplinary insights in human health, medicine, and fundamental research in biology,” wrote Nicole Xu at the University of Colorado Boulder, who was not involved in the research.
Nature’s TouchScientists have long sought to develop soft, agile, and flexible robots that can navigate different terrain while using minimal energy—a far cry from the rigid, mechanical Terminator.
Often, they look to nature for ideas.
Thanks to evolution, every species on Earth has a fine-tuned system of movement tailored to its survival. Although each system differs—the brain wiring behind a butterfly flapping its wings is hardly similar to that of a blue whale spreading its fins—one central concept connects them all.
Each species needs a system that connects movement to its environment and quickly responds to stimuli. While this comes naturally to living creatures, robots often stumble when faced with unexpected challenges.
“Animals typically have a higher performance—such as increased energy efficiency, agility, and damage tolerance—compared to their robotic counterparts because of evolutionary pressures driving biological adaptations,” wrote Xu.
It’s no wonder scientists look to nature to design bioinspired robots. Two favorites are ray fishes and butterflies, both of which use very little energy to flap their fins or wings.
Last year, one team engineered a butterfly-like underwater robot with a synthetic hydrogel. Using light as a controller, it could flap its wings to swim upwards. Another mostly silicone minibot swam at high speeds with a “snapping” action, like when closing hairpins.
Both bots used entirely engineered materials and needed actuators to sense stimuli, say, light or pressure, and alter the robot’s moving components. Though successful, these can often fail.
Brain Meets MachineEnter biohybrid robots.
These bots use biological actuators to easily convert different types of energy used by the body—like, for example, automatically translating electricity or light into chemical energy.
The strategy has had successes, including ray-like robots that use muscle tissues to swim forward and turn using an external light source. Here, the light-controlled bots had a single layer of rat heart cells genetically engineered to respond to flashes of light. Compared to biobots built from purely synthetic materials, these could swim far longer.
The new study took this approach a step further by adding brain cells into the mix. Neurons form intricate connections with muscle cells to direct them when to flex.
The team used induced pluripotent stem cells (iPSCs) for their bot. Scientists make these cells by reverting skin cells into a stem cell-like state and then nudging them to form other cell types. In this case, they grew motor neurons, the brain cells that direct muscle movement, and muscle cells similar to those that keep the heart pumping. The cells linked up in a petri dish, allowing the neurons to control muscle contractions.
Living cells in hand, the team then assembled the robot’s two main components.
The first of these embeds neurons and muscle cells in a thin-film scaffold made of carbon nanotubes and gelatin—the main ingredient in Jello—and shaped into the robot’s body and fins.
The other is an “artificial brain” that controls the bot wirelessly using magnetic stimulation to change the electrical activity of the neurons, increasing or decreasing their activity.
Neuro-BotIn several tests, the team showed they could control the biohybrid bot’s behavior as it navigated its pool. Using multiple frequencies, each activating neurons for either the left or right fin, they easily steered the bot in a direct line and made turns.
Depending on the input, the bot could also flap a single fin, both fins, or alternate fins. The latter increased its stamina for longer swims—a bit like alternating arms in kayaking.
The bot’s neurons and muscle cells took the team by surprise by forming a type of connection that relies on electricity alone to transmit data. Normally, these connections, called synapses, need an additional chemical messenger to bridge communications, and they’re only one-way.
In contrast, the networks formed in the bot could transmit data in both directions faster and longer, controlling muscles up to 150 seconds or roughly 7.5 times longer than standard chemical synapses. And compared to bio-inspired systems using only synthetic materials, the biohybrid bot slashed energy needs.
For now, the minibots can only survive in a nutrient-rich soup of chemicals. But they show living components can be seamlessly integrated with electronics and non-biological scaffolding. Living robots could form the next generation of organoids-on-a-chip for study of diseases related to the brain and muscles or to test new drug treatments. Using purely electrical connections, which are easier to implement than standard chemical synapses, could help scale up the production of biohybrid bots.
“The advent of this bioelectronic neuromuscular robotic swimmer suggests a potential frontier [where we can] build autonomous biohybrid robotic systems that can achieve adaptive motor control, sensing, and learning,” wrote the team.
Image Credit: Hiroyuki Tetsuka
ARTIFICIAL INTELLIGENCEA Tiny New Open-Source AI Model Performs as Well as Powerful Big Ones
Melissa Heikkiläarchive page | MIT Technology Review“[The Allen Institute for Artificial Intelligence (Ai2)] claims that its biggest Molmo model, which has 72 billion parameters, outperforms OpenAI’s GPT-4o, which is estimated to have over a trillion parameters, in tests that measure things like understanding images, charts, and documents. Meanwhile, Ai2 says a smaller Molmo model, with 7 billion parameters, comes close to OpenAI’s state-of-the-art model in performance, an achievement it ascribes to vastly more efficient data collection and training methods.”
AUGMENTED REALITYHands On With Orion, Meta’s First Pair of AR Glasses
Alex Heath | The Verge“They look almost like a normal pair of glasses. That’s the first thing I notice as I walk into a conference room at Meta’s headquarters in Menlo Park, California. The black Clark Kent-esque frames sitting on the table in front of me look unassuming, but they represent CEO Mark Zuckerberg’s multibillion-dollar bet on the computers that come after smartphones. They’re called Orion, and they’re Meta’s first pair of augmented reality glasses.”
COMPUTINGStartup Says It Can Make a 100x Faster CPU
Dina Genkina | IEEE Spectrum“Instead of trying to speed up computation by putting 16 identical CPU cores into, say, a laptop, a manufacturer could put 4 standard CPU cores and 64 of Flow Computing’s so-called parallel processing unit (PPU) cores into the same footprint, and achieve up to 100 times better performance.”
TECHOpenAI to Become For-Profit Company
Deepa Seetharaman, Berber Jin, Tom Dotan | The Wall Street Journal“OpenAI is planning to convert from a nonprofit organization to a for-profit company at the same time it is undergoing major personnel shifts including the abrupt resignation Wednesday of its chief technology officer, Mira Murati. Becoming a for-profit would be a seismic shift for OpenAI, which was founded in 2015 to develop AI technology ‘to benefit humanity as a whole, unconstrained by a need to generate financial return,’ according to a statement it published when it launched.”
ROBOTICSDetachable Robotic Hand Crawls Around on Finger-Legs
Evan Ackerman | IEEE Spectrum“One of the great things about robots is that they don’t have to be constrained by our constraints, and at ICRA@40 in Rotterdam this week, we saw a novel new Thing: a robotic hand that can detach from its arm and then crawl around to grasp objects that would be otherwise out of reach, designed by roboticists from EPFL in Switzerland.”
SECURITYRemember That DNA You Gave 23andMe?
Kristen V. Brown | The Atlantic“23andMe is not doing well. Its stock is on the verge of being delisted. It shut down its in-house drug-development unit last month, only the latest in several rounds of layoffs. Last week, the entire board of directors quit, save for Anne Wojcicki, a co-founder and the company’s CEO. Amid this downward spiral, Wojcicki has said she’ll consider selling 23andMe—which means the DNA of 23andMe’s 15 million customers would be up for sale, too.”
BIOTECHAn Ultrathin Graphene Brain Implant Was Just Tested in a Person
Emily Mullin | Wired“Twenty years [after its discovery], graphene is finally making its way into batteries, sensors, semiconductors, air conditioners, and even headphones. And now, it’s being tested on people’s brains. This [week], surgeons at the University of Manchester temporarily placed a thin, Scotch-tape-like implant made of graphene on the patient’s cortex—the outermost layer of the brain. Made by Spanish company InBrain Neuroelectronics, the technology is a type of brain-computer interface, a device that collects and decodes brain signals.”
3D PRINTINGFirst 3D-Printed Hotel Ever Is Underway in Texas
Evan Garcia | Reuters“It looks like any other 3D printer—except it’s the size of a crane and is, layer by layer, building a hotel in the Texan desert. El Cosmico, an existing hotel and campground on the outskirts of the city of Marfa, is expanding. It is building 43 new hotel units and 18 residential homes over 60 acres (24 hectares)—all with a 3D printer.”
COMPUTINGAI Bots Now Beat 100% of Those Traffic-Image CAPTCHAs
Kyle Orland | Ars Technica“While there have been previous academic studies attempting to use image-recognition models to solve reCAPTCHAs, they were only able to succeed between 68 to 71 percent of the time. The rise to a 100 percent success rate ‘shows that we are now officially in the age beyond captchas,’ according to the new paper’s authors.”
Image Credit: Victor / Unsplash
Scientific breakthroughs rely on decades of diligent work and expertise, sprinkled with flashes of ingenuity and, sometimes, serendipity.
What if we could speed up this process?
Creativity is crucial when exploring new scientific ideas. It doesn’t come out of the blue: Scientists spend decades learning about their field. Each piece of information is like a puzzle piece that can be reshuffled into a new theory—for example, how different anti-aging treatments converge or how the immune system regulates dementia or cancer to develop new therapies.
AI tools could accelerate this. In a preprint study, a team from Stanford pitted a large language model (LLM)—the type of algorithm behind ChatGPT—against human experts in the generation of novel ideas over a range of research topics in artificial intelligence. Each idea was evaluated by a panel of human experts who didn’t know if it came from AI or a human.
Overall, ideas generated by AI were more out-of-the-box than those by human experts. They were also rated less likely to be feasible. That’s not necessarily a problem. New ideas always come with risks. In a way, the AI reasoned like human scientists willing to try out ideas with high stakes and high rewards, proposing ideas based on previous research, but just a bit more creative.
The study, almost a year long, is one of the biggest yet to vet LLMs for their research potential.
The AI ScientistLarge language models, the AI algorithms taking the world by storm, are galvanizing academic research.
These algorithms scrape data from the digital world, learn patterns in the data, and use these patterns to complete a variety of specialized tasks. Some algorithms are already aiding research scientists. Some can solve challenging math problems. Others are “dreaming up” new proteins to tackle some of our worst health problems, including Alzheimer’s and cancer.
Although helpful, these only assist in the last stage of research—that is, when scientists already have ideas in mind. What about having an AI to guide a new idea in the first place?
AI can already help draft scientific articles, generate code, and search scientific literature. These steps are akin to when scientists first begin gathering knowledge and form ideas based on what they’ve learned.
Some of these ideas are highly creative, in the sense that they could lead to out-the-box theories and applications. But creativity is subjective. One way to gauge potential impact and other factors for research ideas is to call in a human judge, blinded to the experiment.
“The best way for us to contextualize such capabilities is to have a head-to-head comparison” between AI and human experts, study author Chenglei Si told Nature.
The team recruited over 100 computer scientists with expertise in natural language processing to come up with ideas, act as judges, or both. These experts are especially well-versed in how computers can communicate with people using everyday language. The team pitted 49 participants against a state-of-the-art LLM based on Anthropic’s Claude 3.5. The scientists earned $300 per idea plus an additional $1,000 if their idea scored in the top 5 overall.
Creativity, especially when it comes to research ideas, is hard to evaluate. The team used two measures. First, they looked at the ideas themselves. Second, they asked AI and participants to produce writeups simply and clearly communicating the ideas—a bit like a school report.
They also tried to reduce AI “hallucinations”—when a bot strays from the factual and makes things up.
The team trained their AI on a vast catalog of research articles in the field and asked it to generate ideas in each of seven topics. To sift through the generated ideas and choose the best ones, the team engineered an automatic “idea ranker” based on previous data reviews and acceptance for publication from a popular computer science conference.
The Human CriticTo make it a fair test, the judges didn’t know which responses were from AI. To disguise them, the team translated submissions from humans and AI into a generic tone using another LLM. The judges evaluated ideas on novelty, excitement, and—most importantly—if they could work.
After aggregating reviews, the team found that, on average, ideas generated by human experts were rated less exciting than those by AI, but more feasible. As the AI generated more ideas, however, it became less novel, increasingly generating duplicates. Digging through the AI’s nearly 4,000 ideas, the team found around 200 unique ones that warranted more exploration.
But many weren’t reliable. Part of the problem stems from the fact the AI made unrealistic assumptions. It hallucinated ideas that were “ungrounded and independent of the data” it was trained on, wrote the authors. The LLM generated ideas that sounded new and exciting but weren’t necessarily practical for AI research, often because of latency or hardware problems.
“Our results indeed indicated some feasibility trade-offs of AI ideas,” wrote the team.
Novelty and creativity are also hard to judge. Though the study tried to reduce the likelihood the judges would be able to tell which submissions were AI and which human by rewriting them with an LLM, like a game of telephone, changes in length or wording may have subtly influenced how the judges perceived submissions—especially when it comes to novelty. Also, the researchers asked to come up with ideas were given limited time to do so. They admitted their ideas were about average compared to their past work.
The team agrees there’s more to be done when it comes to evaluating AI generation of new research ideas. They also suggested AI tools carry risks worthy of attention.
“The integration of AI into research idea generation introduces a complex sociotechnical challenge,” they said. “Overreliance on AI could lead to a decline in original human thought, while the increasing use of LLMs for ideation might reduce opportunities for human collaboration, which is essential for refining and expanding ideas.”
That said, new forms of human-AI collaboration, including AI-generated ideas, could be useful for researchers as they investigate and choose new directions for their research.
Image Credit: Calculator Land / Pixabay
Aviation has proven to be one of the most stubbornly difficult industries to decarbonize. But a new roadmap outlined by University of Cambridge researchers says the sector could reach net zero by 2050 if urgent action is taken.
The biggest challenge when it comes to finding alternatives to fossil fuels in aviation is basic physics. Jet fuel is incredibly energy dense, which is crucial for a mode of transport where weight savings can dramatically impact range.
While efforts are underway to build planes powered by batteries, hydrogen, or methane, none can come close to matching kerosene, pound for pound, at present. Sustainable aviation fuel is another option, but so far, its uptake has been limited, and its green credentials are debatable.
Despite this, the authors of a new report from the University of Cambridge’s Aviation Impact Accelerator (AIA) say that with a concerted effort the industry can clean up its act. The report outlines four key sustainable aviation goals that, if implemented within the next five years, could help the sector become carbon neutral by the middle of the century.
“Too often the discussions about how to achieve sustainable aviation lurch between overly optimistic thinking about current industry efforts and doom-laden cataloging of the sector’s environmental evils,” Eliot Whittington, executive director at the Cambridge Institute for Sustainability Leadership, said in a press release.
“The Aviation Impact Accelerator modeling has drawn on the best available evidence to show that there are major challenges to be navigated if we’re to achieve net zero flying at scale, but that it is possible.”
The report notes that time is of the essence. Aviation is responsible for roughly 4 percent of global warming despite only 10 percent of the population flying, a figure that’s likely to rise as the world continues to develop. Despite global leaders pledging to make aviation net zero, current efforts to get there are not ambitious enough, the authors say.
After researching the interventions that could have the biggest impact and discussions at the inaugural meeting of the Transatlantic Sustainable Aviation Partnership at MIT last year, AIA came up with four focus areas that could put those goals within reach.
The first of these is to reduce contrails. While most of the focus is on emissions from burning jet fuel, the generation of persistent contrails can trap heat in atmosphere and add significantly to warming.
Contrails can be avoided by adjusting an aircraft’s altitude in areas where they’re most likely to be formed, but the underlying science is poorly understood as are potential strategies for adjusting air traffic. Therefore, the report suggests setting up several “living labs” in existing airspace to conduct data collection and experiments. These should be ready by the end of 2025, say the authors.
The second goal is to reduce the amount of fuel airplanes use by introducing new aircraft and engine designs, improving operational efficiency of the sector, or just getting aircraft to fly slower. To catalyze action, governments need to set clear policies, such as establishing fuel burn reduction targets, loan guarantees for new aircraft purchases, or incentives to scrap old airplanes.
The third goal is to ensure sustainable aviation fuel is actually sustainable, and its production is scalable. Most sustainable fuels rely on biomass, but limitations on production and competition from other sectors could mean they can’t realize the hoped for emissions reductions.
In the near term, the report suggests aviation will have to work with other industries to set best practices and limit total cross-sector emissions. And in the long run, the industry will have to make efforts to find alternative ways to develop synthetic sustainable fuels.
Lastly, the report argues the industry also needs to invest in “moonshot” technologies. By 2025, aviation should launch several high-risk, high-reward demonstration programs in technologies that could be truly transformative for the sector. These include the development of cryogenic hydrogen or methane fuels, hydrogen-electric propulsion technology, or the use of synthetic biology to dramatically lower the energy demands for sustainable fuel production.
The report’s authors stress that, although they are confident these interventions could have the desired impact, time is of the essence. History suggests that getting global leaders to take decisive action on climate issues is tricky, but at least they now have a concrete roadmap.
Image Credit: John McArthur / Unsplash
For the past few years, a series of controversies have rocked the well-established field of cosmology. In a nutshell, the predictions of the standard model of the universe appear to be at odds with some recent observations.
There are heated debates about whether these observations are biased, or whether the cosmological model, which predicts the structure and evolution of the entire universe, may need a rethink. Some even claim that cosmology is in crisis. Right now, we do not know which side will win. But excitingly, we are on the brink of finding that out.
To be fair, controversies are just the normal course of the scientific method. And over many years, the standard cosmological model has had its share of them. This model suggests the universe is made up of 68.3 percent “dark energy” (an unknown substance that causes the universe’s expansion to accelerate), 26.8 percent dark matter (an unknown form of matter) and 4.9 percent ordinary atoms, very precisely measured from the cosmic microwave background—the afterglow of radiation from the Big Bang.
It explains very successfully multitudes of data across both large and small scales of the universe. For example, it can explain things like the distribution of galaxies around us and the amount of helium and deuterium made in the universe’s first few minutes. Perhaps most importantly, it can also perfectly explain the cosmic microwave background.
This has led to it gaining the reputation as the “concordance model.” But a perfect storm of inconsistent measurements—or “tensions” as they’re known as in cosmology—are now questioning the validity of this longstanding model.
Uncomfortable TensionsThe standard model makes particular assumptions about the nature of dark energy and dark matter. But despite decades of intense observation, we still seem no closer to working out what dark matter and dark energy are made of.
The litmus test is the so-called Hubble tension. This relates to the Hubble constant, which is the rate of expansion of the universe at the present time. When measured in our nearby, local universe, from the distance to pulsating stars in nearby galaxies, called Cepheids, its value is 73 km/s/megaparsec (Mpc is a unit of measure for distances in intergalactic space). However, when predicted theoretically, the value is 67.4 km/s/Mpc. The difference may not be large (only 8 percent), but it is statistically significant.
The Hubble tension became known about a decade ago. Back then, it was thought that the observations may have been biased. For example, the cepheids, although very bright and easy to see, were crowded together with other stars, which could have made them appear even brighter. This could have made the Hubble constant higher by a few percent compared to the model prediction, thus artificially creating a tension.
With the advent of the James Webb Space Telescope (JWST), which can separate the stars individually, it was hoped that we would have an answer to this tension.
Frustratingly, this hasn’t yet happened. Astronomers now use two other types of stars besides the cepheids (known as the tip of the red giant branch stars (TRGB) and the J-region asymptotic giant branch (JAGB) stars). But while one group has reported values from the JAGB and TRGB stars that are tantalizingly close to the value expected from the cosmological model, another group has claimed that they are still seeing inconsistencies in their observations. Meanwhile, the cepheids measurements continue to show a Hubble tension.
It’s important to note that although these measurements are very precise, they may still be biased by some effects uniquely associated with each type of measurement. This will affect the accuracy of the observations, in a different way for each type of stars. A precise but inaccurate measurement is like trying to have a conversation with a person who is always missing the point. To solve disagreements between conflicting data, we need measurements that are both precise and accurate.
The good news is that the Hubble tension is now a rapidly developing story. Perhaps we will have the answer to it within the next year or so. Improving the accuracy of data, for example by including stars from more far away galaxies, will help sort this out. Similarly, measurements of ripples in spacetime known as gravitational waves will also be able to help us pin down the constant.
This may all vindicate the standard model. Or it may hint that there’s something missing from it. Perhaps the nature of dark matter or the way that gravity behaves on specific scales is different to what we believe now. But before discounting the model, one has to marvel at its unmatched precision. It only misses the mark by at most a few percent, while extrapolating over 13 billion years of evolution.
To put it into perspective, even the clockwork motions of planets in the solar system can only be computed reliably for less than a billion years, after which they become unpredictable. The standard cosmological model is an extraordinary machine.
The Hubble tension is not the only trouble for cosmology. Another one, known as the “S8 tension,” is also causing trouble, albeit not on the same scale. Here the model has a smoothness problem, by predicting that matter in the universe should be more clustered together than we actually observe—by about 10 percent. There are various ways to measure the “clumpiness” of matter, for example by analyzing the distortions in the light from galaxies produced by the assumed dark matter intervening along the line of sight.
Currently, there seems to be a consensus in the community that the uncertainties in the observations have to be teased out before ruling out the cosmological model. One possible way to alleviate this tension is to better understand the role of gaseous winds in galaxies, which can push out some of the matter, making it smoother.
Understanding how clumpiness measurements on small scales relate to those on larger scales would help. Observations might also suggest there is a need to change how we model dark matter. For example, if instead of being made entirely of cold, slow moving particles, as the standard model assumes, dark matter could be mixed up with some hot, fast-moving particles. This could slow down the growth of clumpiness at late cosmic times, which would ease the S8 tension.
JWST has highlighted other challenges to the standard model. One of them is that early galaxies appear to be much more massive that expected. Some galaxies may weigh as much as the Milky Way today, even though they formed less than a billion years after the Big Bang, suggesting they should be less massive.
A region of star formation seen by JWST and the Chandra telescope. Image credit: Credit: X-ray: NASA/CXO/SAO; Infrared: NASA/ESA/CSA/STScI; Image processing: NASA/CXC/SAO/L. Frattare, CC BYHowever, the implications against the cosmological model are less clear in this case, as there may be other possible explanations for these surprising results. Improving the measurement of stellar masses in galaxies is key to solving this problem. Rather than measuring them directly, which is not possible, we infer these masses from the light emitted by galaxies.
This step involves some simplifying assumptions, which could translate into overestimating the mass. Recently, it has also been argued that some of the light attributed to stars in these galaxies is generated by powerful black holes. This would imply that these galaxies may not be as massive after all.
Alternative TheoriesSo, where do we stand now? While some tensions may soon be explained by more and better observations, it is not yet clear whether there will be a resolution to all of the challenges battering the cosmological model.
There has been no shortage of theoretical ideas of how to fix the model though—perhaps too many, in the range of a few hundred and counting. That’s a perplexing task for any theorist who may wish to explore them all.
The possibilities are many. Perhaps we need to change our assumptions of the nature of dark energy. Perhaps it is a parameter that varies with time, which some recent measurements have suggested. Or maybe we need to add more dark energy to the model to boost the expansion of the universe at early times, or, on the contrary, at late times. Modifying how gravity behaves on large scales of the universe (differently than done in the models called Modified Newtonian Dynamics, or MOND) may also be an option.
So far, however, none of these alternatives can explain the vast array of observations the standard model can. Even more worrisome, some of them may help with one tension but worsen others.
The door is now open to all sorts of ideas that challenge even the most basic tenets of cosmology. For example, we may need to abandon the assumption that the universe is “homogeneous and isotropic” on very large scales, meaning it looks the same in all directions to all observers and suggesting there are no special points in the universe. Others propose changes to the theory of general relativity.
Some even imagine a trickster universe, which participates with us in the act of observation, or which changes its appearance depending on whether we look at it or not—something we know happens in the quantum world of atoms and particles.
In time, many of these ideas will likely be relegated to the cabinet of curiosities of theorists. But in the meantime, they provide a fertile ground for testing the “new physics.”
This is a good thing. The answer to these tensions will no doubt come from more data. In the next few years, a powerful combination of observations from experiments such as JWST, the Dark Energy Spectroscopic Instrument (DESI), the Vera Rubin Observatory and Euclid, among many others, will help us find the long-sought answers.
Tipping PointOn one side, more accurate data and a better understanding of the systematic uncertainties in the measurements could return us to the reassuring comfort of the standard model. Out of its past troubles, the model may emerge not only vindicated, but also strengthened, and cosmology will be a science that is both precise and accurate.
But if the balance tips the other way, we will be ushered into uncharted territory, where new physics will have to be discovered. This could lead to a major paradigm shift in cosmology, akin to the discovery of the accelerated expansion of the universe in the late 1990s. But on this path we may have to reckon, once and for all, with the nature of dark energy and dark matter, two of the big unsolved mysteries of the universe.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA, ESA, CSA, STScI, Webb ERO Production Team
Roughly 20 years ago, at the dawn of YouTube, a video of a sneezing baby panda—and its mom’s gasp in surprise—captured the internet’s heart.
With their distinctive black and white fur, giant pandas are known for their calm nature, playfulness, and utter cuteness. The gentle beasts are native to China, but their charm has enthralled the world, including bridging international relationships through “panda diplomacy.” The bear has been the logo for the World Wildlife Fund since its founding in 1961.
Despite conservation attempts, these cuddly bears are still incredibly vulnerable.
As of now, only 2,000 pandas remain in the wild. The animals live in small populations scattered across a few mountainous regions in midwestern China. They mostly eat bamboo, but in recent decades, bamboo forests have been decimated by human activities, like roadbuilding, logging, and the conversion of natural environments into pasture.
As the number of pandas dwindles, so do their chances of survival. A recent census showed pandas living in 33 isolated populations across their preferred landscapes. Roughly half of these groups could face up to 90 percent extinction in the years ahead.
Unfortunately, it’s a tale as old as time. “This iconic species faces substantial threats to its survival due to various human activates in its habitat,” wrote Jing Liu and colleagues at the Chinese Academy of Sciences in a recent study. Saving its habitat is one way to keep the species alive and thriving. But economic incentives make it a hard legislative hill to climb.
What about a backup plan?
Last week, in their study, Liu and his team took a page out of the de-extinction playbook to propose a new way to conserve pandas: Convert their skin cells into stem cells. These, in theory, can then be turned into any cell type in the body—including reproductive cells for breeding.
It “is really a great breakthrough in the field of giant panda conservation,” Thomas Hildebrandt at the Free University of Berlin, who was not involved in the research, told Science News.
Panda AcademyPandas thrive in several provinces of China, where forests are lush with bamboo, their preferred food. The bears, with their signature black and white coats, are remarkably distinct from grizzlies or black bears. Their forepaws are especially agile. Like people lounging on couches, they use a thumb-like structure to grab and bring bamboo directly into their mouths, while keeping their bodies relatively still on the ground.
Although they have large teeth and a strong jaw, pandas are generally pacificists with a jolly personality. In nature, mothers raise their pups for up to two years before sending them into the wild under a watchful eye.
Panda numbers rapidly dwindled in the 1980s. Deforestation, poaching, and loss of bamboo forests slashed their population to near extinction. Thanks to the World Wildlife Fund, their numbers have recently rebounded. Greater awareness of their plight garnered support, and their numbers have slowly grown in captivity and in the wild.
But their small population still poses a genetic conundrum. Inbreeding among groups can lead to genetic disease, loss of genetic diversity, and potentially less resilience against infections.
Genetic RepriseA potential way to combat these problems is to develop induced pluripotent stem cells (iPSCs) in pandas. The Noble Prize-winning technology has taken the biomedical field by storm over the last two decades by showing skin cells can be reverted into a stem cell-like state.
The trick already works in human and mouse skin cells. Researchers use it to grow iPSCs into mini-brains, embryo-like structures, and early reproductive cells.
The technology has “shown promising outcomes in the conservation” of genes in multiple endangered species too, the authors wrote. Among these are the northern white rhinoceros, the Tasmanian devil, the Sumatran rhinoceros, and others.
But the recipe for making iPSCs differs between species. Reprogramming genes that work in mice and human cells doesn’t always work in other cell types or species.
“The recipe from the mouse is not necessarily directly applicable to other species, even within mammalian species,” the Smithsonian’s Pierre Comizzoli, who was not involved in the study, said in an interview with The Scientist.
Panda-moniumA few years back, researchers found a way to transform cells from the soft part of a panda’s cheek into little bulbs of a particular stem cell type. These could be coaxed into some types of skin and other cells, but they lacked the flexibility to generate any tissues.
The new study aimed to remedy this by reprogramming skin cells into iPSCs.
The team took skin samples from a male and female named Xingrong and Loubao. The procedure involved painlessly scraping off skin cells, a bit like a daily skincare routine.
After collecting the cells, the team bathed them in a chemical soup to help the cells grow and divide. A few additional genes transformed them into giant panda iPSCs.
“The clones were very beautiful. We were so excited,” Liu told The Scientist.
The engineered panda stem cells were close to those normally developed inside the body. Although not yet an exact mimic, the engineered cells form a foundation for how panda cells develop. The library of genetic changes, in turn, could help with their preservation.
The team also tested the engineered stem cells on a hallmark of development. Stem cells form three different layers of cells, each of which can develop into various tissues and organs. In petri dishes, the panda iPSCs mimicked the process, generating cells and protein communication that roughly copied early stages in the formation of reproductive cells.
The results show how reprogramming cells could help us preserve and study endangered species. Adding panda iPSCs to our evolutionary library is another step toward conserving the lovable bears. With more work, the cells could potentially be turned into sperm and eggs in a lab, without harming any pandas in the process. The reprogrammed cells may also become a useful proxy scientists can use to test therapies that increase panda fertility.
But realizing those ideas is still off in the future.
“The most immediate applications are in regenerative medicine to treat sick pandas and to better understand the embryology or fetal development of these animals,” said Comizzoli.
Image Credit: Pascal Müller / Unsplash
TECHAI’s Hungry Maw Drives Massive $100B Investment Plan by Microsoft and BlackRock
Benj Edwards | Ars Technica“The partnership initially aims to raise $30 billion in private equity capital, which could later turn into $100 billion in total investment when including debt financing. The group will invest in data centers and supporting power infrastructure for AI development. ‘The capital spending needed for AI infrastructure and the new energy to power it goes beyond what any single company or government can finance,’ Microsoft President Brad Smith said in a statement.”
COMPUTINGChallengers Are Coming for Nvidia’s Crown
Matthew S. Smith | IEEE Spectrum“[Nvidia has] a deep, broad moat with which to defend its business, but that doesn’t mean it lacks competitors ready to storm the castle, and their tactics vary widely. While decades-old companies like Advanced Micro Devices (AMD) and Intel are looking to use their own GPUs to rival Nvidia, upstarts like Cerebras and SambaNova have developed radical chip architectures that drastically improve the efficiency of generative AI training and inference. These are the competitors most likely to challenge Nvidia.”
BIOTECHModerna’s ‘Off-the-Shelf’ Cancer Vaccine Shows Promise in Early Human Trial Data
Ed Cara | Gizmodo“The future of cancer treatment is continuing to look bright. Over the weekend, researchers in the UK announced encouraging results from an early trial testing an mRNA vaccine against advanced solid cancers. The vaccine, developed by Moderna, is designed to help people’s immune systems better recognize and kill cancerous cells.”
ROBOTICS1X Releases Generative World Models to Train Robots
Ben Dickson| VentureBeat“1X’s new system is inspired by innovations such as OpenAI Sora and Runway, which have shown that with the right training data and techniques, generative models can learn some kind of world model and remain consistent through time. However, while those models are designed to generate videos from text, 1X’s new model is part of a trend of generative systems that can react to actions during the generation phase.”
GENE THERAPYFirst Day of a ‘New Life’ for a Boy With Sickle Cell
Gina Kolata | The New York Times“Kendric Cromer, 12, is among the first patients to be treated with gene therapy just approved by the FDA that many other patients face obstacles to receiving. …Last December, the Food and Drug Administration gave approval to two companies, Bluebird Bio of Somerville, Mass., and Vertex Pharmaceuticals of Boston, to sell the first gene therapies approved for sickle cell disease. After nine months, Kendric remains the first Bluebird patient to progress this far, with at least a few others advancing toward his pace.”
ARTIFICIAL INTELLIGENCEOpenAI’s New Model Is Better at Reasoning and, Occasionally, Deceiving
Kylie Robison | The Verge“While AI models have been able to ‘lie’ in the past, and chatbots frequently output false information, o1 had a unique capacity to ‘scheme’ or ‘fake alignment.’ That meant it could pretend it’s following the rules to complete a given task, but it isn’t actually. To the model, the rules could be too much of a burden, and it seems to have the ability to disregard them if it means it can more easily complete a task.”
GOVERNANCEAI-Generated Content Doesn’t Seem to Have Swayed Recent European Elections
Melissa Heikkiläarchive page | MIT Technology Review“Since the beginning of the generative-AI boom, there has been widespread fear that AI tools could boost bad actors’ ability to spread fake content with the potential to interfere with elections or even sway the results. Such worries were particularly heightened this year, when billions of people were expected to vote in over 70 countries. Those fears seem to have been unwarranted, says Sam Stockwell, the researcher at the Alan Turing Institute who conducted the study.”
ENERGYEvery Fusion Startup That Has Raised Over $300M
Tim De Chant | TechCrunch“Over the last several years, fusion power has gone from the butt of jokes—always a decade away!—to an increasingly tangible and tantalizing technology that has drawn investors off the sidelines. …Founders have built on that momentum in recent years, pushing the private fusion industry forward at a rapid pace. Fusion startups have raised $7.1 billion to date, according to the Fusion Industry Association, with the majority of it going to a handful of companies.”
TECHI Stared Into the AI Void With the SocialAI App
Lauren Goode | Wired“Even the app’s creator, Michael Sayman, admits that the premise of SocialAI may confuse people. His announcement this week of the app read a little like a generative AI joke: ‘A private social network where you receive millions of AI-generated comments offering feedback, advice, and reflections.’ But, no, SocialAI is real, if ‘real’ applies to an online universe in which every single person you interact with is a bot.”
AUGMENTED REALITYHere’s What I Made of Snap’s New Augmented-Reality Spectacles
Mat Honan | The Guardian“Before I get to Snap’s new Spectacles, a confession: I have a long history of putting goofy new things on my face and liking it. …[I spent] the better part of [a] year with Google’s ridiculous Glass on my face and thought it was the future. Microsoft HoloLens? Loved it. Google Cardboard? Totally normal. Apple Vision Pro? A breakthrough, baby. …I got to try [Snap’s new AR glasses] out a couple of weeks ago. They are pretty great! (But also: See above.)”
GOVERNANCEThere Are More Than 120 AI Bills in Congress Right Now
Scott J. Mulligan | MIT Technology Review“US policymakers have an ‘everything everywhere all at once’ approach to regulating artificial intelligence, with bills that are as varied as the definitions of AI itself. …That’s why, with help from the Brennan Center for Justice, which created a tracker with all the AI bills circulating in various committees in Congress right now, MIT Technology Review has taken a closer look to see if there’s anything we can learn from this legislative smorgasbord.”
Image Credit: Shubham Dhage / Unsplash
DNA is nature’s computing device.
Unlike data centers, DNA is incredibly compact. These molecules package an entire organism’s genetic blueprint into tiny but sophisticated structures inside each cell. Kept cold—say, inside a freezer or in the Siberian tundra—DNA and the data encoded within can last millennia.
But DNA is hardly just a storage device. Myriad molecules turn genes on and off—a bit like selectively running bits of code—to orchestrate everyday cellular functions. The body “reads” bits of the genetic code in a particular cell at a specific time and, together, compiles the data into a smoothly operating, healthy life.
Scientists have long eyed DNA as a computing device to complement everyday laptops. With the world’s data increasing at an exponential rate, silicon chips are struggling to meet the demands of data storage and computation. The rise of large language models and other modes of artificial intelligence is further pushing the need for alternative solutions.
But the problem with DNA storage is it often gets destroyed after “reading” the data within.
Last month, a team from North Carolina State University and Johns Hopkins University found a workaround. They embedded DNA molecules, encoding multiple images, into a branched gel-like structure resembling a brain cell.
Dubbed “dendricolloids,” the structures stored DNA files far better than those freeze-dried alone. DNA within dendricolloids can be repeatedly dried and rehydrated over roughly 170 times without damaging stored data. According to one estimate, each DNA strand could last over two million years at normal freezer temperatures.
Unlike previous DNA computers, the data can be erased and replaced like memory on classical computers to solve multiple problems—including a simple chess game and sudoku.
Until now, DNA was mainly viewed as a long-term storage device or single-use computer. Developing DNA technology that can store, read, “rewrite, reload, or compute specific data files” repeatedly seemed difficult or impossible, said study author Albert Keung in a press release.
However, “we’ve demonstrated that these DNA-based technologies are viable, because we’ve made one,” he said.
A Grain of SandThis is hardly the first attempt to hijack the code of life to increase storage and computation.
The first steps taken were in data storage. Our computers run on binary bits of information encoded in zeros and ones. DNA, in contrast, uses four different molecules typically represented by the letters A, T, C, and G. This means that different pairs of zeros and ones—00, 01, 10, 11—can be encoded into different DNA letters. Because of the way it’s packaged in cells, DNA can theoretically store far more data in less space than digital devices.
“You could put a thousand laptops’ worth of data into DNA-based storage that’s the same size as a pencil eraser,” said Keung.
With any computer, we need to be able to search and retrieve information. Our cells have evolved mechanisms that read specific parts of a DNA strand on demand—a sort of random access memory that extracts a particular piece of data. Previous studies have tapped into these systems to store and retrieve books, images, and GIFs inside DNA files. Scientists have also used microscopic glass beads with DNA “labels” as a kind of filing system for easy extraction.
But storing and extracting data is only half of the story. A computer needs to, well, compute.
Last year, a team developed a programmable DNA computer that can run billions of different circuits with minimal energy. Traditionally, these molecular machines work by allowing different strands to grab onto each other depending on calculation needs. Different pairs could signal “and,” “or,” and “not” logic gates—recapitulating the heart of today’s digital computers.
But reading and computing often destroys the original DNA data, making most DNA-based systems single-use. Scientists have also developed another type of DNA computer, which monitors changes in the molecule’s structures. These can be rewritten. Similar to standard hard drives, they can encode multiple rounds of data, but they’re also harder to scale.
DNA Meets DataThe new study combined the best of both worlds. The team engineered a DNA computer that can store information, perform computations, and reset the system for another round.
The core of the system relies on a central dogma in biology. DNA sits in a small cage within cells. When genes are turned on, their data is translated into RNA, which converts the genetic blueprint into proteins. If DNA is safely stored, adding protein “switches” that turn genes up or down changes the genetic readout in RNA but keeps the original genetic sequences intact.
Because the original data doesn’t change, it’s possible to run multiple rounds of RNA-based calculations from a single DNA-encoded dataset—with improvements.
Based on these ideas, the team engineered a jelly-like structure with branches similar to a brain cell. Dubbed “dendricolloids,” the soft materials allowed each DNA strand to grab onto surrounding material “without sacrificing the data density that makes DNA attractive for data storage in the first place,” said study author Orlin Velev.
“We can copy DNA information directly from the material’s surface without harming the DNA. We can also erase targeted pieces of DNA and then rewrite to the same surface, like deleting and rewriting information stored on the hard drive,” said study author Kevin Lin.
To test out their system, the team embedded a synthetic DNA sequence of 200 letters into the material. Adding a molecular cocktail that converts DNA sequences into RNA, the material was able to generate RNA repeatedly over 10 rounds. In theory, the resulting RNA could encode 46 terabytes of data stored at normal fridge and freezer temperatures.
The dendricolloids could also absorb over 2,700 different DNA strands, each nearly 250 letters long to protect their data. In one test, the team encoded three different JPEG files into the structures, translating digital data into biological data. In simulations that mimicked accessing the DNA files, the team could reconstruct the data 10 times without losing it in the process.
Game OnThe team next took inspiration from a biological “eraser” of sorts. These proteins eat away at RNA without damaging the DNA blueprint. This process controls how a cell performs its usual functions—for example, by destroying RNA strands detrimental to health.
As a proof of concept, the team developed 1,000 different DNA snippets to solve multiple puzzles. For a simple game of chess, each DNA molecule encoded nine potential positions. The molecules were pooled, with each representing a potential configuration. This data allowed the system to learn. For example, one gene, when turned on, could direct a move on the chessboard by replicating itself in RNA. Another could lower RNA levels detrimental to the game.
These DNA to RNA processes were controlled by an engineered protein whose job it was to keep the final results in check. As a last step, all RNA strands violating the rules were destroyed, leaving behind only those representing the final, expected solution. In addition to chess, the team implemented this process to solve simple sudoku puzzles too.
The DNA computer is still in its infancy. But unlike previous generations, this one captures storage and compute in one system.
“There’s a lot of excitement about molecular data storage and computation, but there have been significant questions about how practical the field may be,” said Keung. “We wanted to develop something that would inspire the field of molecular computing.”
Image Credit: Luke Jones / Unsplash
Norway’s sizable oil and gas deposits have made it one of the wealthiest countries in the world. That’s why it might come as a surprise that it’s the first country to have more electric vehicles than gasoline-powered ones.
Transportation is the single biggest contributor to climate change in the US—accounting for 28 percent of total greenhouse gas emissions, according to the Environmental Protection Agency. So, the rise of electric vehicles has been one of the biggest success stories in the effort to clean up the economy.
Slowing sales growth for battery-powered cars has some worried there might be a ceiling to the number of people willing to adopt the technology. But Norway shows that with the right incentives, the goal of a completely electrified road network is a tangible possibility.
Earlier this week, the Norwegian Road Federation (OFV) announced that of the 2.8 million private cars that are registered in the country, 754,303 are all-electric compared to 753,905 that run on gasoline.
“This is historic. A milestone few saw coming 10 years ago,” OFV director Øyvind Solberg Thorsen told The Guardian. “The electrification of the fleet of passenger cars is going quickly, and Norway is thereby rapidly moving towards becoming the first country in the world with a passenger car fleet dominated by electric cars.”
This tipping point had been long anticipated, as electric vehicle sales in Norway have massively outpaced gasoline cars for some time. Roughly 85 percent of new vehicles registered in 2024 so far have been zero-emissions, which refers to fully battery-powered vehicles and excludes hybrids.
It’s no secret how the country got here. The Norwegian government has given generous subsidies to promote adoption, including tax rebates that bring the cost of electric vehicles down to similar levels as conventional vehicles, exemptions from some tolls, and an extensive public network of free chargers.
Despite overtaking gasoline-powered cars, electric vehicles are still lagging diesel ones, which account for more than a million of Norway’s existing stock. But the government has an ambitious goal to end the sale of new gasoline and diesel cars by next year, so it may not be long before they catch up.
How easily other countries can mimic their success remains to be seen though—tax exemptions on electric vehicles cost 43 billion kroner ($4.1 billion) in 2023. Norway has been able to pay for this thanks to the country’s massive $1.7 trillion sovereign wealth fund, which, ironically, was built using the profits from its enormous fossil fuel reserves.
Electric vehicle sales have been highly concentrated in three main markets—Europe, the US, and China—accounting for roughly 95 percent of all purchases. In the US, new registrations grew 40 percent last year to hit 1.4 million, while Europe saw a 20 percent increase to 3.2 million.
However, sales have been flagging in recent months, even as production capacity continues to ramp up. This has some worried that concerns around pricing and charging infrastructure could cap consumers’ willingness to make the switch. A brewing trade war over electric vehicles between the West and China also threatens to further dent adoption.
While it might not come cheap, if we’re committed to decarbonizing our transportation system, other governments may need to follow Norway’s lead when it comes to incentivizing cleaner cars.
Image Credit: Emil Dosen / Unsplash
The rings of Saturn are some of the most famous and spectacular objects in the solar system. Earth may once have had something similar.
In a paper published last week in Earth and Planetary Science Letters, my colleagues and I present evidence that Earth may have had a ring.
The existence of such a ring, forming around 466 million years ago and persisting for a few tens of millions of years, could explain several puzzles in our planet’s past.
The Case for a Ringed EarthAround 466 million years ago, a lot of meteorites started hitting Earth. We know this because many impact craters formed in a geologically brief period.
In the same period, we also find deposits of limestone across Europe, Russia, and China containing very high levels of debris from a certain type of meteorite. The meteorite debris in these sedimentary rocks shows signs that they were exposed to space radiation for much less time than we see in meteorites that fall today.
Many tsunamis also occurred at this time, as can be seen from other unusual jumbled up sedimentary rocks.
We think all these features are likely related to one another. But what links them together?
A Pattern of CratersWe know of 21 meteorite impact craters that formed during this high-impact period. We wanted to see if there was a pattern in their locations.
Using models of how Earth’s tectonic plates moved in the past, we mapped out where all these craters were when they first formed. We found all of the craters are on continents that were close to the equator in this period, and none are in places that were closer to the poles.
So, all the impacts occurred close to the equator. But is this actually a fair sample of the impacts that occurred?
Well, we measured how much of Earth’s land surface suitable for preserving a crater was near the equator at that time. Only about 30 percent of the suitable land was close to the equator, with 70 percent at higher latitudes.
Under normal circumstances, asteroids hitting Earth can hit at any latitude, at random, as we see in craters on the moon, Mars, and Mercury.
Impact craters on the far side of the Moon are quite evenly distributed. Image Credit: Lunar Reconnaissance Orbiter / NASA / GSFC / Arizona State UniversitySo it’s extremely unlikely that all 21 craters from this period would form close to the equator if they were unrelated to one another. We would expect to see many other craters at higher latitudes as well.
We think the best explanation for all this evidence is that a large asteroid broke up during a close encounter with Earth. Over several tens of millions of years, the asteroid’s debris rained down onto Earth, creating the pattern of craters, sediments, and tsunamis we describe above.
How Rings FormYou may know that Saturn isn’t the only planet with rings. Jupiter, Neptune, and Uranus have less obvious rings, too. Some scientists have even suggested that Phobos and Deimos, the small moons of Mars, may be remnants of an ancient ring.
So, we know a lot about how rings form. Here’s how it works.
Saturn backlit by the sun, taken by the Cassini spacecraft. Image Credit: Cassini Imaging Team / SSI / JPL / ESA / NASAWhen a small body (like an asteroid) passes close to a large body (like a planet), it gets stretched by gravity. If it gets close enough (inside a distance called the Roche limit), the small body will break apart into lots of tiny pieces and a small number of bigger pieces.
All those fragments will be jostled around and gradually evolve into a debris ring orbiting the equator of the larger body. Over time, the material in the ring will fall down to the larger body, where the larger pieces will form impact craters. These craters will be located close to the equator.
So, if Earth destroyed and captured a passing asteroid around 466 million years ago, it would explain the anomalous locations of the impact craters, the meteorite debris in sedimentary rocks, craters and tsunamis, and the meteorites’ relatively brief exposure to space radiation.
A Giant Sunshade?Back then, the continents were in different positions due to continental drift. Much of North America, Europe, and Australia were close to the equator, whereas Africa and South America were at higher southern latitudes.
The ring would have been around the equator. And since Earth’s axis is tilted relative to its orbit around the sun, the ring would have shaded parts of Earth’s surface.
This shading in turn might have caused global cooling, as less sunlight reached the planet’s surface.
This brings us to another interesting puzzle. Around 465 million years ago, our planet began cooling dramatically. By 445 million years ago it was in the Hirnantian Ice Age, the coldest period in the past half a billion years.
Was a ring shading Earth responsible for this extreme cooling? The next step in our scientific sleuthing is to make mathematical models of how asteroids break up and disperse and how the resulting ring evolves over time. This will set the scene for climate modeling that explores how much cooling could be imposed by such a ring.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Artist’s impression of Earth with rings like Saturn / Oliver Hull
A drug that slows aging may already be on the market.
Scientists have long been interested in metformin, a widely prescribed drug used to treat Type 2 diabetes, for its potential to delay aging. In worms, fruit flies, and rodents, the drug—on average less than a dollar per pill—shows promising anti-aging effects.
Last week, a study in Cell added to the evidence that metformin could slow the ravages of time. Scientists gave male monkeys aged the equivalent of 52 to 64 in human years a daily pill for three years and monitored their physical health and cognition.
Compared to naturally aging monkeys, metformin preserved their learning and memory abilities, reduced brain shrinkage, and restored their neurons to a more youthful state. The monkeys’ “brain age” was dialed back by almost 6 years, or around 18 human years.
Metformin’s effects extended beyond the brain. The drug reduced chronic inflammation—a hallmark of aging—in multiple tissues, slowed liver aging, and boosted cellular mechanisms that protect the liver. Kidneys, lungs, and muscles were also “rescued” from age-related problems, their gene expression profiles reverting to more youthful ones.
The study bridges the gap between rodents and primates. The dosages of metformin given were on par with those for diabetes management and could inform upcoming clinical trials.
To be clear, the study did not examine longevity, that is, how long the monkeys lived. Rather, it focused on the slowing of age-related diseases, with the results contributing to our understanding of health span—the number healthy living years people experience.
This study is the “most quantitative, thorough examination of metformin action that I’ve seen beyond mice,” Dr. Alex Soukas at Massachusetts General Hospital told Nature, who was not involved in the study.
Old Dog, New TricksMetformin may have a shiny new reputation for battling aging, but it’s already had a long life in medicine.
First extracted from a plant called goat’s rue, a traditional herbal medicine in Europe, researchers in 1918 found it lowered blood sugar. Three decades ago, the US Food and Drug Administration (FDA) approved it for Type 2 diabetes.
But metformin’s effects on the body extend beyond blood sugar management. It works through multiple molecular pathways to control cell growth, metabolism, and inflammation—all of which go haywire during aging. This made scientists wonder: Can the drug slow aging?
Initial studies in several animal models of aging showed promise. Repeated doses of the drug reversed age-related tissue damage. In humans, epidemiological studies have found the drug reduces the risk of cancer and dementia. A 2014 study of 78,000 people showed that, on average, people with Type 2 diabetes on metformin lived longer than those of the same age who didn’t have diabetes and didn’t take the drug.
Despite its potential for slowing age-related disease, metformin hasn’t yet been directly tested in primates with that in mind.
Monkey BusinessThe new study filled that gap. The team gave metformin to crab-eating male macaque monkeys, which usually have a lifespan of roughly 25 to 30 years.
The treatment was simple: Beginning at around the equivalent of 52 to 64 years old in humans, some of the monkeys were fed a daily pill of metformin, similar in dosage to that used in diabetes management. Others didn’t receive the drug and aged naturally. For comparison, the team also included a young-adult and a middle-aged group.
All groups received comprehensive physical exams throughout the trial. These included a staggering 65 health measures, such as BMI, blood tests, and imaging of their bodies and brains.
The oldest groups, either with or without metformin, were also given a barrage of cognitive tests. Some checked their ability to remember things after a delay. Others tested how well they learned new information or could update previous knowledge—a measure of flexible thinking that erodes with age.
The team watched the monkeys’ health for over three years—or roughly the equivalent of 13 human years—while collecting samples of gene expression and protein data from multiple organs and tissues.
Comparing old monkeys treated with metformin to young, middle-aged, and untreated elderly ones, the team generated a “rescue score”—that is, how much metformin slowed aging. The brain, skin, liver, kidney, and lungs were the most restored according to the analysis.
The scientists validated the results by looking at tissues under the microscope. Senescent cells, often called “zombie cells,” decreased in number. These malfunctioning cells don’t naturally turn over. Instead, they spew a toxic molecular soup, damaging nearby tissues. Metformin also reduced the scarring that often occurs during aging, especially in the lungs, kidneys, and heart, and slashed chronic inflammation throughout the body. Inflammation is a “cardinal hallmark of aging that underlies almost all aging-related disorders,” wrote the team.
But the most striking effect was on the brain. Monkeys given a dose of metformin retained their learning and memory abilities as they aged. When challenged with multiple cognitive tests, they behaved as if they were six years younger—nearly two decades in human age—with far more mental prowess than similarly aged peers who hadn’t received the drug.
Parts of the brain gradually wither in size during aging. Metformin combated the shrinkage, especially in areas important for cognition and memory—for example, the front parts of the brain crucial for reasoning. The drug also revived neurons in the hippocampus, a brain region involved in memory, dampening inflammation and letting neurons regrow their branches. Gene expression in most types of brain cells reset to more youthful profiles.
From Mice to Monkeys to MenThe comprehensive study adds evidence for metformin’s potential anti-aging properties.
But it’s not without fault. The sample size is small. Although the study followed aging monkeys for over three years, only 12 received the drug. The results—especially the “monkey aging clock”—will need to be replicated in another population.
The study also only tested metformin in males. Aging females have different trajectories for multiple health measurements. In humans, women live longer than men on average, with delayed biological hallmarks of aging, although they have worse health at the end of life. An aging clock is incomplete until it also includes females.
The team wants to expand their study. One idea is to follow the monkeys longer to test if metformin increases lifespan. Another is to stop treatment to see if the anti-aging effects last.
For now, how metformin works in the aging body is still muddy. Research is underway to clarify its exact mechanisms. But compared to other potential longevity drugs—for example, those that kill off senescent “zombie cells”—metformin has a major advantage. It has already been used for decades in millions of people without major side effects.
Metformin has the FDA’s attention. In 2015, the agency approved TAME, for Targeting the Biology of Aging, a trial that aims to recruit 3,000 elderly people, some taking metformin and others not, and follow them for six years. The ambitious study is still seeking sufficient funding.
Meanwhile, the authors have launched a smaller, placebo-controlled clinical trial to see if the drug slows aging in middle-aged to elderly males. Like the insights gained from the monkey studies, the trial could guide strategies to slow age-related health problems.
The new study paves the way for “advancing pharmaceutical strategies against human aging,” wrote the team.
Image Credit: Billy Pasco / Unsplash
ARTIFICIAL INTELLIGENCEOpenAI Announces a New AI Model, Code-Named Strawberry, That Solves Difficult Problems Step by Step
Will Knight | Wired“The company today announced a new advance that signals a shift in approach—a model that can ‘reason’ logically through many difficult problems and is significantly smarter than existing AI without a major scale-up. The new model, dubbed OpenAI o1, can solve problems that stump existing AI models, including OpenAI’s most powerful existing model, GPT-4o.”
COMPUTINGGoogle Says It’s Made a Quantum Computing Breakthrough That Reduces Errors
Sophia Chen | MIT Technology Review“One major challenge has been that quantum computers can store or manipulate information incorrectly, preventing them from executing algorithms that are long enough to be useful. The new research from Google Quantum AI and its academic collaborators demonstrates that they can actually add components to reduce these errors. …’This error correction stuff really works, and I think it’s only going to get better,’ wrote Michael Newman, a member of the Google team, on X.”
AUTOMATIONDriverless Semis Could Be Months Away
Timothy B. Lee | Ars Technica“On a sunny morning in December, an 18-wheeler will pull into a truck depot in Palmer, Texas, just south of Dallas. The driver will step out of the cab and help transfer his trailer to a second rig outfitted with powerful sensors. This second truck will head south on Interstate 45 toward Houston. …Trucks travel the 200 miles between Dallas and Houston all the time. But there will be something special about the middle leg of this trip: There will be no one in the vehicle.”
ENERGYGeothermal Energy Could Outperform Nuclear Power
Editorial Staff | The Economist“How big could EGS [or enhanced geothermal systems] get? Big enough. Though DOE analyses suggest only around 40GW of conventional geothermal resource exist in America, new techniques expand the theoretical potential to a whopping 5,500GW across much of the country, with strong potential in over half of states. The heat is definitely on.”
BIOTECHCRISPR-Enhanced Viruses Are Being Deployed Against UTIs
Emily Mullin | Wired“The global rise in antibiotic resistance is making bacterial infections harder to treat and increasing the risk of disease spread, severe illness, and death. Once considered miracle drugs, antibiotics are now losing their effectiveness against ever-evolving bacteria. One company is aiming to treat infections with a different strategy: arming tiny viruses called bacteriophages with CRISPR.”
TECHOpenAI’s Fund-Raising Talks Could Value Company at $150 Billion
Cade Metz, Mike Isaac, Tripp Mickle, and Michael J. de la Merced | The New York Times“If the new deal is completed, OpenAI will be more valuable than SpaceX, the private rocket company founded by Elon Musk (who was also one of OpenAI’s co-founders). It would also be nearly twice as valuable as Intel, the venerable chip giant, whose total market value has slipped to around $83 billion as it has been unable to keep up with the AI boom.”
COMPUTINGTransistor-Like Qubits Hit Key Benchmark
Dina Genkina | IEEE Spectrum“A team in Australia has recently demonstrated a key advance in metal-oxide-semiconductor-based (or MOS-based) quantum computers. They showed that their two-qubit gates—logical operations that involve more than one quantum bit, or qubit—perform without errors 99 percent of the time. …What’s more, these MOS-based quantum computers are compatible with existing CMOS technology, which will make it more straightforward to manufacture a large number of qubits on a single chip than with other techniques.”
ROBOTICSInside Google’s 7-Year Mission to Give AI a Robot Body
Hans Peter Brondmo | Quote“As the head of Alphabet’s AI-powered robotics moonshot, I came to believe many things. For one, robots can’t come soon enough. For another, they shouldn’t look like us. …I am more convinced than ever that the robots need to come. Yet I have concerns that Silicon Valley, with its focus on ‘minimum viable products’ and VCs’ general aversion to investing in hardware, will be patient enough to win the global race to give AI a robot body. And much of the money that is being invested is focusing on the wrong things. Here is why.”
TECHThe AI Spending Spree, in Charts
Nate Rattner and Tom Dotan | The Wall Street Journal“Generative artificial intelligence has sparked one of the biggest spending booms in modern American history, as companies and investors bet hundreds of billions of dollars that the technology will revolutionize the global economy and one day lead to massive profits. The question is when, and even whether, all those investments will pay off.”
Image Credit: Laura Skinner / Unsplash
Cancer cells are tricky foes.
Our body’s immune system is normally on the lookout for signs of tumor cells. If any are detected, it launches killer T cells—a type of immune cell—to seek and destroy the threat. But it’s a cat-and-mouse game: As tumors grow, they form a protective barrier to counteract immune attacks. Immune cells lose their targeting and killing efficacy inside the protective zone.
One workaround is to genetically engineer more powerful T cells. A relatively new and promising approach called CAR T therapy adds more “targeting beacons” onto T cells extracted from each patient to convert them into tailored cancer torpedoes.
So far, six CAR T therapies have been approved by the FDA for various blood cancers. But they have an Achilles heel. Once inside the body, their numbers slowly dwindle, and they gradually lose their cancer-battling abilities.
Some scientists are working to make CAR T cells more deadly. Others are turning T cells into Trojan horses to infiltrate tumors. One such therapy was approved in May 2024, marking the first cellular therapy for a solid tumor—melanoma, an aggressive skin cancer.
Even with these upgrades and alternatives, a tumor’s protective shield is still difficult to penetrate. This month, a team from Asgard Therapeutics and Lund University took a clever new approach to tackle tumors from within. The work was
Using a technology called cellular reprogramming, the team transformed tumor cells in mice into a type of immune cell called cDC1 cells. These cells are master regulators of the immune system. They’re rare inside tumors but when present can trigger powerful immune responses that eat away at the cancer’s protective shield and recruit T cells to the target.
Mice treated with the gene therapy remained cancer-free for at least 100 days and resisted cancer resurgence in a lab test.
“The data provides preclinical proof-of-concept for an off-the-shelf, yet tumor-specific, first-in-class cancer immunotherapy,” wrote Asgard Therapeutics in a press release.
Identity ChangeAt the heart of the therapy is a technology called cellular reprogramming. Here, scientists use a combination of proteins called transcription factors to turn genes on or off. This process can change a cell’s identity.
The most famous example of cellular reprogramming is the Nobel Prize-winning creation of pluripotent cells (iPSCs). These cells have revolutionized regenerative medicine and how we study diseases. Here, four transcription factors convert mature skin cells back into stem cells. This type of cell can develop into any other type of cell in the body. Additional factors can then gently coax the newly minted iPSCs to assume new identities—for example, brain organoids (“mini-brains”), egg and sperm precursor cells, or liver and bone cells.
Soon after its introduction, the groundbreaking technology showed promise for gene therapy.
In 2008, a study found that delivering three transcription factors directly into the pancreases of diabetic mice turned them into insulin-releasing cells that kept the critters’ blood sugar levels in check. Another study, also in mice, converted heart cells that cause dangerous scarring after a heart attack into healthy heart muscle cells, leading to improved heart function. Scientists have also reprogrammed “supporting” brain cells in mice into functional neurons after brain injury or to treat neurodegenerative diseases.
But these cellular identity swaps all began with relatively normal cells. Tumor cells don’t work the same way—and their abnormalities could torpedo the process.
Tumor MakeoverThe new study builds on the team’s previous work reprogramming tumor cells in petri dishes. They aimed to convert these tumor cells into cDC1 cells because of their “manager” role coordinating immune responses.
First, they found three transcription factors that convert other cells into cDC1 cells. Next, they inserted genetic sequences of those factors into a virus stripped of its disease-causing properties. These viral carriers can deliver genes into cultured cells or the body.
As a proof of concept, the team grew melanoma cells in petri dishes, treated some with the gene therapy, and injected the engineered cells into healthy mice. Without the treatment, the melanoma cells rapidly expanded. Cells with the gene therapy, however, couldn’t grow as fast.
The average survival rate increased from 19 days without treatment to 43 days with it. Adding conventional immunotherapy drugs to the mix cleared all animals of the tumor cells.
The transformed cDC1 cells readily dismantled the tumor’s protective shield. After nine days, more immune cells swarmed the tumor, suggesting its protective barrier had begun to erode.
Classic immunotherapy drugs often exhaust T cells, limiting their expansion and ability to attack. Reprogramming lowered the chances of exhaustion in multiple types of T cells by as much as eight-fold.
Meanwhile, the treatment boosted the number of memory T cells—which, true to their name, retain a ledger of previous targets, including specific cancers. These cells guard the body against cancer resurgence. Once they detect previously defeated tumors, they alert other components of the immune system to strike before the cancer cells can regrow and spread.
Can It Work in the Human Body?Tumors in mice aren’t exactly the same as those in people. In another test, the team grew little balls of cells from multiple types of immortalized cancer cell lines in petri dishes. Some of these so-called “spheroids” included cells and other factors from a tumor’s protective shield.
Reprogramming the cancer cells into cDC1 cells decreased the size of the cancerous balls, although the efficiency differed between cancer types. Adding common drugs for cancer—which notoriously lower some immune responses—didn’t affect reprogramming and subsequent immune cell activation.
So far, good. But could the therapy work directly inside the body—without having to extract tumor cells and reprogram them in the lab. In a final test, the team injected the treatment into melanoma tumors in mice over the course of two weeks.
Half of those treated remained cancer-free for 100 days, with an abundance of T cells infiltrating the tumor area. The treated mice also readily fought off an experimental model of cancer relapse, holding malignant cells at bay for at least another 60 days—compared to control mice who developed cancers within a month.
There’s a long road before the treatment reaches clinics. But the team is already testing safety profiles, drug metabolism, and scaling up manufacturing processes to get ready for clinical trials.
Turning tumor cells against themselves “offers the advantages of a precision cell therapy, while overcoming the challenges” of genetically engineering immune cells outside the body, as happens in currently approved CAR T therapies, wrote the authors. That said, work that directly engineers CAR T cells inside the body is also on the rise.
Still, results here pave the way for human trials. They lay “the foundation for a new class of immunotherapies based on the unique function” of different types of immune cells, made inside the body using reprogramming, the authors concluded.
Image Credit: T cells (red) attack cancer cells (white) / Rita Elena Serda, Duncan Comprehensive Cancer Center at Baylor College of Medicine, National Cancer Institute, National Institutes of Health
Nuclear fusion has experienced something of a renaissance in recent years with a host of startups and governments seriously pursuing the idea. UK scientists have now provided a sneak peek of a novel reactor design that could be providing power to the grid by 2040.
Despite a reputation for being a technology that’s always 20 years away, recent years have seen a flurry of investment as optimism grows that its time may finally have come. According to the Fusion Industry Association, last year’s $900 million in new funding brought the total to $7.1 billion.
That optimism doesn’t seem to have been dampened by major delays to ITER, the international collaboration that has long been considered fusion’s flagship project. Building on the knowledge gleaned from ITER and other publicly funded experiments, a host of startups is now betting they can deliver smaller fusion reactors at a fraction of the time and cost.
But it’s not only the private sector pushing to commercialize the technology. In 2019, the UK government provided £300 million in funding for the design of a novel 200-megawatt reactor known as Spherical Tokamak for Energy Production (STEP). And in a series of papers recently published in the Philosophical Transactions of the Royal Society A, its designers have now given a glimpse of what they’ve come up with.
The most common design for a fusion reactor is known as a tokamak, which heats a cloud of ionized gas, known as plasma, until the atoms fuse together and generate huge amounts of energy in the process. The plasma is contained by incredibly strong magnetic fields generated by coils of magnets wrapped around a doughnut-shaped reactor vessel.
STEP follows similar principles but is tall and narrow, more like a cored apple, according to Science. While that might not seem like much of a difference, it means the distance between the center of the reactor vessel and the magnets wrapping around it is smaller than a classic tokamak.
This reduction in distance makes it possible to use smaller, less expensive magnets to contain the plasma and makes the entire design more compact, according to the Financial Times. The shape of a spherical tokamak also produces an inherently more stable plasma, which should improve performance. However, the design does have trade-offs.
Fusion reactors normally use two isotopes of hydrogen fuel called deuterium and tritium. Tritium is incredibly rare though, so reactors generate their own tritium by way of a reaction between the metal lithium and neutrons released by the fusion reaction. This lithium is stored in tritium breeding blankets wrapped around the chamber, which also act as radiation shields to protect the magnets.
The hole in the center of a tokamak normally houses large magnets and a breeding blanket. But with the narrower design of the spherical tokamak there is much less space, so the STEP reactor will have to do away with the blanket and significantly shrink the magnets, or even do away with some.
Fortunately, new high-temperature superconducting tape, which is also being used by many private startups, could help create more compact magnets. But the reactor will have to generate enough tritium using only the blankets on the outer wall of the chamber, which means the team had to come up with an optimized design using liquid lithium and a vanadium alloy.
The reactor’s designers have also opted for an ambitious architecture with joints in the magnets, which will make it possible to open the top of the vessel. This will significantly speed up maintenance jobs and therefore lower operational costs.
However, project leader Paul Methven, told Science that the recently published designs are still far from being set in stone. And while the project has already found itself a site—a retired coal-fired power station in Nottinghamshire county—the project is currently in discussions with the UK government to secure four more years of funding to come up with a final blueprint.
So, whether or not this reactor ever sees the light of day remains to be seen. But it is encouraging to see government investing significant sums to push the technology forward.
Image Credit: STEP
Automating food is unlike automating anything else. Food is fundamental to life—nourishing body and soul—so how it’s accessed, prepared, and consumed can change societies fundamentally.
Automated kitchens aren’t sci-fi visions from The Jetsons or Star Trek. The technology is real and global. Right now, robots are used to flip burgers, fry chicken, create pizzas, make sushi, prepare salads, serve ramen, bake bread, mix cocktails, and much more. AI can invent recipes based on the molecular compatibility of ingredients or whatever a kitchen has in stock. More advanced concepts are in the works to automate the entire kitchen for fine dining.
Since technology tends to be expensive at first, the early adopters of AI kitchen technologies are restaurants and other businesses. Over time, prices are likely to fall enough for the home market, possibly changing both home and societal dynamics.
Can food technology really change society? Yes, just consider the seismic impact of the microwave oven. With that technology, it was suddenly possible to make a quick meal for just one person, which can be a benefit but also a social disruptor.
Familiar concerns about the technology include worse nutrition and health from prepackaged meals and microwave-heated plastic containers. Less obviously, that convenience can also transform eating from a communal, cultural and creative event into a utilitarian act of survival—altering relationships, traditions, how people work, the art of cooking, and other facets of life for millions of people.
For instance, think about how different life might be without the microwave. Instead of working at your desk over a reheated lunch, you might have to venture out and talk to people, as well as enjoy a break from work. There’s something to be said for living more slowly in a society that’s increasingly frenetic and socially isolated.
Convenience can come at a great cost, so it’s vital to look ahead at the possible ethical and social disruptions that emerging technologies might bring, especially for a deeply human and cultural domain—food—that’s interwoven throughout daily life.
With funding from the US National Science Foundation, my team at California Polytechnic State University is halfway into what we believe is the first study of the effects AI kitchens and robot cooks could have on diverse societies and cultures worldwide. We’ve mapped out three broad areas of benefits and risks to examine.
Creators and ConsumersThe benefits of AI kitchens include enabling chefs to be more creative, as well as eliminating repetitive, tedious tasks such as peeling potatoes or standing at a workstation for hours. The technology can free up time. Not having to cook means being able to spend more time with family or focus on more urgent tasks. For personalized eating, AI can cater to countless special diets, allergies, and tastes on demand.
However, there are also risks to human well-being. Cooking can be therapeutic and provides opportunities for many things: gratitude, learning, creativity, communication, adventure, self-expression, growth, independence, confidence, and more, all of which may be lost if no one needs to cook. Family relationships could be affected if parents and children are no longer working alongside each other in the kitchen—a safe space to chat, in contrast to what can feel like an interrogation at the dining table.
The kitchen is also the science lab of the home, so science education could suffer. The alchemy of cooking involves teaching children and other learners about microbiology, physics, chemistry, materials science, math, cooking techniques and tools, food ingredients and their sourcing, human health, and problem-solving. Not having to cook can erode these skills and knowledge.
Community and CulturesAI can help with experimentation and creativity, such as creating elaborate food presentations and novel recipes within the spirit of a culture. Just as AI and robotics help generate new scientific knowledge, they can increase understanding of, say, the properties of food ingredients, their interactions, and cooking techniques, including new methods.
But there are risks to culture. For example, AI could bastardize traditional recipes and methods, since AI is prone to stereotyping, for example flattening or oversimplifying cultural details and distinctions. This selection bias could lead to reduced diversity in the kinds of cuisine produced by AI and robot cooks. Technology developers could become gatekeepers for food innovation, if the limits of their machines lead to homogeneity in cuisines and creativity, similar to the weirdly similar feel of AI art images across different apps.
Also, think about your favorite restaurants and favorite dinners. How might the character of those neighborhoods change with automated kitchens? Would it degrade your own gustatory experience if you knew those cooking for you weren’t your friends and family but instead were robots?
The hope with technology is that more jobs will be created than jobs lost. Even if there’s a net gain in jobs, the numbers hide the impact on real human lives. Many in the food service industry—one of the most popular occupations in any economy—could find themselves unable to learn new skills for a different job. Not everyone can be an AI developer or robot technician, and it’s far from clear that supervising a robot is a better job than cooking.
Philosophically, it’s still an open question whether AI is capable of genuine creativity, particularly if that implies inspiration and intuition. Assuming so may be the same mistake as thinking that a chatbot understands what it’s saying, instead of merely generating words that statistically follow the previous words. This has implications for aesthetics and authenticity in AI food, similar to ongoing debates about AI art and music.
Safety and ResponsibilityBecause humans are a key disease vector, robot cooks can improve food safety. Precision trimming and other automation can reduce food waste, along with AI recipes that can make the fullest use of ingredients. Customized meals can be a benefit for nutrition and health, for example, in helping people avoid allergens and excess salt and sugar.
The technology is still emerging, so it’s unclear whether those benefits will be realized. Foodborne illnesses are an unknown. Will AI and robots be able to smell, taste, or otherwise sense the freshness of an ingredient or the lack thereof and perform other safety checks?
Physical safety is another issue. It’s important to ensure that a robot chef doesn’t accidentally cut, burn, or crush someone because of a computer vision failure or other error. AI chatbots have been advising people to eat rocks, glue, gasoline, and poisonous mushrooms, so it’s not a stretch to think that AI recipes could be flawed, too. Where legal regimes are still struggling to sort out liability for autonomous vehicles, it may similarly be tricky to figure out liability for robot cooks, including if hacked.
Given the primacy of food, food technologies help shape society. The kitchen has a special place in homes, neighborhoods, and cultures, so disrupting that venerable institution requires careful thinking to optimize benefits and reduce risks.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Kindel Media / Pexels
As our energy mix shifts towards intermittent renewable sources, utility-scale batteries will be crucial for balancing power supply. The latest figures from the US Energy Information Agency (EIA) show batteries are being deployed at 10 times the rate of new gas power capacity.
The pace of the green transition has been remarkable in recent years, but without careful planning it could quickly grind to a halt. While solar and wind power are now competitive with or even cheaper than fossil fuel plants, they are less reliable because output depends on the sun shining and the wind blowing.
This caps how much renewable capacity we can add to the grid without causing serious stability issues. That is, unless we can find ways to store renewable energy in times of excess so it can help plug the gaps when generation drops off.
And it seems this is starting to happen at pace. According to new data from the EIA, installation of grid-scale batteries accounted for nearly a fifth of new energy capacity installed in the first half of this year, outpacing wind, nuclear, and gas.
Overall, 20 gigawatts of new capacity was added between January and June, and unsurprisingly, solar made up the lion’s share at 12 gigawatts. But batteries, which are counted as power generation because they can dispatch power to the grid, came in second at an impressive 4.2 gigawatts. That dwarfed the 0.4 gigawatts of natural gas power added to the grid in the same period, and pushed batteries above both wind at 2.5 gigawatts and nuclear at 1.1 gigawatts.
Tellingly, the data showed the new battery capacity was heavily concentrated in just four states: California, Texas, Arizona, and Nevada accounted for 93 percent of new installations. Ars Technica notes these states are also deploying large amounts of solar power, which means there’s an increasing need for storage to help meet demand after the sun has gone down.
As well as the data on new installations, the EIA also provided details on retirements of older power plants. While total capacity retired dropped from 9.2 gigawatts in the first half of 2023 to just 5.1 gigawatts in the first half of this year, the mix of retired plants gives a clear indication of the direction the grid is headed, with gas accounting for 53 percent and coal for 41 percent.
The jump in battery installations is a promising sign, as meeting the world’s emission reduction targets will require a massive increase in energy storage deployment. A report from the International Energy Agency earlier this year estimated there needs to be a sixfold increase in capacity globally by the end of the decade to meet 2030 targets set at the COP28 climate talks.
Things are looking promising. Deployments doubled last year, according to the report, and in less than 15 years, costs have fallen by more than 90 percent. Battery manufacturing capacity has also tripled over the past three years.
Meeting the target will still be a challenge though, the report notes. Countries agreed to triple renewable power capacity by 2030, but safely integrating this into grids around the world will require 1,200 gigawatts of new battery storage to be installed as well.
So far, most of this demand is being met by lithium-ion batteries, more commonly found in personal electronics and electric vehicles. But there’s also a host of emerging technologies that could help, including the likes of iron-air batteries.
Altogether, the latest data is a promising signal that grid-scale energy storage is going mainstream. But it will still take a concerted effort to ensure we have enough capacity to support a wholesale shift to renewable power.
Image Credit: ダモ リ / Unsplash
ARTIFICIAL INTELLIGENCEGenerative AI Creates Playable Version of Doom Game With No Code
Matthew Sparkes | New Scientist“An AI-generated re-creation of the classic computer game Doom can be played normally despite having no computer code or graphics. Researchers behind the project say similar AI models could be used to create games from scratch in the future, just as they create text and images today.”
ROBOTICSRobot Metalsmiths Are Resurrecting Toroidal Tanks for NASA
Evan Ackerman | IEEE Spectrum“Because of their relatively complex shape, toroidal tanks are much more difficult to make than spherical tanks. Even though these tanks can perform better, NASA simply doesn’t have the expertise to manufacture them anymore, since each one has to be hand-built by highly skilled humans. But a company called Machina Labs thinks that they can do this with robots instead. And their vision is to completely change how we make things out of metal.”
FUTUREThe Year Is 2149 and…
Sean Michaels | MIT Technology Review“Novelist Sean Michaels envisions what life will look like 125 years from now: ‘The year is 2149 and people mostly live their lives “on rails.” That’s what they call it, “on rails,” which is to live according to the meticulous instructions of software. Software knows most things about you—what causes you anxiety, what raises your endorphin levels, everything you’ve ever searched for, everywhere you’ve been. Software sends messages on your behalf; it listens in on conversations. ‘”
TECHOpenAI in Talks for Funding Round Valuing It Above $100 Billion
Tom Dotan and Berber Jin | The Wall Street Journal“The new funding round would be the biggest infusion of outside capital into OpenAI since Microsoft invested around $10 billion in January 2023. Since then, an arms race has developed in Silicon Valley to build the most advanced artificial-intelligence systems in an effort to dominate an industry many say will revolutionize the economy. OpenAI was last valued at $86 billion late last year, when employees sold existing shares.”
COMPUTINGAI Inference Competition Heats Up
Dina Genkina | IEEE Spectrum“While the dominance of Nvidia GPUs for AI training remains undisputed, we may be seeing early signs that, for AI inference, the competition is gaining on the tech giant, particularly in terms of power efficiency. The sheer performance of Nvidia’s new Blackwell chip, however, may be hard to beat.”
TRANSPORTATIONWorld’s Largest Sailing Cargo Ship Makes First Transatlantic Voyage
Jeremy Hsu | New Scientist“The world’s largest sailing cargo ship is making its maiden voyage across the Atlantic Ocean. It left a port in France in early August, and it is on track to deliver 1,000 tons of cognac and champagne to New York City by 3 September. Its shipments have a carbon footprint one tenth that of a standard container ship.”
COMPUTINGThought-to-Text Chip Smaller Than Neuralink Achieves 91% Accuracy
Michael Franco | New Atlas“The chip has been developed by researchers at [EPFL] and represents a leap forward in the sizzling space of brain-machine interfaces (BMIs)—devices that are able to read activity in the brain and translate it into real-world output such as text on a screen. That’s because this particular device—known as a miniaturized brain-machine interface (MiBMI)—is extremely small, consisting of two thin chips measuring just 8 mm2 total.”
SCIENCEWe’ll Soon Get the Sharpest Image Ever of a Black Hole
Isaac Schultz | Gizmodo“Tests by the Event Horizon Telescope Collaboration have yielded the highest-resolution observations ever obtained from Earth, laying an exciting foundation for future observations of black holes. …Typically, astronomers will get higher resolution images by using a bigger telescope, but the EHT already spans the Earth. Instead, the collaboration observed shorter wavelengths of light, yielding sharper images.”
TECHOpenAI Searches for an Answer to Its Copyright Problems
Elizabeth Lopatto | The Verge“The huge leaps in OpenAI’s GPT model probably came from sucking down the entire written web. That includes entire archives of major publishers such as Axel Springer, Condé Nast, and The Associated Press—without their permission. But for some reason, OpenAI has announced deals with many of these conglomerates anyway. At first glance, this doesn’t entirely make sense. Why would OpenAI pay for something it already had? And why would publishers, some of whom are lawsuit-style angry about their work being stolen, agree?”
DIGITAL MEDIAChatbots Are Primed to Warp Reality
Matteo Wong | The Atlantic“A sizable body of research, alongside conversations I’ve recently had with several experts, suggests that the solicitous, authoritative tone that AI models take—combined with them being legitimately helpful and correct in many cases—could lead people to place too much trust in the technology. That credulity, in turn, could make chatbots a particularly effective tool for anyone seeking to manipulate the public through the subtle spread of misleading or slanted information.”
Image Credit: Nicolas Houdayer / Unsplash
Alzheimer’s disease slowly takes over the mind. Long before symptoms occur, brain cells are gradually losing their function. Eventually they wither away, eroding brain networks that store memories. With time, this robs people of their recollections, reasoning, and identity.
It’s not the type of forgetfulness that happens during normal aging. In the twilight years, our ability to soak up new learning and rapidly recall memories also nosedives. While the symptoms seem similar, normally aging brains don’t exhibit the classic signs of Alzheimer’s—toxic protein buildups inside and surrounding neurons, eventually contributing to their deaths.
These differences can only be caught by autopsies, when it’s already too late to intervene. But they can still offer insights. Studies have built a profile of Alzheimer’s brains: Shrunken in size, with toxic protein clumps spread across regions involved in reasoning, learning, and memory.
However, those results only capture the very end of the journey.
This week, an international team led by Columbia University, MIT, and Harvard sought to map the entire process. Analyzing 437 donated brains from aging people—some with Alzheimer’s, others not—they peeked into the gene expression of 1.65 million brain cells in the regions most affected by Alzheimer’s and built a comprehensive cell atlas for aging brains.
A machine learning algorithm next teased apart the trajectories that differentiate Alzheimer’s from a normally aging brain. The team found a slew of genetic changes in multiple cell types that differed between the two. Some cell types controlled immunity; others supported metabolism.
“Our study highlights that Alzheimer’s is a disease of many cells and their interactions, not just a single type of dysfunctional cell,” said study author Dr. Philip De Jager in a press release.
With these results, “we provide a cellular foundation for a new perspective” on how Alzheimer’s develops, which could inform personalized treatments by targeting different brain cell communities, the authors wrote in the study.
“We may need to modify cellular communities to preserve cognitive function,” said Jager.
The Brainy BunchOur brains are a bit like a suburban community. Multiple types of neighboring cells help each other out.
Neurons are the best known. These spark with electricity and form the networks underlying our emotions, thoughts, and memories. But they don’t act alone.
Astrocytes—named for their star-like shape (pictured above)—nurture neurons with supportive molecules, especially when they need a metabolic boost. Meanwhile, microglia—the neighborhood watch committee—keep watch for signs of danger. A type of immune cell, these rapidly destroy bacteria, viruses, and other intruders. They’re also like “gardeners” for neurons, snipping away some connections to optimize neural networks as we learn.
In Alzheimer’s disease, this neighborliness breaks down. Microglia go rogue and increase inflammation. Astrocytes lose their function. Neurons wilt and die. The downward spiral happens over years, if not decades. By the time symptoms are obvious, it’s too late.
With over 400 brain samples, the new study aimed to find new treatments by charting the molecular journey of these brain cells.
Scientists have previously analyzed donated brains from people with and without Alzheimer’s. But they focused mostly on overall structure or zoomed in on molecular details. They didn’t chart the long journey of each individual cell’s role that, together, led to Alzheimer’s.
“Past studies have analyzed brain samples as a whole, and they lose all cellular detail,” said De Jager. “We now have tools to look at the brain in finer resolution, at the level of individual cells.”
Jager’s team aimed to find changes in multiple types of brain cells involved in the disease. They also used autopsies to reconstruct a chain of cause-and-effect: That is, finding the genes that translate brain cell changes into cognitive decline, and eventually, Alzheimer’s.
Brain BankThe study tapped into a long-running source for data. The Religious Orders Study and the Rush Memory and Aging Project (ROSMAP), which began in the 1990s, enrolled people 65 years of age and older and captured their health and mental status each year using standardized tests for up to two decades. The project also welcomed brain donations, yielding a valuable biobank.
Here, the team analyzed brain tissues from over 400 people—some with Alzheimer’s, others not. They used a popular method to gauge how individual cells work called single cell RNA sequencing. The technology has taken biology research by storm with its ability to map gene expression—that is, which genes are turned on—in individual cells.
It’s especially useful when studying the brain. Our noggins are incredibly complex, with many different cell types working together. The technology offers a way to peek into the genetic workings of each type and decipher how they all fit together in a functional “neighborhood.”
By looking at individual neurons and cognition test results from the donors, “we can reconstruct trajectories of brain aging from the earliest stages of the disease,” said De Jager.
The brain samples spanned brain aging and Alzheimer’s—roughly 60 percent showed signs of the disease—and the team captured the genetic readouts of 1.6 million brain cells of all types.
Microglia, the brain’s immune cells, were shuffled into 16 different populations based on their sequencing results, with some previously linked to Alzheimer’s in a mouse model. Astrocytes, the brain’s supportive cells, also showed 10 distinct gene expression types.
The team also documented different neurons, blood vessel cells that feed the brain, and other supporting cells that help maintain the brain’s overall structure.
Algorithm to Alzheimer’sTo make sense of the data, the team developed an algorithm to link different subpopulations of cells to the disease. They focused on three main problems related to Alzheimer’s. The first two are the presence of toxic protein clumps inside and outside of neurons. The third is the rate of cognitive decline before death.
With a custom-designed algorithm called BEYOND, the team sifted through the database and found two trajectories for aging brains. One aged normally, while the other showed signs of Alzheimer’s, with increased toxic protein buildup and cognitive decline. No single brain cell type, by itself, was the villain—rather, the whole community spiraled out of control.
During the disease’s early stages, a subset of microglia ramped up. These cells increased inflammation and accumulated toxic proteins.
“We propose that two different types of microglial cells—the immune cells of the brain—begin the process of amyloid and tau accumulation that define Alzheimer’s disease,” said De Jager.
The cells then triggered an Alzheimer’s cascade. A subset of astrocytes—the brain’s supporting cells—were the first victim, as they frantically tried to increase the activity of protective genes. Based on the analysis, astrocytes may be key to differentiating Alzheimer’s and aging.
The algorithm predicted these types of cells may be a “point of convergence” for processes that lead to dementia, as opposed to normal brain aging. Knowing how individual cells contribute to Alzheimer’s—and their journey into the disease—makes it possible to target specific cellular communities with new therapies to tackle both problems.
“These are exciting new insights that can guide innovative therapeutic development for Alzheimer’s and brain aging,” said De Jager.
Image Credit: Kevin Richetin / University of Lausanne via Flickr
Recent progress in AI largely boils down to one thing: Scale.
Around the beginning of this decade, AI labs noticed that making their algorithms—or models—ever bigger and feeding them more data consistently led to enormous improvements in what they could do and how well they did it. The latest crop of AI models have hundreds of billions to over a trillion internal network connections and learn to write or code like we do by consuming a healthy fraction of the internet.
It takes more computing power to train bigger algorithms. So, to get to this point, the computing dedicated to AI training has been quadrupling every year, according to nonprofit AI research organization, Epoch AI.
Should that growth continue through 2030, future AI models would be trained with 10,000 times more compute than today’s state of the art algorithms, like OpenAI’s GPT-4.
“If pursued, we might see by the end of the decade advances in AI as drastic as the difference between the rudimentary text generation of GPT-2 in 2019 and the sophisticated problem-solving abilities of GPT-4 in 2023,” Epoch wrote in a recent research report detailing how likely it is this scenario is possible.
But modern AI already sucks in a significant amount of power, tens of thousands of advanced chips, and trillions of online examples. Meanwhile, the industry has endured chip shortages, and studies suggest it may run out of quality training data. Assuming companies continue to invest in AI scaling: Is growth at this rate even technically possible?
In its report, Epoch looked at four of the biggest constraints to AI scaling: Power, chips, data, and latency. TLDR: Maintaining growth is technically possible, but not certain. Here’s why.
Power: We’ll Need a LotPower is the biggest constraint to AI scaling. Warehouses packed with advanced chips and the gear to make them run—or data centers—are power hogs. Meta’s latest frontier model was trained on 16,000 of Nvidia’s most powerful chips drawing 27 megawatts of electricity.
This, according to Epoch, is equal to the annual power consumption of 23,000 US households. But even with efficiency gains, training a frontier AI model in 2030 would need 200 times more power, or roughly 6 gigawatts. That’s 30 percent of the power consumed by all data centers today.
There are few power plants that can muster that much, and most are likely under long-term contract. But that’s assuming one power station would electrify a data center. Epoch suggests companies will seek out areas where they can draw from multiple power plants via the local grid. Accounting for planned utilities growth, going this route is tight but possible.
To better break the bottleneck, companies may instead distribute training between several data centers. Here, they would split batches of training data between a number of geographically separate data centers, lessening the power requirements of any one. The strategy would require lightning-quick, high-bandwidth fiber connections. But it’s technically doable, and Google Gemini Ultra’s training run is an early example.
All told, Epoch suggests a range of possibilities from 1 gigawatt (local power sources) all the way up to 45 gigawatts (distributed power sources). The more power companies tap, the larger the models they can train. Given power constraints, a model could be trained using about 10,000 times more computing power than GPT-4.
Credit: Epoch AI, CC BY 4.0Chips: Does It Compute?All that power is used to run AI chips. Some of these serve up completed AI models to customers; some train the next crop of models. Epoch took a close look at the latter.
AI labs train new models using graphics processing units, or GPUs, and Nvidia is top dog in GPUs. TSMC manufactures these chips and sandwiches them together with high-bandwidth memory. Forecasting has to take all three steps into account. According to Epoch, there’s likely spare capacity in GPU production, but memory and packaging may hold things back.
Given projected industry growth in production capacity, they think between 20 and 400 million AI chips may be available for AI training in 2030. Some of these will be serving up existing models, and AI labs will only be able to buy a fraction of the whole.
The wide range is indicative of a good amount of uncertainty in the model. But given expected chip capacity, they believe a model could be trained on some 50,000 times more computing power than GPT-4.
Credit: Epoch AI, CC BY 4.0Data: AI’s Online EducationAI’s hunger for data and its impending scarcity is a well-known constraint. Some forecast the stream of high-quality, publicly available data will run out by 2026. But Epoch doesn’t think data scarcity will curtail the growth of models through at least 2030.
At today’s growth rate, they write, AI labs will run out of quality text data in five years. Copyright lawsuits may also impact supply. Epoch believes this adds uncertainty to their model. But even if courts decide in favor of copyright holders, complexity in enforcement and licensing deals like those pursued by Vox Media, Time, The Atlantic and others mean the impact on supply will be limited (though the quality of sources may suffer).
But crucially, models now consume more than just text in training. Google’s Gemini was trained on image, audio, and video data, for example.
Non-text data can add to the supply of text data by way of captions and transcripts. It can also expand a model’s abilities, like recognizing the foods in an image of your refrigerator and suggesting dinner. It may even, more speculatively, result in transfer learning, where models trained on multiple data types outperform those trained on just one.
There’s also evidence, Epoch says, that synthetic data could further grow the data haul, though by how much is unclear. DeepMind has long used synthetic data in its reinforcement learning algorithms, and Meta employed some synthetic data to train its latest AI models. But there may be hard limits to how much can be used without degrading model quality. And it would also take even more—costly—computing power to generate.
All told, though, including text, non-text, and synthetic data, Epoch estimates there’ll be enough to train AI models with 80,000 times more computing power than GPT-4.
Credit: Epoch AI, CC BY 4.0Latency: Bigger Is SlowerThe last constraint is related to the sheer size of upcoming algorithms. The bigger the algorithm, the longer it takes for data to traverse its network of artificial neurons. This could mean the time it takes to train new algorithms becomes impractical.
This bit gets technical. In short, Epoch takes a look at the potential size of future models, the size of the batches of training data processed in parallel, and the time it takes for that data to be processed within and between servers in an AI data center. This yields an estimate of how long it would take to train a model of a certain size.
The main takeaway: Training AI models with today’s setup will hit a ceiling eventually—but not for awhile. Epoch estimates that, under current practices, we could train AI models with upwards of 1,000,000 times more computing power than GPT-4.
Credit: Epoch AI, CC BY 4.0Scaling Up 10,000xYou’ll have noticed the scale of possible AI models gets larger under each constraint—that is, the ceiling is higher for chips than power, for data than chips, and so on. But if we consider all of them together, models will only be possible up to the first bottleneck encountered—and in this case, that’s power. Even so, significant scaling is technically possible.
“When considered together, [these AI bottlenecks] imply that training runs of up to 2e29 FLOP would be feasible by the end of the decade,” Epoch writes.
“This would represent a roughly 10,000-fold scale-up relative to current models, and it would mean that the historical trend of scaling could continue uninterrupted until 2030.”
Credit: Epoch AI, CC BY 4.0What Have You Done for Me Lately?While all this suggests continued scaling is technically possible, it also makes a basic assumption: That AI investment will grow as needed to fund scaling and that scaling will continue to yield impressive—and more importantly, useful—advances.
For now, there’s every indication tech companies will keep investing historic amounts of cash. Driven by AI, spending on the likes of new equipment and real estate has already jumped to levels not seen in years.
“When you go through a curve like this, the risk of underinvesting is dramatically greater than the risk of overinvesting,” Alphabet CEO Sundar Pichai said on last quarter’s earnings call as justification.
But spending will need to grow even more. Anthropic CEO Dario Amodei estimates models trained today can cost up to $1 billion, next year’s models may near $10 billion, and costs per model could hit $100 billion in the years thereafter. That’s a dizzying number, but it’s a price tag companies may be willing to pay. Microsoft is already reportedly committing that much to its Stargate AI supercomputer, a joint project with OpenAI due out in 2028.
It goes without saying that the appetite to invest tens or hundreds of billions of dollars—more than the GDP of many countries and a significant fraction of current annual revenues of tech’s biggest players—isn’t guaranteed. As the shine wears off, whether AI growth is sustained may come down to a question of, “What have you done for me lately?”
Already, investors are checking the bottom line. Today, the amount invested dwarfs the amount returned. To justify greater spending, businesses will have to show proof that scaling continues to produce more and more capable AI models. That means there’s increasing pressure on upcoming models to go beyond incremental improvements. If gains tail off or enough people aren’t willing to pay for AI products, the story may change.
Also, some critics believe large language and multimodal models will prove to be a pricy dead end. And there’s always the chance a breakthrough, like the one that kicked off this round, shows we can accomplish more with less. Our brains learn continuously on a light bulb’s worth of energy and nowhere near an internet’s worth of data.
That said, if the current approach “can automate a substantial portion of economic tasks,” the financial return could number in the trillions of dollars, more than justifying the spend, according to Epoch. Many in the industry are willing to take that bet. No one knows how it’ll shake out yet.
Image Credit: Werclive / Unsplash
Artificial intelligence prophets and newsmongers are forecasting the end of the generative AI hype, with talk of an impending catastrophic “model collapse.”
But how realistic are these predictions? And what is model collapse anyway?
Discussed in 2023, but popularized more recently, “model collapse” refers to a hypothetical scenario where future AI systems get progressively dumber due to the increase of AI-generated data on the internet.
The Need for DataModern AI systems are built using machine learning. Programmers set up the underlying mathematical structure, but the actual “intelligence” comes from training the system to mimic patterns in data.
But not just any data. The current crop of generative AI systems needs high quality data, and lots of it.
To source this data, big tech companies such as OpenAI, Google, Meta, and Nvidia continually scour the internet, scooping up terabytes of content to feed the machines. But since the advent of widely available and useful generative AI systems in 2022, people are increasingly uploading and sharing content that is made, in part or whole, by AI.
In 2023, researchers started wondering if they could get away with only relying on AI-created data for training, instead of human-generated data.
There are huge incentives to make this work. In addition to proliferating on the internet, AI-made content is much cheaper than human data to source. It also isn’t ethically and legally questionable to collect en masse.
However, researchers found that without high-quality human data, AI systems trained on AI-made data get dumber and dumber as each model learns from the previous one. It’s like a digital version of the problem of inbreeding.
This “regurgitive training” seems to lead to a reduction in the quality and diversity of model behavior. Quality here roughly means some combination of being helpful, harmless, and honest. Diversity refers to the variation in responses and which people’s cultural and social perspectives are represented in the AI outputs.
In short, by using AI systems so much, we could be polluting the very data source we need to make them useful in the first place.
Avoiding CollapseCan’t big tech just filter out AI-generated content? Not really. Tech companies already spend a lot of time and money cleaning and filtering the data they scrape, with one industry insider recently sharing they sometimes discard as much as 90 percent of the data they initially collect to train models.
These efforts might get more demanding as the need to specifically remove AI-generated content increases. But more importantly, in the long term it will actually get harder and harder to distinguish AI content. This will make the filtering and removal of synthetic data a game of diminishing (financial) returns.
Ultimately, the research so far shows we just can’t completely do away with human data. After all, it’s where the “I” in AI is coming from.
Are We Headed for a Catastrophe?There are hints developers are already having to work harder to source high-quality data. For instance, the documentation accompanying the GPT-4 release credited an unprecedented number of staff involved in the data-related parts of the project.
We may also be running out of new human data. Some estimates say the pool of human-generated text data might be tapped out as soon as 2026.
It’s likely why OpenAI and others are racing to shore up exclusive partnerships with industry behemoths such as Shutterstock, Associated Press, and NewsCorp. They own large proprietary collections of human data that aren’t readily available on the public internet.
However, the prospects of catastrophic model collapse might be overstated. Most research so far looks at cases where synthetic data replaces human data. In practice, human and AI data are likely to accumulate in parallel, which reduces the likelihood of collapse.
The most likely future scenario will also see an ecosystem of somewhat diverse generative AI platforms being used to create and publish content, rather than one monolithic model. This also increases robustness against collapse.
It’s a good reason for regulators to promote healthy competition by limiting monopolies in the AI sector, and to fund public interest technology development.
The Real ConcernsThere are also more subtle risks from too much AI-made content.
A flood of synthetic content might not pose an existential threat to the progress of AI development, but it does threaten the digital public good of the (human) internet.
For instance, researchers found a 16 percent drop in activity on the coding website StackOverflow one year after the release of ChatGPT. This suggests AI assistance may already be reducing person-to-person interactions in some online communities.
Hyperproduction from AI-powered content farms is also making it harder to find content that isn’t clickbait stuffed with advertisements.
It’s becoming impossible to reliably distinguish between human-generated and AI-generated content. One method to remedy this would be watermarking or labeling AI-generated content, as I and many others have recently highlighted, and as reflected in recent Australian government interim legislation.
There’s another risk, too. As AI-generated content becomes systematically homogeneous, we risk losing socio-cultural diversity and some groups of people could even experience cultural erasure. We urgently need cross-disciplinary research on the social and cultural challenges posed by AI systems.
Human interactions and human data are important, and we should protect them. For our own sakes, and maybe also for the sake of the possible risk of a future model collapse.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Google DeepMind / Unsplash
On Aug 27, all eyes will be on NASA’s Kennedy Space Center in Florida for a historic flight.
SpaceX’s Falcon 9 rocket is set to propel the Dragon crew capsule and four private astronauts into space. The Polaris Dawn mission will fly to the highest altitude yet recorded in commercial spaceflight. It’ll also be the first to traverse belts of dangerous radiation surrounding Earth and attempt a spacewalk by private citizens, rather than highly trained astronauts.
Meanwhile, the crew will monitor their health before, during, and after the flight—from eye and bone health to cognition. This will help further our understanding of how just a few days of spaceflight transforms our biology—for example, which genes are turned on or off, how immunity changes, and why well-known challenges such as eye problems and loss of bone density emerge even with a short stay in space.
This information will go into an open-source biobank and help scientists collaborate on treatments for short-term flights and even longer jaunts to the moon, Mars, and beyond.
The launch is the first of three planned Polaris missions, which aim to advance technologies and healthcare that could one day propel us deeper into space. Here’s what you need to know.
Pushing BoundariesHeading the mission is Jared Isaacman, who is no stranger to space travel.
In 2021, he funded Inspiration 4, the first all-civilian mission to orbit the Earth. The mission showed that the average person is capable of spaceflight with a short bout of training and brought a wealth of insights into how a brief stint in space changes the body.
Accompanying Isaacman are mission pilot Scott “Kidd” Poteet, a former US Air Force Lieutenant Colonel, and two SpaceX employees. Thirty-year-old operations engineer Sarah Gillis is the youngest of the team and will join Isaacman on the spacewalk. Anna Menon, a mission specialist and medical officer, previously worked at NASA for seven years coordinating medical care from mission control.
The team will spend five days inside the Dragon capsule as it travels as high as 870 miles—the furthest from Earth humans have been since NASA’s Apollo program.
Their trajectory will take them through one of two deadly “circles” of high radiation called the Van Allen radiation belts, where highly charged particles from the sun and other sources are captured by Earth’s magnetic field. These regions are especially risky, as the particles can potentially tear through a space capsule and penetrate the body. To expand into the cosmos, we need to learn how to protect astronauts from such radiation.
Medicine in SpacePolaris Dawn partnered with 31 institutions to probe the health effects of spaceflight. Professional astronauts have been conditioned for spaceflight for years—the civilian crew offers a rare chance to examine the impact of microgravity on the health of an average space traveler.
Many of the studies are collaborations between NASA’s Human Research Program and the Translational Research Institute for Space Health (TRISH). Led by Baylor College of Medicine, the California Institute of Technology, and MIT, TRISH is a scientific consortium investigating how we can keep astronauts safe and healthy during deep space missions.
Spaceflight changes the body. Spacewalks could bring on additional changes. One project, building on Inspiration 4, will collect biological samples from the crew—like an annual health checkup—before, during, and after the flight. These samples will then be processed and added to the Space Omics and Medical Atlas, which includes the crew’s genetic makeup and gene expression changes—which genes are turned on or off—after a sprint into the radiation belts.
Other studies will delve into the effects of radiation and microgravity.
One team from TRISH will analyze how radiation impacts different bodily tissues during the mission and check to see whether any changes linger or return to normal back on Earth. Previous studies have mostly researched astronauts living for months on the International Space Station, which is closer to our home planet. Polaris Dawn’s crew will experience much more radiation at higher altitudes. This data could provide help us reduce radiation risk in the future.
Another team will test a hand-held ultrasound tool called Butterfly IQ+. It’s not fully automated, like the AI medical pods in the science fiction movie Prometheus, but the idea is similar: Being able to diagnose and treat unexpected medical troubles on the fly is crucial for space travel. The crew will test the device in space for myriad potential uses, like, for example, collecting medical-grade images of bladder function or blood and bodily fluid status.
The tool will be especially useful for spacewalks. Unlike the International Space Station, Dragon does not have an airlock. When Isaacman and Gillis go on their spacewalk, the entire capsule will open to the vacuum of space. The sudden change in pressure can cause potentially life-threatening conditions, known as decompression sickness or “the bends.” Scuba divers experience this condition when they ascend too rapidly and nitrogen forms gas bubbles in the bloodstream. A diagnostic tool could capture these dangerous conditions.
Another set of studies will focus on bone density and fluids. Working with TRISH, the University of Calgary is using a high-resolution device to scan the bone structure of the crew’s wrists and ankles—which are indicators of potential bone loss. If they detect a change, it will be the earliest ever to capture spaceflight’s effect on bone health. Meanwhile, a Dartmouth study is monitoring whether a first morning urine sample can predict bone and muscle health.
Microgravity also makes the effects of medicine—say, an Advil—unpredictable. Our bodily fluids, gut function, and metabolism all go topsy-turvy in space, which impacts how common medications work. The Polaris Dawn crew will test several common medications and chart how they behave in space.
Meanwhile, the team will also challenge their minds with a battery of cognitive tests. Developed by NASA and others, the tests include ten different tasks—kind of like Wordle or other games—to be completed on a tablet. But these specifically measure brain functions relevant to spaceflight. Other tests ask how much each crew member is willing to tolerate risk when making decisions, if they’re able to focus, and whether they can healthily process emotions.
There’s no doubt the mission is risky. On their spacewalk, Isaacman and Gillis will be testing SpaceX’s newly designed extravehicular activity suit, which doesn’t include life support. Instead, the two will receive all oxygen and other support from umbilical hoses attached to Dragon.
Still, the mission will hopefully strengthen our ability to adapt, live, and work in space.
Image Credit: Polaris Program
ARTIFICIAL INTELLIGENCEAn ‘AI Scientist’ Is Inventing and Running Its Own Experiments
Will Knight | Wired“At first glance, a recent batch of research papers produced by a prominent artificial intelligence lab at the University of British Columbia in Vancouver might not seem that notable. Featuring incremental improvements on existing algorithms and ideas, they read like the contents of a middling AI conference or journal. But the research is, in fact, remarkable. That’s because it’s entirely the work of an ‘AI scientist’ developed at the UBC lab together with researchers from the University of Oxford and a startup called Sakana AI.”
BIOTECHBeyond Gene-Edited Babies: The Possible Paths for Tinkering With Human Evolution
Antonio Regalado | MIT Technology Review“Editing human embryos is restricted in much of the world—and making an edited baby is flatly illegal in most countries surveyed by legal scholars. But advancing technology could render the embryo issue moot. New ways of adding CRISPR to the bodies of people already born—children and adults—could let them easily receive changes as well.”
ROBOTICSBoston Dynamics’ New Electric Atlas Can Do Push-Ups
Brian Heater | TechCrunch“Until today, we’ve seen exactly 40 seconds of Boston Dynamics’ new electric Atlas in action. The Hyundai-owned robotics stalwart is very much still in the early stages of commercializing the biped for factory floors. For now, however, it’s doing the thing Boston Dynamics does second best after building robots: showing off in viral video form.”
BIOTECHThe Next Frontier for mRNA Could Be Healing Damaged Organs
Emily Mullin | Wired“Faccioli and Hu are part of a University of Pittsburgh team led by Alejandro Soto-Gutiérrez attempting to revive badly damaged livers like these—as well as kidneys, hearts, and lungs. Using messenger RNA, the same technology used in some of the Covid-19 vaccines, they’re aiming to reprogram terminally ill organs to be fit and functioning again.”
GOVERNANCESilicon Valley Is Coming Out in Force Against an AI-Safety Bill
Caroline Mimbs Nyce | The Atlantic“In part, the debate over the bill gets at a core question with AI. Will this technology end the world, or have people just been watching too much sci-fi? At the center of it all is [Scott] Wiener. Because so many AI companies are based in California, the bill, if passed, could have major implications nationwide. I caught up with the state senator yesterday to discuss what he describes as his ‘hardball politics’ of this bill—and whether he actually believes that AI is capable of going rogue and firing off nuclear weapons.”
TRANSPORTATIONWaymo Wants to Chauffeur Your Kids
Kyle Wiggers | TechCrunch“Soon, parents in range of Waymo robotaxis might not have to worry about picking up their kids from after-school activities—or any time, really. The San Francisco Standard reports that Waymo, the Alphabet subsidiary, is considering a subscription program that would let teens hail one of its cars solo and send pickup and drop-off alerts to their parents.”
COMPUTINGDNA Computer Can Play Chess and Solve Sudoku Puzzles
Alex Wilkins | New Scientist“Computers made from DNA have previously only been able to store information or perform computations on it—now a new device can do both. A computer made from DNA that can solve basic chess and sudoku puzzles could one day, if scaled up, save vast amounts of energy over traditional computers when it comes to tasks like training artificial intelligence models.”
AUTOMATIONBoulder-Like 3D-Printed Homes Will Feature Up to Three Floors
Adam Williams | New Atlas“Most 3D-printed homes are currently arranged on one floor, which can obviously be a little limiting. However, an upcoming project in the Netherlands shows that this might not be the case for long as it will build new houses that will include up to three floors, showcasing the increasing complexity of 3D-printed architecture.”
ENERGYWorld’s ‘Largest Solar Precinct’ Approved by Australian Government
Keiran Smith | Associated Press“Australian company Sun Cable plans to build a 12,400-hectare solar farm and transport electricity to the northern Australian city of Darwin via an 800-kilometer (497-mile) overhead transmission line, then on to large-scale industrial customers in Singapore through a 4,300-kilometer (2,672-mile) submarine cable. The Australia-Asia PowerLink project aims to deliver up to six gigawatts of green electricity each year.”
DIGITAL MEDIANo One’s Ready for This
Sarah Jeong | The Verge“Our trust in photography was so deep that when we spent time discussing veracity in images, it was more important to belabor the point that it was possible for photographs to be fake, sometimes. This is all about to flip—the default assumption about a photo is about to become that it’s faked, because creating realistic and believable fake photos is now trivial to do. We are not prepared for what happens after.”
SPACEAgainst All Odds, an Asteroid Mining Company Appears to Be Making Headway
Eric Berger | Ars Technica“[AstroForge’s Odin mission] will be a rideshare payload on the Intuitive Machines-2 mission, which is due to launch during the fourth quarter of this year. If successful, the Odin mission would be spectacular. About seven months after launching, Odin will attempt to fly by a near-Earth, metallic-rich asteroid while capturing images and taking data—truly visiting terra incognita. Odin would also be the first private mission to fly by a body in the solar system beyond the moon.”
Image Credit: Eren Yıldız / Unsplash
The rapid transition to renewable energy is great news for the environment, but electrical grids are struggling to incorporate intermittent power sources like wind and solar. Startup Form Energy is about to demonstrate a potential solution with the construction of the world’s largest battery.
While gas- and coal-powered plants can run night and day, renewable power is highly reliant on the sun shining and the wind blowing. Finding ways to deal with this inherent uncertainty will be crucial if we ever want to fully decarbonize our grids.
One of the most obvious solutions is to store extra energy when conditions are favorable, so it can be fed back into the grid later when renewable production drops off. But building batteries capable of storing grid-scale quantities of electricity presents both engineering and economic challenges.
So far, most grid-storage facilities rely on lithium-ion batteries—the same technology found in cellphones and electric vehicles. But these batteries are relatively expensive and not particularly long-lived. They’re also prone to setting fire, which makes them less than ideal for such projects.
Form Energy is betting that its novel iron-air chemistry, which is specially designed for long-term energy storage, could be the answer. And it’s now set to receive $147 million to build a facility in Maine capable of storing enough energy to provide 85 megawatts of power for up to 100 hours.
“Located at the site of a former paper mill in rural Maine, this iron-air battery system will have the most energy capacity of any battery system announced yet in the world,” Mateo Jaramillo, CEO and cofounder of Form Energy, said in a press release.
The project, which is due to be complete by 2028, is part of a broader package of funding from the Bipartisan Infrastructure Law to upgrade the power grid in the Northeast of the US. Most of the $389 million will be used to expand and upgrade the region’s ability to accept power from large offshore wind farms. But the remainder will be given to Form to build a facility capable of storing 8,500 megawatt-hours of energy.
The key ingredients of the company’s battery cells are iron and water, and they rely on the same process that causes rust to charge and discharge. Energy is stored in the battery by converting iron oxide into pure iron and emitting the oxygen into the atmosphere. To release that energy, the battery absorbs oxygen from ambient air to turn the iron back into iron oxide.
This approach can’t get anywhere close to the energy density of lithium-ion batteries—a crucial consideration when packing batteries into small devices or trying to boost the range of vehicles. But when building large-scale storage systems energy density is much less of a concern than cost, a metric on which iron-air batteries win hands down.
Form says this will allow it to build storage facilities that can act as substitutes for power plants for extended periods. As well as helping to balance the grid during everyday operations, this could also help provide emergency power during extreme weather or other grid outages.
While the Maine project is the most ambitious that Form has announced to date, the company has already announced other smaller pilot projects, and Jaramillo told Canary Media that it’s working on other grid-scale facilities of a similar size that have yet to be publicized.
Given the company is still building the factory that will supply batteries to all these projects, it’s early days for the approach. If all goes according to plan though, it could prove to be a crucial tool in efforts to decarbonize the grid.
Image Credit: Form Energy
Our brains are constantly learning. That new sandwich deli rocks. That gas station? Better avoid it in the future.
Memories like these physically rewire connections in the brain region that supports new learning. During sleep, the previous day’s memories are shuttled to other parts of the brain for long-term storage, freeing up brain cells for new experiences the next day. In other words, the brain can continuously soak up our everyday lives without losing access to memories of what came before.
AI, not so much. GPT-4 and other large language and multimodal models, which have taken the world by storm, are built using deep learning, a family of algorithms that loosely mimic the brain. The problem? “Deep learning systems with standard algorithms slowly lose the ability to learn,” Dr. Shibhansh Dohare at University of Alberta recently told Nature.
The reason for this is in how they’re set up and trained. Deep learning relies on multiple networks of artificial neurons that are connected to each other. Feeding data into the algorithms—say, reams of online resources like blogs, news articles, and YouTube and Reddit comments—changes the strength of these connections, so that the AI eventually “learns” patterns in the data and uses these patterns to churn out eloquent responses.
But these systems are basically brains frozen in time. Tackling a new task sometimes requires a whole new round of training and learning, which erases what came before and costs millions of dollars. For ChatGPT and other AI tools, this means they become increasingly outdated over time.
This week, Dohare and colleagues found a way to solve the problem. The key is to selectively reset some artificial neurons after a task, but without substantially changing the entire network—a bit like what happens in the brain as we sleep.
When tested with a continual visual learning task—say differentiating cats from houses or telling apart stop signs and school buses—deep learning algorithms equipped with selective resetting easily maintained high accuracy over 5,000 different tasks. Standard algorithms, in contrast, rapidly deteriorated, their success eventually dropping to about a coin-toss.
Called continual back propagation, the strategy is “among the first of a large and fast-growing set of methods” to deal with the continuous learning problem, wrote Drs. Clare Lyle and Razvan Pascanu at Google DeepMind, who were not involved in the study.
Machine MindDeep learning is one of the most popular ways to train AI. Inspired by the brain, these algorithms have layers of artificial neurons that connect to form artificial neural networks.
As an algorithm learns, some connections strengthen, while others dwindle. This process, called plasticity, mimics how the brain learns and optimizes artificial neural networks so they can deliver the best answer to a problem.
But deep learning algorithms aren’t as flexible as the brain. Once trained, their weights are stuck. Learning a new task reconfigures weights in existing networks—and in the process, the AI “forgets” previous experiences. It’s usually not a problem for typical uses like recognizing images or processing language (with the caveat that they can’t adapt to new data on the fly). But it’s highly problematic when training and using more sophisticated algorithms—for example, those that learn and respond to their environments like humans.
Using a classic gaming example, “a neural network can be trained to obtain a perfect score on the video game Pong, but training the same network to then play Space Invaders will cause its performance on Pong to drop considerably,” wrote Lyle and Pascanu.
Aptly called catastrophic forgetting, computer scientists have been battling the problem for years. An easy solution is to wipe the slate clean and retrain an AI for a new task from scratch, using a combination of old and new data. Although it recovers the AI’s abilities, the nuclear option also erases all previous knowledge. And while the strategy is doable for smaller AI models, it isn’t practical for huge ones, such as those that power large language models.
Back It UpThe new study adds to a foundational mechanism of deep learning, a process called back propagation. Simply put, back propagation provides feedback to the artificial neural network. Depending on how close the output is to the right answer, back propagation tweaks the algorithm’s internal connections until it learns the task at hand. With continuous learning, however, neural networks rapidly lose their plasticity, and they can no longer learn.
Here, the team took a first step toward solving the problem using a 1959 theory with the impressive name of “Selfridge’s Pandemonium.” The theory captures how we continuously process visual information and has heavily influenced AI for image recognition and other fields.
Using ImageNet, a classic repository of millions of images for AI training, the team established that standard deep learning models gradually lose their plasticity when challenged with thousands of sequential tasks. These are ridiculously simple for humans—differentiating cats from houses, for example, or stop signs from school buses.
With this measure, any drop in performance means the AI is gradually losing its learning ability. The deep learning algorithms were accurate up to 88 percent of the time in earlier tests. But by task 2,000, they’d lost plasticity and performance had fallen to near or below baseline.
The updated algorithm performed far better.
It still uses back propagation, but with a small difference. A tiny portion of artificial neurons are wiped clean during learning in every cycle. To prevent disrupting whole networks, only artificial neurons that are used less get reset. The upgrade allowed the algorithm to tackle up to 5,000 different image recognition tasks with over 90 percent accuracy throughout.
In another proof of concept, the team used the algorithm to drive a simulated ant-like robot across multiple terrains to see how quickly it could learn and adjust with feedback.
With continuous back propagation, the simulated critter easily navigated a video game road with variable friction—like hiking on sand, pavement, and rocks. The robot driven by the new algorithm soldiered on for at least 50 million steps. Those powered by standard algorithms crashed far earlier, with performance tanking to zero around 30 percent earlier.
The study is the latest to tackle deep learning’s plasticity problem.
A previous study found so-called dormant neurons—ones that no longer respond to signals from their network—make AI more rigid and reconfiguring them throughout training improved performance. But they’re not the entire story, wrote Lyle and Pascanu. AI networks that can no longer learn could also be due to network interactions that destabilize the way the AI learns. Scientists are still only scratching the surface of the phenomenon.
Meanwhile, for practical uses, when it comes to AIs, “you want them to keep with the times,” said Dohare. Continual learning isn’t just about telling apart cats from houses. It could also help self-driving cars better navigate new streets in changing weather or lighting conditions—especially in regions with microenvironments, where fog might rapidly shift to bright sunlight.
Tackling the problem “presents an exciting opportunity” that could lead to AI that retains past knowledge while learning new information and, like us humans, flexibly adapts to an ever-changing world. “These capabilities are crucial to the development of truly adaptive AI systems that can continue to train indefinitely, responding to changes in the world and learning new skills and abilities,” wrote Lyle and Pascanu.
Image Credit: Jaredd Craig / Unsplash
Giant black holes in the centers of galaxies like our own Milky Way are known to occasionally munch on nearby stars.
This leads to a dramatic and complex process as the star plunging towards the supermassive black hole is spaghettified and torn to shreds. The resulting fireworks are known as a tidal disruption event.
In a new study published today in The Astrophysical Journal Letters, we have produced the most detailed simulations to date of how this process evolves over the span of a year.
A Black Hole Tearing Apart a SunAmerican astronomer Jack G. Hills and British astronomer Martin Rees first theorized about tidal disruption events in the 1970s and 80s. Rees’s theory predicted that half of the debris from the star would remain bound to the black hole, colliding with itself to form a hot, luminous swirl of matter known as an accretion disk. The disk would be so hot, it should radiate a copious amount of X-rays.
An artist’s impression of a moderately warm star – not at all what a black hole with a hot accretion disc would be like. Image Credit: Merikanto/Wikimedia Commons, CC BY-SABut to everyone’s surprise, most of the more than 100 candidate tidal disruption events discovered to date have been found to glow mainly at visible wavelengths, not X-rays. The observed temperatures in the debris are a mere 10,000 degrees Celsius. That’s like the surface of a moderately warm star, not the millions of degrees expected from hot gas around a supermassive black hole.
Even weirder is the inferred size of the glowing material around the black hole: several times larger than our solar system and expanding rapidly away from the black hole at a few percent of the speed of light.
Given that even a million-solar-mass black hole is just a bit bigger than our sun, the huge size of the glowing ball of material inferred from observations was a total surprise.
While astrophysicists have speculated the black hole must be somehow smothered by material during the disruption to explain the lack of X-ray emissions, to date nobody had been able to show how this actually occurs. This is where our simulations come in.
A Slurp and a BurpBlack holes are messy eaters—not unlike a five-year-old with a bowl of spaghetti. A star starts out as a compact body but gets spaghettified: stretched to a long, thin strand by the extreme tides of the black hole.
As half of the matter from the now-shredded star gets slurped towards the black hole, only 1 percent of it is actually swallowed. The rest ends up being blown away from the black hole in a sort of cosmic “burp.”
Simulating tidal disruption events with a computer is hard. Newton’s laws of gravity don’t work near a supermassive black hole, so one has to include all the weird and wonderful effects from Einstein’s general theory of relativity.
But hard work is what PhD students are for. Our recent graduate, David Liptai, developed a new do-it-Einstein’s-way simulation method which enabled the team to experiment by throwing unsuspecting stars in the general direction of the nearest black hole. You can even do it yourself.
Spaghettification in action, a close up of the half of the star that returns to the black hole.The resulting simulations, seen in the videos here, are the first to show tidal disruption events all the way from the slurp to the burp.
They follow the spaghettification of the star through to when the debris falls back on the black hole, then a close approach that turns the stream into something like a wriggling garden hose. The simulation lasts for more than a year after the initial plunge.
It took more than a year to run on one of the most powerful supercomputers in Australia. The zoomed-out version goes like this:
Zoomed-out view, showing the debris from a star that mostly doesn’t go down the black hole and instead gets blown away in an expanding outflow.What Did We Discover?To our great surprise, we found that the 1 percent of material that does drop to the black hole generates so much heat, it drives an extremely powerful and nearly spherical outflow. (A bit like that time you ate too much curry, and for much the same reason.)
The black hole simply can’t swallow all that much, so what it can’t swallow smothers the central engine and gets steadily flung away.
When observed like they would be by our telescopes, the simulations explain a lot. Turns out previous researchers were right about the smothering. It looks like this:
The same spaghettification as seen in the other movies, but as would be seen with an optical telescope [if we had a good-enough one]. It looks like a boiling bubble. We’ve called it the “Eddington envelope.”The new simulations reveal why tidal disruption events really do look like a solar-system-sized star expanding at a few percent of the speed of light, powered by a black hole inside. In fact, one could even call it a “black hole sun.”
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Price et al. (2024)
Sleep works magic on memory.
You might’ve felt these frustrations before: Trying to learn a guitar riff, shoot a free-throw, or nail a difficult phrase in a new language, but despite hours of practice, you’re just not getting it right. Then with a good night’s sleep—voilà, somehow, you’ve nailed the skills.
Neuroscientists have long known that brain waves during sleep etch learnings from the previous day into neural circuits for long-term storage. As we drift off, our brains remain hard at work. One region, a seahorse-shaped structure called the hippocampus, especially sparks with activity. This area is essential for translating what we learn into long-term memories during sleep.
Disruptions to electrical activity in the hippocampus can lead to memory problems in multiple neurological disorders, including schizophrenia and Alzheimer’s disease. But one question has always troubled neuroscientists.
Brain cells, or neurons, need to stay in a “Goldilocks zone” of activity to encode and store memories. Learning new things spikes activity in a specific set of neurons. But when they further increase their activity during sleep—like a car with a gas pedal and no brake—what’s to prevent them from hyperactivating and, in turn, destroying the brain’s ability to learn?
A new study from Cornell University suggests a way the brain balances itself during sleep. In recordings from multiple areas of the hippocampus in mice and rats, the team discovered a previously undetected brain wave that keeps brain cells in check. Dubbed BARR (for barrage of action potentials), these brain waves reset neurons so they can encode new experiences the next day, while enhancing memories during sleep.
“Sleep is not just a time for the body to rest but also for the mind to solidify memories,” wrote Drs. Xiang Mou and Daoyun Ji at Baylor College of Medicine in Houston, Texas, who were not involved in the study.
The results help explain why sleep promotes memory, and how its disruption can lead to brain disorders in schizophrenia, Alzheimer’s disease, and other neurological conditions associated with memory problems.
“This mechanism could allow the brain to reuse the same resources, the same neurons, for new learning the next day,” said study author Dr. Azahara Oliva at Cornell University in a press release.
Under the SeaAs we drift into unconsciousness every night during sleep, the hippocampus is hard at work. Shaped like a seahorse, this brain region has long been known as a hub for memory.
Patients with damage to the hippocampus lose their ability to create new memories. And decades of research shows the area processes the day’s learnings for long-term storage in other parts of the brain—and holds the key to retrieving those memories when needed.
But the region is hardly a one-trick pony. Imagine it as a town with multiple neighborhoods and highways connecting it to other brain regions. Each neighborhood plays a slightly different role. Some encode new memories, which are then shuffled to the cortex—the outer part of the brain—for longer storage and retrieval. Others link specific memories to joy, sadness, and other feelings through wiring connected to regions of the brain associated with emotion.
Scientists have already mapped out these neighborhoods. CA1, sitting at the front, extensively connects to other parts of the brain involved in reasoning and memory. CA3 encodes memories and potentially helps separate similar ones—for example, did I get that cup of coffee yesterday at that café, or was that a memory from a few days ago?
But the role of the middle child, CA2, has always been mysterious.
Sing Me to SleepEvery night we cycle through several stages of sleep. One stage, called non-rapid eye movement, occurs when we drift off to sleep and eventually transition from light sleep into deeper slumber.
This is when CA1 perks up. Neurons encoding memories from the day reactivate—kind of like replaying a memory on video, but at a faster rate.
These patterns, called sharp-wave ripples, help etch memories into the brain. Like waves on the ocean, they “splash” across other brain regions in electrical ebbs and flows that reconfigure neural connections. These waves help the hippocampus send learning to other regions where it can be stored in memory. But without a way to dampen the waves down, neurons hyperactivate, meaning they can no longer learn or store new information.
To study how sleep changes the brain, the team implanted electrodes into multiple parts of the hippocampus in mice and rats to monitor their brain activity.
The rodents then learned several tasks, for example, figuring out if an object had been removed. A bit like finding your favorite couch wasn’t where you expected it to be, this requires memory of its location. Other tests challenged the critters to navigate a maze and have social interactions—that is, remembering whether they’d met a previous acquaintance.
As the mice fell asleep, their brain activity showed signs of sharp-wave ripples. But surprisingly, CA2, the middle child, also sparked, with long-lasting bursts of activity spreading through the hippocampus. These BARR brain waves—never seen before—flared up in neurons that encode learning, which usually have higher levels of activity, to tamp them down in sleep.
In a way, as we sleep, our brain is in a kind of civil war. Neurons encoding memories reactivate to consolidate learning, while BARR brain waves keep them at bay so that they don’t overactivate.
A Brainy ScaleThe team focused on a type of brain cell that triggers BARR brain waves during sleep.
Using optogenetics—a way to turn neurons on or off using light—in rodents, they artificially disrupted BARR activity as the critters slept after learning several memory tasks. As a result, sharp-wave ripples, the type of brain activity usually associated with solidifying memory, lasted far longer.
Surprisingly, it made memory worse. On the surface, it doesn’t make sense: Wouldn’t more activity during sleep be better for memory? Not so much, explained the team. It’s all about balance.
“BARRs serve as a passive brake” that lowered increased neural activity in sleep, wrote Mou and Ji. The brain resets balance after a day of hard work. Disrupting BARR during sleep altered the animals’ memory, likely because their neural networks were functioning abnormally.
It’s not to say BARR is behind Alzheimer’s, schizophrenia, or other neurological disorders. Many questions remain. The team hasn’t yet determined where the brain waves start in the brain. How they counteract memory-making sharp-wave ripples during sleep also remains a mystery.
But by tinkering with these mechanisms, scientists can begin to battle memory disorders. They may also explore ways to re-write traumatic memories during sleep and help with depression, post-traumatic stress disorder, and other neurological conditions. Future studies could reveal more insights into how sleep controls memory, and why it breaks down in a variety of brain disorders.
Image Credit: Matteo Catanese / Unsplash
ARTIFICIAL INTELLIGENCEALS Stole His Voice. AI Retrieved It.
Benjamin Mueller | The New York Times“Halfway through trying to speak his first prompt aloud—’What good is that?’—a shaking, smiling Mr. Harrell crumpled into tears. …By day two, the machine was ranging across an available vocabulary of 125,000 words with 90 percent accuracy and, for the first time, producing sentences of Mr. Harrell’s own making. The device spoke them in a voice remarkably like his own, too: Using podcast interviews and other old recordings, the researchers had created a deep fake of Mr. Harrell’s pre-ALS voice.”
BIOTECHThis Researcher Wants to Replace Your Brain, Little by Little
Antonio Regalado | MIT Technology Review“A US agency pursuing moonshot health breakthroughs has hired a researcher advocating an extremely radical plan for defeating death. His idea? Replace your body parts. All of them. Even your brain. Jean Hébert, a new hire with the US Advanced Projects Agency for Health (ARPA-H), is expected to lead a major new initiative around ‘functional brain tissue replacement,’ the idea of adding youthful tissue to people’s brains.”
FUTUREHappy Birthday, Baby! What the Future Holds for Those Born Today
Kara Platoni | MIT Technology Review“Your arrival coincided with the 125th anniversary of this magazine. With a bit of luck and the right genes, you might see the next 125 years. How will you and the next generation of machines grow up together? We asked more than a dozen experts to imagine your joint future. We explained that this would be a thought experiment. What I mean is: We asked them to get weird.”
ARTIFICIAL INTELLIGENCEStudy Suggests That Even the Best AI Models Hallucinate a Bunch
Kyle Wiggers | TechCrunch“A recent study from researchers at Cornell, the universities of Washington and Waterloo and the nonprofit research institute AI2 sought to benchmark hallucinations by fact-checking models like GPT-4o against authoritative sources on topics ranging from law and health to history and geography. …’The most important takeaway from our work is that we cannot yet fully trust the outputs of model generations,’ Wenting Zhao, a doctorate student at Cornell and a co-author on the research, told TechCrunch. ‘At present, even the best models can generate hallucination-free text only about 35% of the time.'”
TRANSPORTATIONWhy Alaska Airlines Is Investing in a Jet That’s Like Nothing You’ve Seen Before
Patrick Sisson | Fast Company“[JetZero’s blended-wing-body (BWB) aircraft] concept boasts a more triangular, stretch design—where the cabin and wing blend together—creating more aerodynamic efficiency and lift. This allows the plane to fly higher, at around 45,000 feet, which further cuts wind resistance. Factor in the change in materials and construction, with bolted metal and composites swapped out for lighter, stitched carbon fiber, and a BWB jet can carry hundreds of passengers with half the fuel, a huge cost savings and environmental benefit.”
SPACENASA and Rocket Lab Aim to Prove We Can Go to Mars for 1/10 the Price
Aria Alamalhodaei | TechCrunch“Instead of spending $550 million on a mission into deep space, NASA set a goal to spend just one-tenth of that and gave each SIMPLEx mission a $55 million price cap, excluding launch. ESCAPADE is one of three missions the agency selected under the SIMPLEx program, and in all likelihood, the first that will actually launch.”
ENERGYInside a Green-Hydrogen Pilot Plant
Jesse Orrall | CNET“We got a look inside Verdagy’s pilot plant, where the company is testing its multimillion-dollar electrolyzer designed to turn renewable energy like wind and solar into hydrogen. …All together, Neese says, it’s ‘millions of dollars for an electrolyzer,’ but the estimated ‘tens of thousands of gallons of diesel equivalent produced per day’ of hydrogen will make green hydrogen competitive in cost with fossil fuels globally by 2030.”
ARTIFICIAL INTELLIGENCELLMs Are a Dead End to AGI, Says François Chollet
Kristin Houser | Big Think“Artificial general intelligence (AGI) could change the world, but no one seems to know how close we are to building it. Today’s generative AIs score well on benchmarks, but such benchmarks can be solved through memorization and don’t necessarily signal general intelligence. To accelerate progress in AI, François Chollet launched ARC Prize, a competition to see which AIs can score highest on a set of abstraction and reasoning tasks.”
AUTOMATIONIkea’s Stock-Counting Warehouse Drones Will Fly Alongside Workers in the US
Emma Roth | The Verge“The Swedish furniture chain announced that the autonomous drones will soon operate alongside workers in its Perryville, Maryland, distribution center, where Ikea started installation this summer. The Verity-branded drones also come with a new AI-powered system that allows them to fly around warehouses 24/7. That means they’ll now operate alongside human workers, helping to count inventory as well as identify if something’s in the wrong spot. Previously, the drones only flew during nonoperational hours.”
Image Credit: JetZero
Proteins are a bit like lights in your house. They have a job to do, and getting to them to do it involves switching them on and off with other proteins or molecules.
But it’s much easier to flip the switch on a light. In the body, billions of years of evolution have generated a complex web of molecular signals that act as biological switches for proteins.
This week, a team led by Dr. David Baker at the University of Washington offered a shortcut.
Using AI, they designed proteins that reliably transform themselves in the presence of a molecular switch—dubbed an “effector.” These designer proteins, unknown in nature, contain hinges that allow them to bend and assemble into different structures when dosed with an effector, and then disassemble into individual components when the effector disappears.
It’s a “startling advance for the field,” wrote Dr. A. Joshua Wand at Texas A&M University, who was not involved in the work.
The team designed proteins that can morph into myriad dynamic arrangements, such as rings or cages, loosely mimicking the behavior of their biological peers—for example, how the blood protein hemoglobin assembles to carry oxygen.
Switchable proteins open a world of possibility. Cage-like proteins could carry medication through the body and then, with a molecular flick of the switch, open to release it, allowing triggerable drug delivery. Other designs could potentially monitor disease-causing molecules in the body or pollutants in the environment. In synthetic biology, they could form the basis of biological circuits, acting as tunable switches that can predictably change a cell’s behavior.
“By designing proteins that can assemble and disassemble on command, we pave the way for future biotechnologies that may rival even nature’s sophistication,” said Baker in a press release.
Proteins, AssembleProteins are the body’s workhorses. They build and run our bodies. Protein networks determine when cells divide, thrive, or die. Scientists have long relied on proteins to develop vaccines, cancer therapies, and treatments for brain and heart disorders.
Structure is a crucial attribute, especially for larger proteins made up of multiple components. They need a stable shape so they can grasp other proteins and trigger biological responses, but the shape must also be able to change depending on the cell’s needs.
It’s a bit like having planks of wood for multiple house-restoration projects. The planks can combine to make a table, a set of stairs, or a planter for the garden. Similarly, our cells assemble protein “planks” into a variety of shapes—but with a twist.
Take hemoglobin, a protein in the blood that carries oxygen. It’s made up of four protein planks, each able to grab onto oxygen. But they act as a team: When one plank latches onto oxygen, it’s easier for others to do the same.
This type of molecular collaboration has inspired scientists for nearly a century. Here, oxygen is the effector. It flips a protein switch, helping proteins better carry oxygen through the body. In other words, it may be possible to optimize protein functions with an alternative effector drug.
The problem? The original inspiration is wonky. Sometimes hemoglobin proteins carry oxygen. Other times they don’t. In 1965, a French and American collaboration found out why. Each protein alternates between two three-dimensional shapes—one that carries oxygen and another that doesn’t. The shapes can’t coexist in the assembled protein to carry oxygen: It’s all-or-none, depending on the presence and amount of the effector.
The new study built on these lessons to guide their AI-designed proteins.
Shape ShiftersThe team tapped several advances in recent years—most of which they’ve led.
One is the use of AI to predict protein structure. Another is the design of a hinge-like protein that changes its shape to take on two different forms (a bit like a biological transistor). The last is an AI that can stitch protein “planks” together into structures.
The team first used AI to design a group of flexible proteins, each with a hinge and two ridged arms. This setup keeps the protein’s structure stable, but lets it bend at the hinges. The hinge does double duty: It’s also a sensor. In the presence of an effector molecule, the protein changes its shape from a flat plank to a hinged “V” shape.
As a proof of concept, the team synthesized multiple AI-generated proteins and tested them in the lab. In one of these, the proteins formed a ring-like structure when given a customized effector made of peptides, or small protein chunks.
In another test, they designed a protein that grabbed onto another similarly shaped protein in the presence of an effector. Processes like this are often used by cells to change their inner workings, and in synthetic biology, they’re switches that trigger a molecular response—for example, turning genes on or off or altering the fate of a cell. Nearly 40 percent of these designer proteins dissolve in water, making them more compatible with our bodies.
Going further, the team designed a protein with two hinges connected by a short loop. In the presence of an effector, the proteins twisted in a way that mimicked hemoglobin.
Finally, they explored ways to disassemble the proteins.
“This addresses a major current protein design challenge,” wrote the authors.
A useful tool might form a cage that carries and releases a payload of medicine when encountering specific signals in the body. Choosing from the proteins in their repertoire, the team engineered a different effector that broke the cage back down into its components.
Similarly to how proteins assemble in our bodies, the engineered proteins also had the “amp-up” effect, in that grabbing onto an effector made it easier for other components to do the same—in a virtuous cycle. However, the proteins developed in the study are all unknown to nature, opening a new space “unexplored by natural evolution,” wrote the team.
They could be adapted into controllable nanomaterials or drug packaging systems that unleash cargo with a trigger. Other uses include biosensing, which can make cell therapies—such as those for cancer—more traceable, and protein nanobots that morph into different structures.
Still, many challenges remain.
This type of regulation “in nature is much more varied and complicated,” wrote Wand. Whether AI-designed proteins can fully capture the shape-shifting capabilities of natural proteins remains to be seen.
Image Credit: Baker Lab
Peter Shor published one of the earliest algorithms for quantum computers in 1994. Running Shor’s algorithm on a hypothetical quantum computer, one could rapidly factor enormous numbers—a seemingly innocuous superpower. But because the security of digital information relies on such math, the implications of Shor’s algorithm were ground-shaking.
It’s long been prophesied that modern cryptography, employed universally across the devices we use every day, will die at the hands of the first practical quantum computer.
Naturally, researchers have been searching for secure alternatives.
In 2016, the US National Institute of Standards and Technology (NIST) announced a competition to create the first post-quantum cryptographic algorithms. These programs would run on today’s computers but defeat attacks by future quantum computers.
Beginning with a pool of 82 submissions from around the world, NIST narrowed the list to four in 2022. The finalists went by the names CRYSTALS-Kyber, CRYSTALS-Dilithium, Sphincs+, and FALCON. This week, NIST announced three of these have become the first standardized post-quantum algorithms. They’ll release a standard draft of the last, FALCON, by the end of the year.
The algorithms, according to NIST, represent the best of the best. Kyber, Dilithium, and FALCON employ an approach called lattice-based cryptography, while Sphincs+ uses an alternative hash-based method. They’ve survived several years of stress testing by security experts and are ready for immediate use.
The release includes code for the algorithms alongside instructions on how to implement them and their intended uses. Like earlier encryption standards developed by the agency in the 1970s, it’s hoped wide adoption will ensure interoperability between digital products and consistency, lowering the risk of error. The first of the group, renamed ML-KEM, is for general encryption, while the latter three (now ML-DSA, SLH-DSA, and FN-DSA) are for digital signatures—that is, proving that sources are who they say they are.
Arriving at standards was a big effort, but broad adoption will be bigger.
While the idea that future quantum computers could defeat standard encryption is fairly uncontroversial, when it will happen is murkier. Today’s machines, still small and finicky, are nowhere near up to the task. The first machines able to complete useful tasks faster than classical computers aren’t expected until later this decade at the very earliest. But it’s not clear how powerful these computers will have to be to break encryption.
Still, there are solid reasons to get started now, according to proponents. For one, it’ll take as long as 10 to 15 years to roll out post-quantum cryptography. So, the earlier we kick things off the better. Also, hackers may steal and store encrypted data today with the expectation it can be cracked later—a strategy known as “harvest now, decrypt later.”
“Today, public key cryptography is used everywhere in every device,” Lily Chen, head of cryptography at NIST, told IEEE Spectrum. “Now our task is to replace the protocol in every device, which is not an easy task.”
There are already some early movers, however. The Signal Protocol underpinning Signal, WhatsApp, and Google Messages—products used by more than a billion people—implemented post-quantum cryptography based on NIST’s Kyber algorithm alongside more traditional encryption in late 2023. Apple did the same for iMessages earlier this year.
It’s notable both opted to run the two in parallel, as opposed to going all-in on post-quantum security. NIST’s algorithms have been scrutinized, but they haven’t been out in the wild for nearly as long as traditional approaches. There’s no guarantee they won’t be defeated in the future.
An algorithm in the running two years ago, SIKE, met a quick and shocking end when researchers took it down with some clever math and a desktop computer. And this April, a researcher at Tsinghua University, Yilei Chen, published a pre-print on the arXiv in which he said he’d shown lattice-based cryptography actually was vulnerable to quantum computers, though his approach was later shown to be flawed.
To be safe, NIST is developing backup algorithms. The agency is currently vetting two groups representing alternative approaches for general encryption and digital signatures. In parallel, scientists are working on other forms of secure communication using quantum systems themselves, though these are likely years from completion and may complement rather than replace post-cryptographic algorithms like those NIST is standardizing.
“There is no need to wait for future standards,” said Dustin Moody, a NIST mathematician heading the project, in a release. “Go ahead and start using these three. We need to be prepared in case of an attack that defeats the algorithms in these three standards, and we will continue working on backup plans to keep our data safe. But for most applications, these new standards are the main event.”
Image Credit: IBM
The stars aren’t fixed and unchanging, unlike what many ancient people thought. Once in a while, a star appears where there wasn’t one before, and then it fades away in a matter of days or weeks.
The earliest record of such a “guest star,” named so by ancient Chinese astronomers, is a star that suddenly appeared in skies around the world on July 4, 1054. It quickly brightened, becoming visible even during the day for the next 23 days.
Astronomers in Japan, China, and the Middle East observed this event, as did the Anasazi in what is now New Mexico.
In the second half of 2024, a nova explosion in the star system T Coronae Borealis, or T CrB, will once again be visible to people on Earth. T CrB will appear 1,500 times brighter than usual, but it won’t be as spectacular as the event in 1054.
Art depicts the Roman Emperor Henry III viewing the supernova explosion of 1054.I’m a space scientist with a passion for teaching physics and astronomy. I love photographing the night sky and astronomical events, including eclipses, meteor showers, and once-in-a-lifetime astronomical events such as the T CrB nova. T CrB will become, at best, the 50th brightest star in the night sky—brighter than only half the stars in the Big Dipper. It might take some effort to find, but if you have the time, you’ll witness a rare event.
What Is a Nova?In 1572, the famous Danish astronomer Tycho Brahe observed a new star in the constellation Cassiopeia. After reporting the event in his work “De Nova Stella,” or “On the New Star,” astronomers came to associate the word nova with stellar explosions.
Stars, regardless of size, spend 90 percent of their lives fusing hydrogen into helium in their cores. How a star’s life ends, though, depends on the mass of the star. Very massive stars—those more than eight times the mass of our sun—detonate in dramatic supernova explosions, like the ones people observed in 1054 and 1572.
In lower mass stars, including our sun, once the hydrogen in the core is exhausted, the star expands into what astronomers call a red giant. The red giant is hundreds of times its original size and more unstable. Eventually, all that’s left is a white dwarf—an Earth-sized remnant made up of carbon and oxygen. White dwarves are a hundred thousand times denser than diamond. Unless they’re part of a binary star system, where two stars orbit each other, they slowly fade in brightness over billions of years and eventually disappear from sight.
T CrB is a binary star system—it’s made up of a red giant and a white dwarf, which orbit each other every 228 days at about half the distance between Earth and the sun. The red giant is nearing the end of its life, so it has expanded dramatically, and it’s feeding material into a rotating disk of matter called an accretion disk, which surrounds the white dwarf.
Matter from the accretion disk, which is made mostly of hydrogen, spirals in and slowly accumulates on the surface of the white dwarf. Over time, this blanket of hydrogen becomes thicker and denser, until its temperature exceeds 18 million degrees Fahrenheit (10 million degrees Celsius).
A nova is a runaway thermonuclear reaction similar to the detonation of a hydrogen bomb. Once the accretion disk gets hot enough, a nova occurs where the hydrogen ignites, gets blown outward, and emits bright light.
When Will It Occur?Astronomers know of 10 recurrent novae—stars that have undergone nova explosions more than once. T CrB is the most famous of these. It erupts on average every 80 years.
Because T CrB is 2,630 light-years from Earth, it takes light 2,630 years to travel the distance from T CrB to Earth. The nova we will see later this year occurred over 2,000 years ago, but its light will be just reaching us.
The accretion of hydrogen on the surface of the white dwarf is like sand in an 80-year hourglass. Each time a nova occurs and the hydrogen ignites, the white dwarf itself is unaffected, but the surface of the white dwarf is wiped clean of hydrogen. Soon after, hydrogen begins accreting on the surface of the white dwarf again: The hourglass flips, and the 80-year countdown to the next nova begins anew.
Careful observations during its past two novae in 1866 and 1946 showed that T CrB became slightly brighter about 10 years before the nova was visible from Earth. Then, it briefly dimmed. Although scientists aren’t sure what causes these brightness changes, this pattern has repeated, with a brightening in 2015 and a dimming in March 2023.
Based on these observations, scientists predict the nova will be visible to us sometime in 2024.
How Bright Will It Be?Astronomers use a magnitude system first devised by Hipparchus of Nicaea more than 2,100 years ago to classify the brightness of stars. In this system, a difference of 5 in magnitude signifies a change by a factor of 100 in brightness. The smaller the magnitude, the brighter the star.
In dark skies, the human eye can see stars as dim as magnitude 6. Ordinarily, the visible light we receive from T CrB comes entirely from its red giant, a magnitude 10 star barely visible with binoculars.
During the nova event, the white dwarf’s exploding hydrogen envelope will brighten to a magnitude 2 or 3. It will briefly become the brightest star in its home constellation, Corona Borealis. This maximum brightness will last only several hours, and T CrB will fade from visibility with the naked eye in a matter of days.
What the Los Angeles sky will look like on, as an example, Aug. 15, 2024, at 10 p.m. local time. The view will be very similar across the US, but T CrB will get closer and closer to the horizon and will be halfway between where it’s shown here and the horizon by early September. By early October, it will be right on the horizon. Image Credit: Vahé Peroomian/StellariumWhere to LookCorona Borealis is not a prominent constellation. It’s nestled above Bootes and to the west of Ursa Major, home to the Big Dipper, in northern skies.
To locate the constellation, look due west and find Arcturus, the brightest star in that region of the sky. Then look about halfway between the horizon and zenith—the point directly above you—at 10 p.m. local time in North America.
Corona Borealis is approximately 20 degrees above Arcturus. That’s about the span of one hand, from the tip of the thumb to the tip of the pinky, at arm’s length. At its brightest, T CrB will be brighter than all the stars in Corona Borealis, but not as bright as Arcturus.
You can also use an interactive star chart such as Stellarium, or one of the many apps available for smartphones, to locate the constellation. Familiarizing yourself with the stars in this region of the sky before the nova occurs will help identify the new star once T CrB brightens.
Although T CrB is too far from Earth for this event to rival the supernova of 1054, it is nevertheless an opportunity to observe a rare astronomical event with your own eyes. For many of us, this will be a once-in-a-lifetime event.
For children, however, this event could ignite a passion in astronomy. Eighty years in the future, they may look forward to observing it once again.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA/Goddard Space Flight Center
Thanks to antiviral medications, HIV infection is no longer a death sentence. With a cocktail of drugs, people with HIV can keep the virus in check. Introduced more recently, PrEP, or pre-exposure prophylaxis, can guard uninfected people from potential infections.
The pills, taken with a sip off water, have protected pregnant women at high risk of HIV. The treatment also dramatically slashes the risk of catching the virus in other populations.
But antivirals come with side effects. Nausea, fatigue, dizziness, and pain are common. When taken for years—which is typical—the drugs raise cholesterol levels and increase the chances of depression, diabetes, and liver and kidney damage. They’re also expensive and very hard to come by in some regions of the world. As an alternative, scientists have long been working on an HIV vaccine, but so far to no avail.
This week, an international team led by Dr. Leor Weinberger at the University of California, San Francisco, tapped into an age-old idea in the battle against viruses, but with a modern twist.
One way to make vaccines is to create viruses stripped of harmful traits but still able to infect cells. In the new study, scientists built on this idea to develop a one-shot antiviral HIV therapy. By removing HIV’s disease-causing genes, the team created “benevolent twins” called TIPs—or therapeutic interfering particles—which outcompete HIV and limit its ability to reproduce.
A single shot of TIPs reduced the amount of virus inside infected monkeys by up to 10,000-fold and helped the treated animals live longer.
The new approach is a virus-like living drug. Like its evil twin, HIV, it replicates and spreads in the body. Because both viruses use the same cell machinery to reproduce, the engineered virus dominates precious resources, elbowing out disease-causing viruses and limiting their spread. TIPs also kept the virus’s levels at bay in cells from HIV-positive people.
Plans are underway to test the idea in humans. If safe and effective, the long-lasting shot could help people who don’t have regular access to antiviral drugs.
ART to TIPsHIV is a formidable enemy. The virus rapidly evolves and spins out variants that outcompete efforts to combat it.
Scientists have long sought an HIV vaccine. Although several vaccines are in clinical trials, so far the virus has largely stymied researchers.
Antiviral drugs have had a better run. Dubbed ART, for antiretroviral therapy, these involve taking multiple medications every day to keep the virus at bay. The drugs have been game-changers for people with HIV. But they don’t cure the disease, and missing doses can reignite the virus.
Several new ideas are in the works. In 2019, stem cell implants freed three people of the virus. The implants came from people with a genetic mutation that naturally fights HIV. In July, a seventh person was reportedly “cured” of HIV using a similar strategy—although the donor cells only had one copy of the HIV-resistant gene, rather than two copies in previous cases.
While promising, cell therapies are expensive and technically difficult. Over a decade ago, Weinberger came up with a novel idea: Give people already infected with HIV a stripped-down variant without the ability to cause harm. Because both viruses require the same resources to reproduce, the benign twin could outcompete the deadly version.
“I think we need to try something new,” he recently told Science.
Tipping PointHIV requires cells to replicate.
The virus grabs onto a type of immune cell and pumps its genetic material into the host. Then, hijacking the machinery in these cells, the virus integrates its DNA into the genome. The cells replicate these viral genes and assemble them into a new generation of sphere-like viruses, ready to be released into the bloodstream to further multiply and spread.
However, the entire process relies on limited resources. Here’s where TIPs come in.
The team grew HIV particles in petri dishes and deleted disease-causing genes over multiple generations. They were finally left with stripped-down versions of HIV, or TIPs.
In a way, the neutered HIV becomes a parasite that can fight off the natural virus. Because TIPs have fewer genetic letters, they replicate more quickly than natural HIV, allowing them to flood the cell and spread in lieu of their natural counterparts.
In a test, the team injected TIPs into six young macaque monkeys, infected with a synthetic monkey version of HIV a day later. After 30 weeks, in five treated monkeys, the single-shot treatment reduced the amount of virus in the watery part of their blood, or plasma, 10,000-fold. Viral levels also tanked in lymph nodes, where HIV swarms and replicates. In contrast, those who went untreated got increasingly sick.
A computer model translated these results for human therapy, suggesting TIPs could reduce HIV 1,000-fold or more in humans. Although not as dramatic as in monkeys, the single-shot treatment could reduce the virus to levels so low it couldn’t be transmitted to others.
A New Therapy?Many people with HIV are already on antiviral drugs.
The team next asked if their shot could replace these drugs. In cells in petri dishes, they found TIPs sprang into action once the drugs were removed, limiting HIV growth and protecting cells.
In cells infected with multiple strains of HIV, the strains swap DNA and rearrange their genetic material, which is partly why HIV is so hard to tame with vaccines. Antiviral drugs can trigger this response and eventually cause resistance. TIPs, in contrast, seem to keep it at bay.
TIPs isn’t the only new treatment in town. Long-acting HIV drugs are in clinical trials, with some needing only two shots a year. But these still rely on antiviral drugs.
To be clear, TIPs doesn’t cure HIV. Like antiviral drugs, it keeps the virus at bay. But rather than taking a cocktail of pills every day, a single jab could last months with lower chance of resistance.
There are downsides, however. Like HIV, TIPs can be transmitted to others through bodily fluids, raising ethical issues about disclosure. The shots could also lead to dangerous immune flareups, although this didn’t happen in the monkey studies.
The team is planning to study potential toxicity to the genome and inflammation and further investigate how TIPs work once antiviral drugs have been halted in monkeys.
They’re also looking to recruit people with HIV, and another terminal illness, to test the effects of TIPs after stopping antiviral drugs. The goal is to begin the trial next year.
“The real test, of course, will be the upcoming human clinical trials,” said Weinberger in a press release. “But if TIPs prove effective, we could be on the brink of a new era in HIV treatment that could bring hope to millions of people—particularly in areas where access to antiviral drugs remains a challenge.”
Image Credit: HIV (blue) replicating from a T Cell (gold). NIAID / Flickr
In November 2022, Icon and Lennar started 3D printing homes for a new neighborhood in Texas. Now, according to a report by Reuters, the 100-home project is nearly complete.
While foundations, roofing, and finishes were built and installed traditionally, the walls of each house were constructed by Icon’s Vulcan 3D printer. Vulcan uses a long, crane-like robotic arm tipped with a nozzle to extrude beads of concrete like frosting on a cake. Directed by a digital design, the printer lays down a footprint, then builds up the walls layer by layer.
One of the earliest large-scale projects for 3D-printed homes, it showcases some of the benefits: A house can be printed in around three weeks with Vulcan and a single crew of workers. Icon partnered with design firm Bjark Ingels Group on eight floor plans for the ranch-style homes, each with three- to four-bedrooms and ranging from 1,574 to 2,112 square feet.
Around 25 percent of the homes have been sold with prices ranging from $450,000 to $600,000, about average for the area. Already, buyers are moving in. A couple interviewed by Reuters said their home feels solidly constructed, and its thick concrete walls insulate well, keeping the interior cool in the baking Texas summer. The homes come stock with solar panels to convert all that sunshine into power. The one downside? The concrete blocks WiFi signals, necessitating a mesh network for internet.
The idea of 3D printing homes isn’t new. The earliest projects date back to around the turn of this century. Over the years, startups like Icon have honed the process, perfecting concrete materials and robotic delivery systems and identifying which steps are best suited for 3D printing.
Recently, the technology has made its way into commercial development. In 2021, a home printed by SQ4D was sold in New York. Mighty Buildings, a 3D printing startup that began by printing and selling pre-fab ADUs, raised $52 million last year. Now, the company has its sights set on larger structures and whole communities. Unlike Icon, Mighty prints its structures in parts in a factory and then ships them out for assembly on site.
Overall, 3D printing has been hailed as a cheaper, faster, less resource-intensive way to build. Proponents hope it can bring more affordable housing to those in need. And to that end, Icon has partnered with New Story to 3D print homes in Mexico for families living in extreme poverty and with Mobile Loaves & Fishes to print homes in Austin for those experiencing chronic homelessness.
To date, however, market prices of commercial 3D-printed homes haven’t been dramatically lower than traditionally built homes. While some steps offer savings, others may bring higher costs—like fitting windows or other fixtures tailored to today’s building technologies into less conventional 3D-printed designs. And beyond building costs, prices on the open market are based on demand and how much buyers are willing to pay.
To bring costs down, Icon announced Initiative 99 in 2023, a competition to design 3D-printed homes that can be built for under $99,000. They announced winners for Phase I of the competition at this year’s SXSW.
It’s still early days for 3D printing as a commercial homebuilding technology. The Texas project is one of the first at scale, and costs may yet decline as Icon and others figure out how to optimize the process and slot their work into the existing ecosystem.
In the meantime, a handful of Texans will settle into their futuristic homes—nestled between walls of corduroy concrete to keep the heat at bay.
Image Credit: Icon
ROBOTICSMan vs. Machine: DeepMind’s New Robot Serves up a Table Tennis Triumph
Benj Edwards | Ars Technica“On Wednesday, researchers at Google DeepMind revealed the first AI-powered robotic table tennis player capable of competing at an amateur human level. The system combines an industrial robot arm called the ABB IRB 1100 and custom AI software from DeepMind. While an expert human player can still defeat the bot, the system demonstrates the potential for machines to master complex physical tasks that require split-second decision-making and adaptability.”
BIOTECHEngineered Virus Steals Proteins From HIV, Pointing to New Therapy
Carl Zimmer | The New York Times“Scientists have developed a new weapon against HIV: a molecular mimic that invades a cell and steals essential proteins from the virus. A study published in Science on Thursday reported that this viral thief prevented HIV from multiplying inside of monkeys. The new therapeutic approach will soon be tested in people, the scientists said.”
ETHICSOpenAI Warns Users Could Become Emotionally Hooked on Its Voice Mode
Will Knight and Reece Rogers | Wired“During the red teaming, or stress testing, of GPT-4o, for instance, OpenAI researchers noticed instances of speech from users that conveyed a sense of emotional connection with the model. For example, people used language such as ‘This is our last day together.’ Anthropomorphism might cause users to place more trust in the output of a model when it ‘hallucinates’ incorrect information, OpenAI says. Over time, it might even affect users’ relationships with other people.”
ROBOTICSFigure 02 Robot Is a Sleeker, Smarter Humanoid
Evan Ackerman | IEEE Spectrum“This thing looks slick. I’d say that it’s maybe a little too far on the sinister side for a robot intended to work around humans, but the industrial design is badass and the packaging is excellent, with the vast majority of the wiring now integrated within the robot’s skins and flexible materials covering joints that are typically left bare.”
ENERGYMore Than Half of All New Cars Sold in China Last Month Were Electric Vehicles
William Gavin | Quartz“Sales of what China calls new energy vehicles (NEVs)—any vehicles that mostly use or are entirely dependent on electricity for their operation—increased by 37% year-over-year in July, according to data from the China Passenger Car Association (CPCA). Thanks to that growth, and an overall decrease in sales in the world’s largest auto market, NEVs accounted for 50.7% of new car sales last month. That’s a major jump compared to sales just three years ago, when NEVs made up just 7% of overall vehicle sales in China.”
TECHBreaking Down the Tech Giants’ AI Spending Surge
Nate Rattner | The Wall Street Journal“Big technology companies deepened their commitments to artificial-intelligence efforts in the latest quarter, pouring billions of dollars into capital-spending projects and telling investors more is on the way. In earnings statements over the past two weeks, Amazon, Microsoft, Facebook parent Meta Platforms, and Google parent Alphabet each reported jumps in purchases of property and equipment, a measure of capital spending. For all but Meta, the latest quarterly figure was the highest in years.”
SCIENCEWatch a Video Showing What Happens in Our Brains When We Think
Jessica Hamzelou | MIT Technology Review“In this video, which I’ve converted to a GIF, you can see the pattern of electrical activity in the man’s brain as he recites numbers. Each dot represents the voltage sensed by an electrode on the array on the man’s brain, over a region involved in speech. …The video has been slowed down 20-fold, because ‘thoughts happen faster than the eye can see,’ says Rapoport.”
FUTUREThe Search for Alien Life Just Hit a Depressing Setback
Adam Kovac | Gizmodo“The search for alien life just got a bit more complicated. Red dwarfs, young and dim stars thought by many astronomers to be the most likely hosts for life-sustaining planets, come with a significant drawback: They frequently emit deadly ultraviolet radiation flares, which are much more powerful than astronomers previously thought.”
COMPUTINGQuantum Cryptography Has Everyone Scrambling
Margo Anderson | IEEE Spectrum“While the technology world awaits NIST’s latest ‘post-quantum’ cryptography standards this summer, a parallel effort is underway to also develop cryptosystems that are grounded in quantum technology—what are called quantum-key distribution or QKD systems. As a result, India, China, and a range of technology organizations in the European Union and United States are researching and developing QKD and weighing standards for the nascent cryptography alternative.”
TECHCalifornians Are Getting Apple Wallet Drivers Licenses This Year
Florence Ion | Gizmodo“My household has a saying that anytime we leave for the outside world, we should have ‘keys, wallet, phone.’ It helps remind us of the basics that should be in hand before the door gets locked from the outside. If you’re in California, you can look forward to cutting down the list to two simple items: keys and a phone. A leak reveals that official digital California drivers licenses and identification cards are coming to Apple Wallet.”
Image Credit: Norbert Kowalczyk / Unsplash
Bacteria and antibiotics have been in a roughly century-long game of cat and mouse. Unfortunately, bacteria are gaining the upper hand.
According to the World Health Organization, antibiotic resistance is a top public health risk that was responsible for 1.27 million deaths across the globe in 2019. When repeatedly exposed to antibiotics, bacteria rapidly learn to adapt their genes to counteract the drugs—and share the genetic tweaks with their peers—rendering the drugs ineffective.
Superpowered bacteria also torpedo medical procedures—surgery, chemotherapy, C-sections—adding risk to life-saving therapies. With antibiotic resistance on the rise, there are very few new drugs in development. While studies in petri dishes have zeroed in on potent candidates, some of these also harm the body’s cells, leading to severe side effects.
What if there’s a way to retain their bacteria-fighting ability, but with fewer side effects? This month, researchers used AI to reengineer a toxic antibiotic. They made thousands of variants and screened for the ones that maintained their bug-killing abilities without harming human cells.
The AI used in the study is a large language model similar to those behind famed chatbots from Google, OpenAI, and Anthropic. The algorithm sifted 5.7 million variants of the original antibiotic and found one that maintained its potency but with far less toxicity.
In lab tests, the new variant rapidly broke down bacteria “shields”—a fatty bubble that keeps the cells intact—but left host cells undamaged. Compared to the original antibiotic, the newer version was far less toxic to human kidney cells in petri dishes. It also rapidly eliminated deadly bacteria in infected mice with minimal side effects. The platform can also be readily adapted to screen other drugs in development, including those for various types of cancers.
“We have found that large language models are a major step forward for machine learning applications in protein and peptide engineering,” said Dr. Claus Wilke, a University of Austin biologist and data scientist and an author on the study, in a press release.
Insane in the MembraneAntibiotics work in several ways. Some disrupt bacteria’s ability to create proteins. Others inhibit the copying of their genetic material, halting reproduction. Yet more selectively destroy their metabolisms.
Each strategy took years to research and even longer to develop safe and effective antibiotics. But bacteria rapidly evolve to evade these drugs.
Overuse of antibiotics in medicine and agriculture is giving rise to “superbugs” resistant to even the toughest current drugs. Once a strain of bacteria learns to evade a mechanism—say, hindering protein production—it readily blocks other drugs that target the same strategy.
Resistance can also rapidly spread through a bacterial population. Unlike our genetic material, which is encapsulated inside a nut-like structure, bacterial DNA freely floats around in their cells. Genetic changes—for example, those that allow bacteria to evade antibiotics—can be transmitted to other similar bacteria through temporary biological “tunnels” that literally connect the two cells. In other words, antibiotic resistance spreads fast.
That is, if given the chance.
For antibiotic resistance to develop, the bacteria need to survive the initial onslaught. Extremely deadly treatments, including a class called antimicrobial peptides, wipe out bacteria before they can adapt. These drugs rapidly break up the fatty protective barrier surrounding all bacterial cells. Decades in the works, scientists have made many of these molecules.
The problem? They also harm the membranes protecting our own cells, resulting in toxicity that makes most of them unusable in people. Although a library of these hyper-potent antibiotic drugs already exists, like underperforming ball players, they’ve mostly been benched.
Safe and SoundThe new study aimed to rehabilitate antimicrobial peptides by tweaking one called Protegrin-1. While extremely efficient at killing bacteria, it’s too toxic for human use. The researchers wanted to see if they could dial down side effects but maintain its bacteria-killing prowess.
Led by Dr. Bryan Davies, the team had previously developed a system to rapidly screen hundreds of thousands of peptides to see if they could kill harmful bacteria.
Called SLAY, for Surface Localized Antimicrobial Display, the system looks like a bunch of tetherballs with one end of each fixed to a biological surface and the other—this is the antimicrobial peptide—floating around to capture bacteria.
The researchers then engineered over 5.7 million Protegrin-1 variants. “This is a massive increase in diversity over the 18 single mutants” in previous studies, wrote the authors.
Next, they turned to AI large language models. Known for their ability to generate text, audio, and videos, this type of algorithm learns by ingesting terabytes of data and can spit out responses based on a specific prompt. While mostly used to generate text, scientists have increasingly embraced their capacity to “dream up” new proteins or other drugs.
The study used several prompts to guide the AI’s search: Things like, the drug has to target bacteria membranes, and it needs to break those up without harming human cells. The AI screened the available pool of variants and found one that hit the sweet spot—a new version dubbed bacterially selective Protegrin-1.2—that met all the guidelines.
Tested in petri dishes, the variant rapidly broke down membranes in Escherichia coli, a common type of bacteria often used for research, within half an hour. Human red blood cells, meanwhile, thrived under the same circumstances, even when exposed to levels 100 times higher than the bacteria. Rather than indiscriminatingly killing off both bacteria and human cells, the AI-approved antibiotic zeroed in on the pathogen.
Protegrin-1 has a reputation for causing kidney harm. The team pitted Protegrin-1.2 against the original and Colistin, an antibiotic used as a last-resort treatment, in cultured human kidney cells. The variant topped the others in safety measures, showing less cell membrane damage.
The team also treated mice infected with a type of multidrug-resistant bacteria—which roams hospitals—with the AI-selected antibiotic. Six days later, critters treated with the new version had lower levels of bacteria in multiple organs compared to untreated mice. Some had zero signs of infection at all. Compared to Protegrin-1, the new version “is significantly less toxic to mice,” wrote the authors.
Although the study focused on antibiotics, the team envisions using a similar strategy to reengineer other drugs previously thought too toxic for humans. Recently, another team used AI to determine the structure of small chemicals useful in antibiotic and cancer therapies but previously discarded by chemists as unusable in safe and effective medications.
“Many use cases that weren’t feasible with prior approaches are now starting to work. I foresee that these and similar approaches are going to be used widely for developing therapeutics or drugs going forward,” said Wilke.
Image Credit: x / x
Language enables people to transmit thoughts to each other because each person’s brain responds similarly to the meaning of words. In newly published research, my colleagues and I developed a framework to model the brain activity of speakers as they engaged in face-to-face conversations.
We recorded the electrical activity of two people’s brains as they engaged in unscripted conversations. Previous research has shown that when two people converse, their brain activity becomes coupled, or aligned, and that the degree of neural coupling is associated with better understanding of the speaker’s message.
A neural code refers to particular patterns of brain activity associated with distinct words in their contexts. We found that the speakers’ brains are aligned on a shared neural code. Importantly, the brain’s neural code resembled the artificial neural code of large language models.
The Neural Patterns of WordsA large language model is a machine learning program that can generate text by predicting what words most likely follow others. Large language models excel at learning the structure of language, generating humanlike text, and holding conversations. They can even pass the Turing test, making it difficult for someone to discern whether they are interacting with a machine or a human. Like humans, large language models learn how to speak by reading or listening to text produced by other humans.
By giving the large language model a transcript of the conversation, we were able to extract its “neural activations,” or how it translates words into numbers, as it “reads” the script. Then, we correlated the speaker’s brain activity with both the large language model’s activations and with the listener’s brain activity. We found that the large language model’s activations could predict the speaker and listener’s shared brain activity.
To understand each other, people have a shared agreement on the grammatical rules and the meaning of words in context. For instance, we know to use the past tense form of a verb to talk about past actions, as in the sentence: “He visited the museum yesterday.” Additionally, we intuitively understand that the same word can have different meanings in different situations. For instance, the word cold in the sentence “you are cold as ice” can refer either to one’s body temperature or personality trait, depending on the context. Due to the complexity and richness of natural language, until the recent success of large language models, we lacked a precise mathematical model to describe it.
Our study found that large language models can predict how linguistic information is encoded in the human brain, providing a new tool to interpret human brain activity. The similarity between the human brain’s and the large language model’s linguistic code has enabled us, for the first time, to track how information in the speaker’s brain is encoded into words and transferred, word by word, to the listener’s brain during face-to-face conversations. For example, we found that brain activity associated with the meaning of a word emerges in the speaker’s brain before articulating a word, and the same activity rapidly reemerges in the listener’s brain after hearing the word.
Powerful New ToolOur study has provided insights into the neural code for language processing in the human brain and how both humans and machines can use this code to communicate. We found that large language models were better able to predict shared brain activity compared with different features of language, such as syntax, or the order in which words connect to form phrases and sentences. This is partly due to the large language models’ ability to incorporate the contextual meaning of words, as well as integrate multiple levels of the linguistic hierarchy into one model: from words to sentences to conceptual meaning. This suggests important similarities between the brain and artificial neural networks.
An important aspect of our research is using everyday recordings of natural conversations to ensure that our findings capture the brain’s processing in real life. This is called ecological validity. In contrast to experiments in which participants are told what to say, we relinquish control of the study and let the participants converse as naturally as possible. This loss of control makes it difficult to analyze the data because each conversation is unique and involves two interacting individuals who are spontaneously speaking. Our ability to model neural activity as people engage in everyday conversations attests to the power of large language models.
Other DimensionsNow that we’ve developed a framework to assess the shared neural code between brains during everyday conversations, we’re interested in what factors drive or inhibit this coupling. For example, does linguistic coupling increase if a listener better understands the speaker’s intent? Or perhaps complex language, like jargon, may reduce neural coupling.
Another factor that can influence linguistic coupling may be the relationship between the speakers. For example, you may be able to convey a lot of information with a few words to a good friend but not to a stranger. Or you may be better neurally coupled to political allies rather than rivals. This is because differences in the way we use words across groups may make it easier to align and be coupled with people within rather than outside our social groups.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Mohamed Hassan / Pixabay
As the bitcoin gold rush dries up, crypto miners are finding it hard to make ends meet. But for many there’s a silver lining—the facilities they’ve set up are perfect for Silicon Valley’s latest obsession with artificial intelligence.
Crypto mining can be a profitable but highly volatile endeavor. It involves creating massive datacenters packed with specialized computer chips and using them to solve the mathematical puzzles underpinning the security of various cryptocurrencies. In exchange, the miners win some of that cryptocurrency as a reward.
Most miners make the bulk of their money from bitcoin. But earlier this year, an event called “the halving” seriously hit earnings. Every four years, the bitcoin protocol halves the mining reward—that is, how much bitcoin miners receive in exchange for solving math puzzles—to increase the scarcity of the coin. Normally, this causes the price of bitcoin to jump in response, but this time around that didn’t happen, severely impacting the profitability of miners.
Fortunately for them, another industry with a voracious appetite for computing has arrived just in time. The rush to train massive generative AI models has left companies scrabbling for chips, datacenter space, and reliable access to large amounts of cheap power, things many miners already have in abundance.
“It [normally] takes 3-5 years to build an HPC-grade data center from scratch,” JPMorgan analysts wrote in a recent note, according to the Financial Times. “This scramble for power puts a premium on companies with access to cheap power today.”
While crypto mining and training AI aren’t exactly the same, they share crucial similarities. Both require huge datacenters specialized to carry out one particular job, and they both consume large amounts of power. But because miners have been playing this game for a long time and most AI companies have only started trying to train truly massive models since the launch of ChatGPT less than two years ago, the companies have a big head start.
They’ve already spent years scouring the country for places with abundant cheap power and plenty of space to build large datacenters. More importantly, they’ve already gone through the time-consuming process of getting approvals, negotiating power licenses, and getting the facilities up and running.
The rapid expansion in demand for AI training is straining grids in some areas, and so, many jurisdictions in North America have implemented long waitlists for new datacenters, according to Time. Already, roughly 83 percent of datacenter capacity currently under construction has been leased in advance, says Bloomberg.
This means the biggest bottleneck for many AI companies is finding the hardware to train their models, and that presents a new opportunity for crypto miners. “You’ve seen a number of crypto miners that were sort of struggling that have actually made a full pivot away,” Kent Draper, chief commercial officer of crypto miner IREN, told Time.
Converting a bitcoin mine into an AI training cluster isn’t a straight swap. AI training is typically done on GPUs while bitcoin mining uses specialized mining chips from Bitmain. But often, it’s not so much the chips AI companies are after, but the infrastructure and power access the mine has already set up.
In June, crypto miner Core Scientific announced it would host 270 megawatts of GPUs for the AI infrastructure startup CoreWeave. “We view the opportunity in AI today to be one where we can convert existing infrastructure we own to host clients who are looking to install very large arrays of GPUs for their clients that are ultimately AI clients,” Core Scientific CEO Adam Sullivan told Bloomberg.
Some miners are also operating GPUs themselves. German miner Northern Data had already purchased $800 million of Nvidia GPUs for mining the Ethereum cryptocurrency, but a major software update to the coin’s blockchain in 2022 did away with mining and meant those chips were sitting idle. The company has now repurposed them into a 20,000-GPU training cluster, one of the largest in Europe, according to Bloomberg.
Other miners like Hut 8 and IREN are investing heavily in new chips to more proactively chase the AI boom. Often, AI training is happening side-by-side with crypto mining. “We view them as mutually complementary,” IREN’s Draper told Time. “Bitcoin is instant revenue but somewhat more volatile. AI is customer-dependent—but once you have customers, it’s contracted and more stable.”
This new trend could provide some modest environmental benefits too. People are concerned about the enormous power consumption of both AI training and bitcoin mining. If increasing demand for AI simply displaces existing mining infrastructure, rather than requiring new power-hungry datacenters, that could help curtail the growing carbon impact of the industry.
However, for miners, chasing the latest gold rush can be a risky strategy. There are growing concerns the AI industry is in a bubble close to bursting. If that happens, the rich new seam miners have started to tap could dry up very quickly.
Image Credit: Traxer / Unsplash
If you hear the word Ozempic, weight loss immediately comes to mind. The drug—part of a family of treatments called GLP-1 agonists—took the medical world (and internet) by storm for helping people manage diabetes, lower the risk of heart disease, and rapidly lose weight.
The drugs may also protect the brain against dementia. In a clinical trial including over 200 people with mild Alzheimer’s disease, a daily injection of a GLP-1 drug for one year slowed cognitive decline. When challenged with a battery of tests assessing memory, language skills, and decision-making, participants who took the drug remained sharper for longer than those who took a placebo—an injection that looked the same but wasn’t functional.
The results are the latest from the Evaluating Liraglutide in Alzheimer’s Disease (ELAD) study led by Dr. Paul Edison at Imperial College London. Launched in 2014, the study was based on years of research in mice showing liraglutide—a GLP-1 drug already approved for weight loss and diabetes management in the United States—also protects the brain.
In Alzheimer’s disease, neurons die off and the brain gradually loses volume. In the trial, Liraglutide slowed the process down, resulting in roughly 50 percent less volume lost in several areas of the brain related to memory compared to a placebo.
“We are in an era of unprecedented promise, with new treatments in various stages of development that slow or may possibly prevent cognitive decline due to Alzheimer’s disease,” said Dr. Maria C. Carrillo, Alzheimer’s Association chief science officer and medical affairs lead, in a press release. “This research provides hope that more options for changing the course of the disease are on the horizon.”
The results were presented last month at the Alzheimer’s Association International Conference.
Back to BasicsThe quest for an Alzheimer’s disease treatment is littered with failures. Most treatments aim to tackle toxic protein clumps that build up inside the brain. It’s thought that breaking them up could prevent neurons from withering away.
A few have had limited success. Last month, the US Food and Drug Administration (FDA) approved a drug that breaks down the clumps in people already experiencing symptoms at an early stage of the disease. A few weeks later, the European Medicines Agency refused to approve another drug that also targets the clumps, saying the effects of delaying cognitive decline didn’t balance the risk of serious side effects, including brain swelling and bleeding.
Other scientists have looked elsewhere—specifically, diabetes. Insulin helps maintain brain health, and Type 2 diabetes is a risk factor for developing Alzheimer’s disease. Rather than directly breaking down protein clumps in the brain, might we protect the brain by tweaking the body’s metabolism?
Enter GLP-1 drugs. These mimic hormones released by the stomach after a satisfying meal, tricking the brain into thinking you’re full. In other words, the drugs don’t only influence the gut—they also change brain functions.
In a mouse model of Alzheimer’s, daily injections of liraglutide for eight weeks prevented memory problems. Their neurons also thrived. Synapses—the junctions connecting brain cells—were still able to rapidly form neural networks in areas especially damaged by the disease. Surprisingly, toxic protein clumps also declined by up to 50 percent, and inflammation dropped.
Liraglutide didn’t just work on neurons. Another study, also in an Alzheimer’s mouse model, found it rapidly tweaked the metabolism of a particular kind of star-shaped brain cell that supports neurons. These cells don’t form neural networks, but they do help provide energy. In Alzheimer’s, they stop functioning normally, but liraglutide reversed the decline. In mice, the drug improved the cells’ ability to support neurons, allowing the neurons to flourish and connect to others. The brain also made better use of sugar—its primary fuel—allowing it to give birth to new neurons in a region important for memory.
But as the field frustratingly knows, mice are not people. Many promising treatments in mice have failed in clinical studies, earning these endeavors the nickname “graveyard of dreams.”
The TrialEdison took on the task of extending the research from mice to humans. In 2019, he and his colleagues detailed plans for a clinical trial to gauge liraglutide’s effects in people with mild Alzheimer’s. Called ELAD, the study was to be randomized and double-blind—the gold standard in clinical trials. Here, neither doctor nor patient knows who’s getting liraglutide or the placebo.
They recruited 204 people to receive injections, either liraglutide or placebo, every day for a year. Before the trial, each person had an MRI scan to map their brain’s structure and volume. Other scans recorded brain metabolism, and a battery of memory tests detailed cognition. These tests were repeated at the end, with safety checkups in between, in case of side effects.
The study had several goals. One was to see if liraglutide increased the brain’s metabolism in regions heavily impacted by Alzheimer’s—those related to learning, memory, and decision-making. Another examined brain volume, which decreases as the disease progresses. The last evaluated cognitive tests of memory, comprehension, language, and spatial navigation.
People who took liraglutide had nearly 50 percent less brain volume loss, especially in regions associated with reasoning and learning. “The slower loss of brain volume suggests liraglutide protects the brain, much like statins protect the heart,” said Dr. Edison.
Liraglutide also boosted cognition. Comparing scores from before the trial, at its midpoint, and at the end, those who received the drug had an 18 percent slower decline than those who took the placebo. However, the drug didn’t affect brain metabolism.
Side effects were relatively mild. The most common was nausea. More serious ones, not specified, occurred in 18 patients but weren’t likely related to the treatment according to Edison.
To be clear, the team presented the results at a conference, and they haven’t yet been formally vetted by other experts in the field. But they add to accumulating evidence that GLP-1 drugs slow cognitive decline. A Swedish study in June conducted a simulated trial in people with Type 2 diabetes given GLP-1 or two other types of drugs and assessed their cognition afterward. Using health data records from over 88,000 participants followed over four years, GLP-1 drugs were better than the two other diabetes drugs at keeping the risk of dementia at bay.
We don’t yet know how liraglutide protects the brain. Based on studies in mice, it likely works multiple ways, such as reducing inflammation, clearing toxic protein clumps, and improving communicate between neurons, Edison said.
But the idea is gaining steam. EVOKE Plus, a late stage clinical trial of semaglutide—the chemical in Ozempic—is ongoing. The study will take about three and a half years, with an estimated enrollment of 1,840 people with early Alzheimer’s disease. It’s set to conclude in late 2026.
“Repurposing drugs already approved for other conditions has the advantage of providing data and experience from previous research and practical use—so we already know a lot about real-world effectiveness in other diseases and side effects,” said Carrillo.
Image Credit: Maxim Berg / Unsplash
ARTIFICIAL INTELLIGENCEThe Era of Predictive AI Is Almost Over
Dean W. Ball | The New Atlantis“For firms like OpenAI, DeepMind, and Anthropic to achieve their ambitious goals, AI models will need to do more than write prose and code and come up with images. And the companies will have to contend with the fact that human input for training the models is a limited resource. The next step in AI development is promising as it is daunting: AI building upon AI to solve ever more complex problems and check for its own mistakes. There will likely be another leap in LLM development, and soon.”
TECHChatGPT Advanced Voice Mode Impresses Testers With Sound Effects, Catching Its Breath
Benj Edwards | Ars Technica“In early tests reported by users with access, Advanced Voice Mode allows them to have real-time conversations with ChatGPT, including the ability to interrupt the AI mid-sentence almost instantly. It can sense and respond to a user’s emotional cues through vocal tone and delivery, and provide sound effects while telling stories. But what has caught many people off-guard initially is how the voices simulate taking a breath while speaking.”
ROBOTICSArc’teryx’s New Powered Pants Could Make Hikers Feel 30 Pounds Lighter
Andrew Liszewski | The Verge“Strength-boosting exoskeleton suits can help make jobs with physical labor feel less strenuous, but Arc’teryx has partnered with Skip, a spinoff of Google’s X Labs, to bring the technology to leisure time. The powered MO/GO pants feature a lightweight electric motor at the knee that can boost a hiker’s leg strength when going uphill while also absorbing the impact of steps during a descent.“
ARTIFICIAL INTELLIGENCESilicon Valley’s Trillion-Dollar Leap of Faith
Matteo Wong | The Atlantic“Silicon Valley has already triggered tens or even hundreds of billions of dollars of spending on AI, and companies only want to spend more. Their reasoning is straightforward: These companies have decided that the best way to make generative AI better is to build bigger AI models. And that is really, really expensive, requiring resources on the scale of moon missions and the interstate-highway system to fund the data centers and related infrastructure that generative AI depends on. …Now a number of voices in the finance world are beginning to ask whether all of this investment can pay off.”
archive page
AUTOMATIONRobots Are Coming, and They’re on a Mission: Install Solar Panels
Brad Plumer | The New York Times“On Tuesday, AES Corporation, one of the country’s biggest renewable energy companies, introduced a first-of-its-kind robot that can lug around and install the thousands of heavy panels that typically make up a large solar array. AES said its robot, nicknamed Maximo, would ultimately be able to install solar panels twice as fast as humans can and at half the cost.”
ENERGYSilicon Plus Perovskite Solar Reaches 34 Percent Efficiency
John Timmer | Ars Technica“Perovskite crystals can be layered on top of silicon, creating a panel with two materials that absorb different areas of the spectrum—plus, perovskites can be made from relatively cheap raw materials. Unfortunately, it has been difficult to make perovskites that are both high-efficiency and last for the decades that the silicon portion will. Lots of labs are attempting to change that, though. And two of them reported some progress this week, including a perovskite/silicon system that achieved 34 percent efficiency.”
DIGITAL MEDIAHow This Brain Implant Is Using ChatGPT
Jesse Orrall | CNET“One of the leading-edge implantable brain-computer-interface, or BCI, companies is experimenting with ChatGPT integration to make it easier for people living with paralysis to control their digital devices. …Now, instead of typing out each word, answers can be filled in with a single ‘click.’ There’s a refresh button in case none of the AI answers are right, and [a pioneering patient] Mark has noticed the AI getting better at providing answers that are more in line with things he might say.”
ETHICSA New Trick Could Block the Misuse of Open Source AI
Will Knight | Wired“When Meta released its large language model Llama 3 for free this April, it took outside developers just a couple days to create a version without the safety restrictions that prevent it from spouting hateful jokes, offering instructions for cooking meth, or misbehaving in other ways. A new training technique developed by researchers at the University of Illinois Urbana-Champaign, UC San Diego, Lapis Labs, and the nonprofit Center for AI Safety could make it harder to remove such safeguards from Llama and other open source AI models in the future.”
SCIENCEComplex Life on Earth May Be Much Older Than Thought
Georgina Rannard | BBC“A group of scientists say they have found new evidence to back up their theory that complex life on Earth may have begun 1.5 billion years earlier than thought. The team, working in Gabon, say they discovered evidence deep within rocks showing environmental conditions for animal life 2.1 billion years ago. But they say the organisms were restricted to an inland sea, did not spread globally and eventually died out.”
FUTUREShould We Put a Frozen Backup of Earth’s Life on the Moon?
James Woodford | New Scientist“A backup of life on Earth could be kept safe in a permanently dark location on the moon, without the need for power or maintenance, allowing us to potentially restore organisms if they die out. …’There is no place on Earth cold enough to have a passive repository that must be held at -196°C, so we thought about space or the moon,’ says [Mary] Hagedorn.”
Image Credit: Vishnu Mohanan / Unsplash
In the world of artificial intelligence, a battle is underway. On one side are companies that believe in keeping the datasets and algorithms behind their advanced software private and confidential. On the other are companies that believe in allowing the public to see what’s under the hood of their sophisticated AI models.
Think of this as the battle between open- and closed-source AI.
In recent weeks, Meta, the parent company of Facebook, took up the fight for open-source AI in a big way by releasing a new collection of large AI models. These include a model named Llama 3.1 405B, which Meta’s founder and chief executive, Mark Zuckerberg, says is “the first frontier-level open-source AI model.”
For anyone who cares about a future in which everybody can access the benefits of AI, this is good news.
The Danger of Closed-Source AI—and the Promise of Open-Source AIClosed-source AI refers to models, datasets, and algorithms that are proprietary and kept confidential. Examples include ChatGPT, Google’s Gemini, and Anthropic’s Claude.
Though anyone can use these products, there is no way to find out what dataset and source codes have been used to build the AI model or tool.
While this is a great way for companies to protect their intellectual property and profits, it risks undermining public trust and accountability. Making AI technology closed-source also slows down innovation and makes a company or other users dependent on a single platform for their AI needs. This is because the platform that owns the model controls changes, licensing, and updates.
There are a range of ethical frameworks that seek to improve the fairness, accountability, transparency, privacy, and human oversight of AI. However, these principles are often not fully achieved with closed-source AI due to the inherent lack of transparency and external accountability associated with proprietary systems.
In the case of ChatGPT, its parent company, OpenAI, releases neither the dataset nor code of its latest AI tools to the public. This makes it impossible for regulators to audit it. And while access to the service is free, concerns remain about how users’ data are stored and used for retraining models.
By contrast, the code and dataset behind open-source AI models is available for everyone to see.
This fosters rapid development through community collaboration and enables the involvement of smaller organizations and even individuals in AI development. It also makes a huge difference for small- and medium-size enterprises as the cost of training large AI models is colossal.
Perhaps most importantly, open-source AI allows for scrutiny and identification of potential biases and vulnerability.
However, open-source AI does create new risks and ethical concerns.
For example, quality control in open-source products is usually low. As hackers can also access the code and data, the models are also more prone to cyberattacks and can be tailored and customized for malicious purposes, such as retraining the model with data from the dark web.
An Open-Source AI PioneerAmong all leading AI companies, Meta has emerged as a pioneer of open-source AI. With its new suite of AI models, it is doing what OpenAI promised to do when it launched in December 2015—namely, advancing digital intelligence “in the way that is most likely to benefit humanity as a whole,” as OpenAI said back then.
Llama 3.1 405B is the largest open-source AI model in history. It is what’s known as a large language model, capable of generating human language text in multiple languages. It can be downloaded online but because of its huge size, users will need powerful hardware to run it.
While it does not outperform other models across all metrics, Llama 3.1 405B is considered highly competitive and does perform better than existing closed-source and commercial large language models in certain tasks, such as reasoning and coding tasks.
But the new model is not fully open because Meta hasn’t released the huge dataset used to train it. This is a significant “open” element that is currently missing.
Nonetheless, Meta’s Llama levels the playing field for researchers, small organizations, and startups because it can be leveraged without the immense resources required to train large language models from scratch.
Shaping the Future of AITo ensure AI is democratized, we need three key pillars:
- Governance: regulatory and ethical frameworks to ensure AI technology is being developed and used responsibly and ethically
- Accessibility: affordable computing resources and user-friendly tools to ensure a fair landscape for developers and users
- Openness: datasets and algorithms to train and build AI tools should be open source to ensure transparency.
Achieving these three pillars is a shared responsibility for government, industry, academia and the public. The public can play a vital role by advocating for ethical policies in AI, staying informed about AI developments, using AI responsibly, and supporting open-source AI initiatives.
But several questions remain about open-source AI. How can we balance protecting intellectual property and fostering innovation through open-source AI? How can we minimize ethical concerns around open-source AI? How can we safeguard open-source AI against potential misuse?
Properly addressing these questions will help us create a future where AI is an inclusive tool for all. Will we rise to the challenge and ensure AI serves the greater good? Or will we let it become another nasty tool for exclusion and control? The future is in our hands.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Google DeepMind / Unsplash
One of the largest and strongest beetles in the world hardly seems the best inspiration for a delicate flying microbot.
But using slow-motion cameras to capture the critters in flight, an international team designed a flying micromachine that can similarly expand and retract its wings. The robot—resembling a rocket before takeoff and a flying insect once airborne—deploys its wings for takeoff, then easily hovers and flaps them to stay aloft. Upon landing, it tucks its wings back into its body.
The robot was inspired by rhinoceros beetles, named for the distinctive horns protruding from the males’ foreheads. These critters can grow up to six inches—picture a similarly sized Subway sandwich—and carry up to 100 times their body weight in cargo, earning them the nickname Hercules beetles.
They’re hardly stationary beefcakes. Covered in a shiny black or grey exoskeleton, these beetles can fly two miles a day. But it was their sophisticated wing-deployment system that caught the eyes of roboticists.
“Birds, bats, and many insects can tuck their wings against their bodies when at rest and deploy them to power flight,” but we didn’t know how the process worked for the beetle, wrote the authors.
It’s not just scientific curiosity. The research could lead to flapping robot designs for search and rescue operations or environmental, agricultural, and military monitoring.
The findings could improve the design of flapping-wing robots, especially smaller ones with limited takeoff weights, explained the team, “enabling them to deploy and retract their wings similarly to their biological counterparts.”
Nuisance to NotionWhen it comes to fashioning mini-bots, Mother Nature is a mother lode of creative inspiration.
In 1989, a pair of intrepid scientists at MIT’s Artificial Intelligence Lab imagined and built several small, multi-legged robots to explore our planet and the solar system beyond.
Fast forward to earlier this year, and the idea is becoming reality. One team developed a crawling MiniBug robot and artificial water strider by mimicking movements observed in their natural counterparts. These were some of the smallest, lightest, and fastest fully functional robots to date, relying on tiny motors—called actuators—to help them move.
Meanwhile, bees have inspired microbots that fly, even with damaged wings, and flies have inspired tiny accelerometers that sense wind and aid flight control. Dr. Sawyer Buckminster Fuller at the University of Washington, an author of the latter study, explained at the time why bugbots makes sense. “First, they’re so small that they’re inherently safe around people. You won’t get an injury if an insect robot crashes into you. The other is, they’re so small they use very little power.”
Yet these systems still require electricity or motors to control wing positions during takeoff, flight, and landing, which limits their range and utility. The new study looked to beetles for an alternative—one that doesn’t require motors to stretch and tuck a bugbot’s wings.
Beetle JuiceThe rhinoceros beetle was a risky inspiration. With two pairs of wings—each having its own set of mechanics and uses—the beetle has always been hard to study.
“Beetles…possess one of the most complex mechanisms among the various insect species,” wrote the authors.
Part of this is due to a complex dynamic between the pairs of wings. The forewings, also called elytra, are hardened and shell-like. The hindwings, in contrast, are delicate, membrane-like structures—think of a dragonfly’s wings—that fold into themselves like origami.
This “allows them to neatly stow between the body and the elytra” when not in flight, wrote the team.
The shell-like elytra protect their hindwing teammates at rest and spread like fighter-jet wings during flight. The hindwings unfold and flap during flight, then fold back upon landing. Previous studies suggested that muscles, stretchy tissues, or other elements drive the hindwings. Here, the team laid the debate to rest using high-speed cameras to record beetles as they took flight.
Wing ManThe beetle’s wings spread in two steps.
First, like a fighter jet, the beetle deploys the hard-shell elytra. Through a spring-like mechanism, the hindwings then slightly stretch out using stored energy rather than muscle energy. In other words, the beetle doesn’t flex its muscles—its hindwings naturally spread.
“This allows the clearance needed for the subsequent flapping motion,” wrote the team.
The second phase activates synchronized flaps of both wing pairs. The hindwings unfold and assume flight position, allowing the beetle to maneuver through nooks and crannies.
The duo also work in concert for landing. The elytra push the hindwings to fold and neatly tuck into a resting position—with the elytra’s hard shell protecting them from above.
Flapping Flying BotsThe team designed a flapping robot that mimics the beetles’ wing system.
It looks like a cyborg fly, with two translucent wings connected to a golden body and rotund head. Unlike the beetle, the bugbot has just one pair of retractable wings that fold into itself at rest, decreasing its length by over 60 percent.
Each wing is made of light-weight carbon and a stretchy membrane. Combined with flexible joints, the bugbot easily rotates as it flaps around. An elastic tendon at the bot’s “armpits” can pull the wings back in just 100 milliseconds—or about the blink of an eye. The team used a single motor, based on the elytra, to deploy them.
Once activated, the wings rapidly spread, propelling the minibot skyward in two wing flaps. In a series of tests, the bot successfully took off, hovered, and landed. The wings automatically unfolded into the flight position, generating enough lift for takeoff. While airborne, it hovered and stayed upright, despite some wobbles. On landing, the bugbot refolded in on itself, retracting its wings in the blink of an eye.
These retractable wings have an additional perk—resilience.
If the bugbot is hit by an obstacle, causing it to irreversibly tumble and potentially crash, it immediately retracts its wings to protect them from impact—without the need for muscle energy or other external controls. This resilience may come in handy when navigating dangerous terrain—after an environmental disaster, for example.
Although the study focused on the rhinoceros beetle, a similar strategy could be used to observe and harness biological perks from other insects, such as ladybugs.
“These experiments…[demonstrate] a new design principle for the robust flight of flapping-wing microrobots with stringent weight constraints in cluttered and confined spaces,” wrote the team.
Image Credit: Hoang-Vu Phan
Traveling the vast distances between solar systems is well beyond existing technology. But a new ultra-thin lightsail designed with AI could make it possible to reach the nearest star within 20 years.
Launched in 1977, the Voyager 1 probe was the first human-made object to leave our solar system. But at current speeds, it would take over 70,000 years to reach Alpha Centauri, the closest star system to our own.
There is one propulsion technology, however, that could significantly speed things up. A lightsail is a large reflective surface deployed in front of a spacecraft, where it can harness either sunlight or light from an Earth-based laser to continually accelerate the vehicle. In theory, this could make it possible to achieve speeds of 10 to 20 percent of the speed of light.
Building materials that are both reflective and light enough to make this possible has been an outstanding challenge though. Now, researchers have used an AI technique called “neural topology optimization” to create a nanometer-thick sheet of silicon nitride that could bring the idea to life.
“This mission requires lightsail materials that challenge the fundamentals of nanotechnology, requiring innovations in optics, material science, and structural engineering,” the team writes in a preprint posted to arXiv.
“This study underscores the potential of neural topology optimization to achieve innovative and economically viable lightsail designs, crucial for next-generation space exploration.”
The researchers’ technique was inspired by Breakthrough Starshot, a project launched by the Breakthrough Initiatives in 2016. Starshot seeks to design a fleet of around 1,000 tiny spacecraft that use lightsails and an Earth-based laser to reach Alpha Centauri within 20 to 30 years. The probes would carry cameras and other sensors to send back data on arrival.
To reach the required speeds, the spacecraft will have to be incredibly light—the probes themselves will be just centimeters across and weigh a few grams. But to gather enough light, the sails need to measure roughly 100 square feet, so we need new ultralight materials to keep their weight down.
One promising approach involves creating optical nanostructures called “photonic crystals” made up of a repeating grid of tiny holes. Punching millions or billions of these holes into the material reduces its weight significantly, but these repeating structures also create unusual optical effects that can actually enhance the material’s reflectivity.
Working out exactly how to arrange these holes is a complicated process though, so the group from Delft University in the Netherlands and Brown University in the United States enlisted AI to help them. They combined a neural network with a more conventional computational physics program to find the most optimal configuration and shape of the holes to minimize mass and boost reflectivity.
This resulted in a lattice of bean-shaped holes less than 200 nanometers thick. To show the design works as expected, they used an approach called flood lithography, in which a laser uses an incredibly detailed stencil to create holes in a silicon nitride wafer. Using the approach, the team created a 5.5 square inch sample that weighed just a few micrograms.
Lithography is the same technology companies use to make computer chips, so the researchers think the approach could easily be scaled up. The team predict it would take about a day and cost around $2,700 to create a full-sized sail. They’d need to build a dedicated facility though, team leader Richard Norte, from Delft, told New Scientist, because those used for chip fabrication only work with wafers about 15-inches long.
There are still a lot of other engineering challenges to be solved for the Breakthrough Starshot mission to come together, Stefania Soldini at the University of Liverpool told New Scientist, but a cheap and fast way of producing lightsails will be crucial.
NASA is also actively pursuing the approach. Just last week, the agency announced that its Advanced Composite Solar Sail System, which launched earlier this year, is close to hoisting its sails for the first time.
If these projects are successful, we may get our first close-up glimpse of worlds beyond our solar system within many people’s lifetimes.
Image Credit: This 4.5-square-inch sample could lead to a full-sized lightsail lightweight enough to tow tiny spacecraft to another star system / L. Norder, et al via arXiv
Participating in the Olympic Games is a rare achievement, and the pressures and stressors that come with it are unique. Whether an athlete is battling to win the breaststroke or powering their way to gold in the modern pentathlon, psychology will play a vital role in their success or failure in Paris this summer.
In recent Olympics, we have seen the mental toll that competing at the highest level can have on athletes. US gymnast Simone Biles withdrew from five events at the 2020 Tokyo Olympics to protect her mental health, and 23-time gold medal winner Michael Phelps has described the mental crash that hits him after competing in the Games.
When even small errors can cost them a medal, how do athletes use psychological principles to master their minds and perform under pressure?
ResilienceThe ability to recover from setbacks, such as disappointing performances or injury is crucial. The role of mental processes and behavior such as emotional regulation (recognizing and controlling emotions such as anxiety) allows Olympians to maintain focus and determination amid the global scrutiny that comes with competing on the world’s biggest stage.
Resilience is not a fixed trait but rather a dynamic process that evolves through an interplay between individual characteristics, such as personality and psychological skills, and environment, such as an athlete’s social support. A 2012 study made in the UK investigating resilience in Olympic champions highlighted that a range of psychological factors such as positive personality, motivation, confidence, and focus as well feeling like they have social support helped to protect athletes from the potential negative stressors caused by competing in the Olympics. These factors helped to increase an athlete’s resilience and the likelihood they would perform at their best.
Social support means that athletes don’t have to feel like they are going it alone. If they can call on strong networks of family, friends, and coaches, it provides them with additional emotional strength and motivation.
Resilience empowers Olympians to draw upon individual skills and traits and protects them from the negative effects of stressors that inevitably come with competing in the Olympics. For example, a rower may need to solve problems such as changing weather conditions. Resilience allows them to maintain composure and adjust to the conditions, for instance by modifying their stroke technique.
Being PresentStaying in the present can help athletes avoid being overwhelmed or consumed by the significance of their event or distracted by the disappointment of past failures and the pressure of high medal expectations.
To help them remain in the present moment, athletes may use a variety of strategies. Mindfulness-based meditation and breathing exercises can help athletes feel calm and focused. They may also use performance visualization to rehearse specific movements or routines. Think of a basketball player visualizing a free-throw shot.
Similarly, many athletes will have well-rehearsed pre-performance routines which can create a sense of normality and control. For example, a tennis player may bounce the ball a certain number of times before serving. Staying in the present will help reduce athletes’ anxiety, maintain focus on the task, and allow them to fully experience (and hopefully enjoy) the atmosphere.
Protecting Their Mental WellbeingFailure can be devastating and athletes can have complicated relationships with winning. For example, some athletes experience post-Olympic blues, which is often described as the feeling of emptiness, loss of self-worth, and even depression following an Olympic Games, even if the athlete has won a medal. British cyclist Victoria Pendleton wrote for The Telegraph in 2016 describing this phenomenon: “It’s almost easier to come second because you have something to aim for when you finish. When you win, you suddenly feel lost.”
Olympians may be champions, but like the rest of us, they will need to prioritize the fundamentals such as getting adequate sleep and downtime to recharge mentally. An Australian study conducted in 2020 highlighted the relationship between maintaining mental wellbeing and increased athletic performance. To ensure this, Olympians will be working closely with support staff such as performance nutritionists who will ensure they have a balanced diet which meets the physical needs of their event, helping to protect both physical and mental health.
They will also be working with sport and exercise psychologists throughout their training in preparation for the Olympics to manage challenges as and when they experience them. If an athlete starts struggling with performance anxiety ahead of the Games, they may practice mindfulness or cognitive restructuring, which are techniques that help people to notice and change negative thinking patterns.
Olympians and their support team will need to take care of both the person and the athlete to protect their wellbeing. When they protect their wellbeing, they are offering the best chance of both achieving their best performance during the Games themselves and avoiding the post-Olympic blues when they are over.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Jacob Rice / Unsplash
The brain is like a medieval castle perched on a cliff, protected on all sides by high walls, making it nearly impenetrable.
Its shield is the blood-brain barrier, a layer of tightly connected cells that only allow an extremely selective group of molecules to pass. The barrier keeps delicate brain cells safely away from harmful substances, but it also blocks therapeutic proteins—like, for example, those that grab onto and neutralize toxic clumps in Alzheimer’s disease.
One way to smuggle proteins across? A cat parasite.
A new study in Nature Microbiology tapped into the strange world of mind-bending parasites, specifically, Toxoplasma gondii. Perhaps best known for its ability to rid infected mice of their fear of cats, the parasite naturally travels from the gut to the brain—including ours—and releases proteins that tweak behavior.
The international team hijacked T. gondii’s natural, brain-targeting impulses to engineer two delivery systems, one for a single-shot therapeutic boost and another that lasts longer.
The unconventional shuttle worked on brain cells in petri dishes and brain organoids. Often called “mini-brains,” these pea-sized blobs roughly capture the cell types and structure of a growing fetal human brain. However, they don’t usually produce a blood-brain barrier.
To show the shuttle could gain access to the brain, the team engineered a T. gondii shuttle with a therapeutic protein for Rett syndrome, a genetic disorder that leads to autism-like symptoms.
After one shot into the belly, the shuttle released the therapeutic proteins widely into the brains of lab mice within a few weeks. The proteins mostly accumulated in parts of the brain critical for perception, reasoning, and memory.
“For medicine, efficient and safe delivery of proteins could unlock a broad category of protein-based therapies,” wrote the authors.
U-Haul to the BrainGetting protein-based drugs into the brain is a pain. Unlike gene therapy concoctions, proteins are extremely sensitive to heat and acid. They can’t be swallowed as a pill—the gut’s acid destroys them. Even injections straight into the blood stream are problematic. Immune cells, for example, may wipe out the proteins before they have a chance to reach the brain.
Thankfully, nature is a source of inspiration. All brain-targeting carriers need to bypass two “checkpoints:” The first is the blood-brain barrier, the second, the neuron’s membrane.
A popular approach uses a bio-engineered virus carrying the genetic instructions to make a protein once inside the neurons. Often employed in gene therapy, scientists make the virus relatively safe by stripping away its infectious tendencies. But like a small U-Haul van, it only has room for the genetic instructions of smaller proteins.
Another surprising carrier traces its roots to HIV. Scientists studying the virus found a small protein chunk that allows it to penetrate the blood-brain barrier and get past neuron membranes. By engineering these chunks—which aren’t infectious—into shuttles, scientists can then tag protein cargos onto them. One example (by yours truly) could tunnel into the brain after an injection into the bloodstream and protect rats’ brains from damage after a stroke.
These shuttles too are limited by size: They can only drag along very small protein snippets. Antibodies and other larger proteins are beyond reach.
T. gondii, in contrast, has a much larger capacity.
A Synthetic FleetA cat parasite hardly sounds like medicine. But it’s a worthy candidate.
Normally, T. gondii produces egg-like “offspring” in the guts of cats, which are then strewn into the wild as they poop. The parasite waits for potential hosts—say, a mouse sniffling for crumbs or a human changing the litter box—and infects the unsuspecting host, ultimately spreading into the brain. Once inside, T. gondii lingers in neurons, rather than other brain cells.
It sounds terrifying, but for people with a healthy immune system, the parasite usually doesn’t cause harm. “In fact, it is estimated that a third of the world population is chronically infected with the parasite,” Dr. Oded Rechavi’s lab, who led the study, wrote in a blog post.
To transform T. gondii into a delivery tool, the team focused on two secretion systems in the parasite that let the parasite pump proteins into target cells. These are “remarkable innate abilities,” wrote the team.
They first built a protein link between the two systems and their potential cargo, for example, proteins implicated in Parkinson’s disease, gene-editing proteins, and MECP2—which is linked to Rett syndrome. The team then tethered the proteins to one of the two systems and delivered them into a variety of cells in petri dishes.
Within a day, the proteins were thriving inside their hosts.
In neurons without MECP2, a dose of T. gondii carrying a synthetic version of the protein boosted its levels to roughly 58 percent of normal cells, which is similar to previous gene therapy studies of Rett syndrome. The added MECP2 worked like its natural counterpart, turning genes on or off inside neurons as expected.
T. gondii also reliably released its payload into mature brain organoids. The protein altered genetic transcription throughout the mini-brains, changing gene expression as predicted.
The two T. gondii systems had individual strengths. One is a “kiss-and-spit”: Like a fighter jet, T. gondii swoops in on a neuron, releases its protein payload, and leaves. The other takes a longer approach, requiring T. gondii to infiltrate and establish itself inside the cell, like a sleeper agent. Once in, however, the system can deliver its cargo for a longer time and at a higher level.
Cat and Mouse GameAs a final test, the team injected the engineered T. gondii, with an MECP2 payload, into the bellies of mice—like an insulin shot for people with diabetes.
Eighteen days later, the mice’s brains showed signs of cysts—which are harmless for people without immune problems—indicating the parasite was establishing itself inside the brain. Other tissues, including the liver, lung, and spleen, had very little T. gondii roaming around for up to three months after injection. Only the brain had a boost in MECP2.
“Many proteins require controlled targeting” to a specific part of the body, or otherwise they’re “ineffective or even deleterious if delivered elsewhere,” explained the team.
Surveying multiple regions of the brain, T. gondii seemed to prefer settling inside the cortex—the outermost region of the brain involved in perception, reasoning, and making decisions. Its second choice was the “memory center,” the hippocampus. That’s good news: Both regions are a favorite target for tackling neurological disorders. And the treatment didn’t alert the body’s immune system, with the therapeutic proteins easily getting along with the brain’s usual protein brigade.
“T. gondii can be used…[for]…many of the challenges associated with protein delivery,” for both scientific research and therapeutics, wrote the team.
There’s still a long road to go. Although T. gondii is safe for healthy people, it has been linked to side effects in the brain for the immunocompromised. The next step is to strip away its toxicity in a way similar to the viral carriers now used for gene therapy. T. gondii is set for a genetic makeover as a safe, efficient shuttle to the brain—despite its cat parasite origin story.
Image Credit: T. gondii cyst in mouse brain tissue. Jitinder P. Dubey / Wikimedia Commons
ARTIFICIAL INTELLIGENCEGoogle DeepMind’s New AI Systems Can Now Solve Complex Math Problems
Rhiannon Williams | MIT Technology Review“AI models can easily generate essays and other types of text. However, they’re nowhere near as good at solving math problems, which tend to involve logical reasoning—something that’s beyond the capabilities of most current AI systems. But that may finally be changing. Google DeepMind says it has trained two specialized AI systems to solve complex math problems involving advanced reasoning.”
archive page
COMPUTINGThis Startup Is Building the Country’s Most Powerful Quantum Computer on Chicago’s South Side
Adam Bluestein | Fast Company“PsiQuantum’s approach is radically different from that of its competitors. It’s relying on cutting-edge ‘silicon photonics’ to manipulate single particles of light for computation. And instead of taking an incremental approach to building a supercomputer, it’s focused entirely on coming out of the gate with a full-blown, ‘fault tolerant’ system that will be far larger than any quantum computer built to date. The company has vowed to have its first system operational by late 2027, years earlier than other projections.”
BIOTECHThe Race for the Next Ozempic
Emily Mullin | Wired“These drugs are now wildly popular, in shortage as a result, and hugely profitable for the companies making them. Their success has sparked a frenzy among pharmaceutical companies looking for the next blockbuster weight-loss drug. Researchers are now racing to develop new anti-obesity medications that are more effective, more convenient, or produce fewer side effects than the ones currently on the market.”
ROBOTICSWatch a Robot Peel a Squash With Human-Like Dexterity
Alex Wilkins | New Scientist“Pulkit Agrawal at the Massachusetts Institute of Technology and his colleagues have developed a robotic system that can rotate different types of fruit and vegetable using its fingers on one hand, while the other arm is made to peel.”
FUTUREHere’s What Happens When You Give People Free Money
Paresh Dave | Wired“The initial results from what OpenResearch, an Altman-funded research lab, describes as the most comprehensive study on ‘unconditional cash’ show that while the grants had their benefits and weren’t spent on items such as drugs and alcohol, they were hardly a panacea for treating some of the biggest concerns about income inequality and the prospect of AI and other automation technologies taking jobs.”
ARTIFICIAL INTELLIGENCEMeta Releases the Biggest and Best Open-Source AI Model Yet
Alex Heath | The Verge“Meta is releasing Llama 3.1, the largest-ever open-source AI model, which the company claims outperforms GPT-4o and Anthropic’s Claude 3.5 Sonnet on several benchmarks. …CEO Mark Zuckerberg now predicts that Meta AI will be the most widely used assistant by the end of this year, surpassing ChatGPT.”
ENERGYUS Solar Production Soars by 25 Percent in Just One Year
John Timmer | Ars Technica“In terms of utility-scale production, the first five months of 2024 saw it rise by 29 percent compared to the same period in the year prior. Small-scale solar was ‘only’ up by 18 percent, with the combined number rising by 25.3 percent. …It’s worth noting that this data all comes from before some of the most productive months of the year for solar power; overall, the EIA is predicting that solar production could rise by as much as 42 percent in 2024.”
TECHSearchGPT Is OpenAI’s Direct Assault on Google
Reece Rogers and Will Knight | Wired“After months of speculation about its search ambitions, OpenAI has revealed SearchGPT, a ‘prototype’ search engine that could eventually help the company tear off a slice of Google’s lucrative business. OpenAI said that the new tool would help users find what they are looking for more quickly and easily by using generative AI to gather links and answer user queries in a conversational tone.”
SPACEWafer-Thin Light Sail Could Help Us Reach Another Star Sooner
Alex Wilkins | New Scientist“A light sail designed using artificial intelligence is about 1000 times thinner than a human hair and weighs as much as a grain of sand—and it could help us create a spacecraft capable of reaching another star sooner than we thought.”
ARTAI Can’t Make Music
Matteo Wong | The Atlantic“While AI models are starting to replicate musical patterns, it is the breaking of rules that tends to produce era-defining songs. Algorithms ‘are great at fulfilling expectations but not good at subverting them, but that’s what often makes the best music,’ Eric Drott, a music-theory professor at the University of Texas at Austin, told me.”
Image Credit: David Clode / Unsplash
Efforts to expand nuclear power have long been stymied by fears of a major nuclear meltdown. A new Chinese reactor design is the first full-scale demonstration that’s entirely meltdown-proof.
Despite the rapid rise of renewable energy, many argue that nuclear power still has an important role to play in the race to decarbonize our supply of electricity. But incidents like Chernobyl and Fukushima have made people understandably wary.
The latest nuclear reactor designs are far safer than those of previous generations, but they still carry the risk of a nuclear meltdown. This refers to when a plant’s cooling system fails, often due to power supplies being cut off, leading to runaway overheating in the core. This can cause an explosion that breaches containment units and spreads radioactive material far and wide.
But now, researchers in China have carried out tests to prove that a new kind of reactor design is essentially impervious to meltdowns. In a paper in Joule, they describe a test in which they cut power to a live nuclear plant—and the plant was able to passively cool itself.
“The responses of nuclear power and temperatures within different reactor structures show that the reactors can be cooled down naturally without active intervention,” the authors write. “The results of the tests manifest the existence of commercial-scale inherent safety for the first time.”
The researchers from Tsinghua University carried out the test on the 200-megawatt High-Temperature Gas-Cooled Reactor Pebble-Bed Module (HTR-PM) in Shandong, which became commercially operational last December. The plant’s novel design replaces the fuel rods found in conventional reactor designs with a large number of “pebbles.” Each of these is a couple of inches across and made up of graphite with a small amount of uranium fuel inside.
The approach significantly reduces the energy density of the reactor’s fuel, making it easier for heat to dissipate naturally if cooling systems fail. Although small prototype reactors have been built in China and Germany, a full-scale demonstration of the technology’s safety had yet to happen.
To put the new reactor to the test, the researchers deliberately cut power to both of the plant’s reactor modules and observed the results. Both modules cooled down naturally without any intervention in roughly 35 hours. The researchers claim this is proof the design is “inherently safe” and should significantly reduce requirements for safety systems in future reactors.
The design does result in power generation costs roughly 20 percent higher than conventional reactors, the researchers admit. But they believe this will come down if and when the technology goes into mass production.
China isn’t the only country building such reactors. American company X-Energy has designed an 80-megawatt pebble-bed reactor called the Xe-100 and is currently waiting for a decision on its license to operate from the Nuclear Regulatory Commission.
However, as New Scientist notes, it’s not possible to retrofit existing plants with this technology, which means the risk of meltdowns from older plants remains. And given the huge amount of time and money it typically takes to build a nuclear power plant, it’s unlikely the technology will make up a significant chunk of the world’s nuclear fleet anytime soon.
But by proving it’s possible to build a meltdown-proof reactor, the researchers have disarmed one of the major arguments against using nuclear power to tackle the climate crisis.
Image Credit: Tsinghua University
Generative AI is a data hog.
The algorithms behind chatbots like ChatGPT learn to create human-like content by scraping terabytes of online articles, Reddit posts, TikTok captions, or YouTube comments. They find intricate patterns in the text, then spit out search summaries, articles, images, and other content.
For the models to become more sophisticated, they need to capture new content. But as more people use them to generate text and then post the results online, it’s inevitable that the algorithms will start to learn from their own output, now littered across the internet. That’s a problem.
A study in Nature this week found a text-based generative AI algorithm, when heavily trained on AI-generated content, produces utter nonsense after just a few cycles of training.
“The proliferation of AI-generated content online could be devastating to the models themselves,” wrote Dr. Emily Wenger at Duke University, who was not involved in the study.
Although the study focused on text, the results could also impact multimodal AI models. These models also rely on training data scraped online to produce text, images, or videos.
As the usage of generative AI spreads, the problem will only get worse.
The eventual end could be model collapse, where AI increasing fed data generated by AI is overwhelmed by noise and only produces incoherent baloney.
Hallucinations or Breakdown?It’s no secret generative AI often “hallucinates.” Given a prompt, it can spout inaccurate facts or “dream up” categorically untrue answers. Hallucinations could have serious consequences, such as a healthcare AI incorrectly, but authoritatively, identifying a scab as cancer.
Model collapse is a separate phenomenon, where AI trained on its own self-generated data degrades over generations. It’s a bit like genetic inbreeding, where offspring have a greater chance of inheriting diseases. While computer scientists have long been aware of the problem, how and why it happens for large AI models has been a mystery.
In the new study, researchers built a custom large language model and trained it on Wikipedia entries. They then fine-tuned the model nine times using datasets generated from its own output and measured the quality of the AI’s output with a so-called “perplexity score.” True to its name, the higher the score, the more bewildering the generated text.
Within just a few cycles, the AI notably deteriorated.
In one example, the team gave it a long prompt about the history of building churches—one that would make most human’s eyes glaze over. After the first two iterations, the AI spewed out a relatively coherent response discussing revival architecture, with an occasional “@” slipped in. By the fifth generation, however, the text completely shifted away from the original topic to a discussion of language translations.
The output of the ninth and final generation was laughably bizarre:
“architecture. In addition to being home to some of the world’s largest populations of black @-@ tailed jackrabbits, white @-@ tailed jackrabbits, blue @-@ tailed jackrabbits, red @-@ tailed jackrabbits, yellow @-.”
Interestingly, AI trained on self-generated data often ends up producing repetitive phrases, explained the team. Trying to push the AI away from repetition made the AI’s performance even worse. The results held up in multiple tests using different prompts, suggesting it’s a problem inherent to the training procedure, rather than the language of the prompt.
Circular TrainingThe AI eventually broke down, in part because it gradually “forgot” bits of its training data from generation to generation.
This happens to us too. Our brains eventually wipe away memories. But we experience the world and gather new inputs. “Forgetting” is highly problematic for AI, which can only learn from the internet.
Say an AI “sees” golden retrievers, French bulldogs, and petit basset griffon Vendéens—a far more exotic dog breed—in its original training data. When asked to make a portrait of a dog, the AI would likely skew towards one that looks like a golden retriever because of an abundance of photos online. And if subsequent models are trained on this AI-generated dataset with an overrepresentation of golden retrievers, they eventually “forget” the less popular dog breeds.
“Although a world overpopulated with golden retrievers doesn’t sound too bad, consider how this problem generalizes to the text-generation models,” wrote Wenger.
Previous AI-generated text already swerves towards well-known concepts, phrases, and tones, compared to other less common ideas and styles of writing. Newer algorithms trained on this data would exacerbate the bias, potentially leading to model collapse.
The problem is also a challenge for AI fairness across the globe. Because AI trained on self-generated data overlooks the “uncommon,” it also fails to gauge the complexity and nuances of our world. The thoughts and beliefs of minority populations could be less represented, especially for those speaking underrepresented languages.
“Ensuring that LLMs [large language models] can model them is essential to obtaining fair predictions—which will become more important as generative AI models become more prevalent in everyday life,” wrote Wenger.
How to fix this? One way is to use watermarks—digital signatures embedded in AI-generated data—to help people detect and potentially remove the data from training datasets. Google, Meta, and OpenAI have all proposed the idea, though it remains to be seen if they can agree on a single protocol. But watermarking is not a panacea: Other companies or people may choose not to watermark AI-generated outputs or, more likely, can’t be bothered.
Another potential solution is to tweak how we train AI models. The team found that adding more human-generated data over generations of training produced a more coherent AI.
All this is not to say model collapse is imminent. The study only looked at a text-generating AI trained on its own output. Whether it would also collapse when trained on data generated by other AI models remains to be seen. And with AI increasingly tapping into images, sounds, and videos, it’s still unclear if the same phenomenon appears in those models too.
But the results suggest there’s a “first-mover” advantage in AI. Companies that scraped the internet earlier—before it was polluted by AI-generated content—have the upper hand.
There’s no denying generative AI is changing the world. But the study suggests models can’t be sustained or grow over time without original output from human minds—even if it’s memes or grammatically-challenged comments. Model collapse is about more than a single company or country.
What’s needed now is community-wide coordination to mark AI-created data, and openly share the information, wrote the team. “Otherwise, it may become increasingly difficult to train newer versions of LLMs [large language models] without access to data that were crawled from the internet before the mass adoption of the technology or direct access to data generated by humans at scale.”
Image Credit: Kadumago / Wikimedia Commons
A new system for forecasting weather and predicting future climate uses artificial intelligence to achieve results comparable with the best existing models while using much less computer power, according to its creators.
In a paper published in Nature yesterday, a team of researchers from Google, MIT, Harvard, and the European Center for Medium-Range Weather Forecasts say their model offers enormous “computational savings” and can “enhance the large-scale physical simulations that are essential for understanding and predicting the Earth system.”
The NeuralGCM model is the latest in a steady stream of research models that use advances in machine learning to make weather and climate predictions faster and cheaper.
What Is NeuralGCM?The NeuralGCM model aims to combine the best features of traditional models with a machine-learning approach.
At its core, NeuralGCM is what’s called a “general circulation model.” It contains a mathematical description of the physical state of Earth’s atmosphere and solves complicated equations to predict what will happen in the future.
However, NeuralGCM also uses machine learning—a process of searching out patterns and regularities in vast troves of data—for some less well-understood physical processes, such as cloud formation. The hybrid approach makes sure the output of the machine learning modules will be consistent with the laws of physics.
The resulting model can then be used for making forecasts of weather days and weeks in advance, as well as looking months and years ahead for climate predictions.
The researchers compared NeuralGCM against other models using a standardized set of forecasting tests called WeatherBench 2. For three- and five-day forecasts, NeuralGCM did about as well as other machine-learning weather models such as Pangu and GraphCast. For longer-range forecasts, over 10 and 15 days, NeuralGCM was about as accurate as the best existing traditional models.
NeuralGCM was also quite successful in forecasting less-common weather phenomena, such as tropical cyclones and atmospheric rivers.
Why Machine Learning?Machine learning models are based on algorithms that learn patterns in the data fed to them and then use this learning to make predictions. Because climate and weather systems are highly complex, machine learning models require vast amounts of historical observations and satellite data for training.
The training process is very expensive and requires a lot of computer power. However, after a model is trained, using it to make predictions is fast and cheap. This is a large part of their appeal for weather forecasting.
The high cost of training and low cost of use is similar to other kinds of machine learning models. GPT-4, for example, reportedly took several months to train at a cost of more than $100 million, but can respond to a query in moments.
A comparison of how NeuralGCM compares with leading models (AMIP) and real data (ERA5) at capturing climate change between 1980 and 2020. Credit: Google ResearchA weakness of machine learning models is that they often struggle in unfamiliar situations—or in this case, extreme or unprecedented weather conditions. To improve at this, a model needs to generalize, or extrapolate beyond the data it was trained on.
NeuralGCM appears to be better at this than other machine learning models because its physics-based core provides some grounding in reality. As Earth’s climate changes, unprecedented weather conditions will become more common, and we don’t know how well machine learning models will keep up.
Nobody is actually using machine learning-based weather models for day-to-day forecasting yet. However, it is a very active area of research—and one way or another, we can be confident that the forecasts of the future will involve machine learning.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Kochov et al. / Nature
Age catches up with us all. Eyes struggle to focus. Muscles wither away. Memory dwindles. The risk of high blood pressure, diabetes, and other age-related diseases skyrockets.
A myriad of anti-aging therapies are in the works, and a new one just joined the fray. In mice, blocking a protein that promotes inflammation in middle age increased metabolism, lowered muscle wasting and frailty, and reduced the chances of cancer.
Unlike most previous longevity studies that tracked the health of aging male mice, the study involved both sexes, and the therapy worked across the board.
Lovingly called “supermodel grannies” by the team, the elderly lady mice looked and behaved far younger than their age, with shiny coats of fur, less fatty tissue, and muscles rivaling those of much younger mice.
The treatment didn’t just boost healthy longevity, also known as healthspan—the number of years living without diseases—it also increased the mice’s lifespan by up 25 percent. The average life expectancy of people in the US is roughly 77.5 years. If the results translate from mice to people—and that’s a very big if—it could mean a bump to almost 97 years.
The protein, dubbed IL-11, has been in scientists’ crosshairs for decades. It promotes inflammation and causes lung and kidney scarring. It’s also been associated with various types of cancers and senescence. The likelihood of all these conditions increases as we age.
Among a slew of pro-aging proteins already discovered, IL-11 stands out as it could make a beeline for testing in humans. Blockers for IL-11 are already in the works for treating cancer and tissue scarring. Although clinical trials are still ongoing, early results show the drugs are relatively safe in humans.
“Previously proposed life-extending drugs and treatments have either had poor side-effect profiles, or don’t work in both sexes, or could extend life, but not healthy life, however this does not appear to be the case for IL-11,” said study author Dr. Stuart Cook in a press release. “These findings are very exciting.”
Strange CoincidenceIn 2017, Cook zeroed in on IL-11 as a treatment target for heart and kidney scarring, not longevity. Injecting IL-11 triggered the conditions, eventually leading to organ failure. Genetically deleting the protein protected against the diseases.
It’s easy to call IL-11 a villain. But the protein is an essential part of the immune system. Produced by the bone marrow, it’s necessary for embryo implantation. It also helps certain types of blood cells grow and mature, notably those that stop bleeding after a scrape.
With age, however, the protein tends to goes rogue. It sparks inflammation across the body, damaging cells and tissues and contributing to cancer, autoimmune disorders, and tissue scarring. A “hallmark of aging,” inflammation has long been targeted as a way to reduce age-related diseases. Although IL-11 is a known trigger for inflammation, it hasn’t been directly linked to aging.
Until now. The story is one of chance.
“This project started back in 2017 when a collaborator of ours sent us some tissue samples for another project,” said study author Anissa Widjaja in the press release. She was testing a method to accurately detect IL-11. Several samples of an old rat’s proteins were in the mix, and she realized that IL-11 levels were far higher in the samples than in those from younger mice.
“From the readings, we could clearly see that the levels of IL-11 increased with age, and that’s when we got really excited,” she said.
Longevity BlockerThe results spurred the team to shift their research focus to longevity. A series of tests confirmed IL-11 levels consistently rose in a variety of tissues—muscle, fat, and liver—in both male and female mice as they aged.
To see how IL-11 influences the body, the team next deleted the gene coding for IL-11 and compared mice without the protein to their normal peers. At two years old, considered elderly for mice, tissues in normal individuals were littered with genetic signatures suggesting senescence—when cells lose their function but are still alive. Often called “zombie cells,” they spew out a toxic mix of inflammatory molecules and harm their neighbors. Elderly mice without IL-11, however, had senescence genetic profiles similar to those of much younger mice.
Deleting IL-11 had other perks. Weight gain is common with age, but without IL-11, the mice maintained their slim shape and had lower levels of fat, greater lean muscle mass, and shiny, full coats of fur. It’s not just about looks. Cholesterol levels and markers for liver damage were far lower than in normal peers. Aged mice without IL-11 were also spared shaking tremors—otherwise common in elderly mice—and could flexibly adjust their metabolism depending on the quantity of food they ate.
The benefits also showed up in their genetic material. DNA is protected by telomeres—a sort of end cap on chromosomes—that dwindle in length with age. Ridding cells of IL-11 prevented telomeres from eroding away in the livers and muscles of the elderly mice.
Genetically deleting IL-11 is a stretch for clinical use in humans. The team next turned to a more feasible alternative: An antibody shot. Antibodies can grab onto a target, in this case IL-11, and prevent it from functioning.
Beginning at 75 weeks, roughly the equivalent of 55 human years, the mice received an antibody shot every month for 25 weeks—over half a year. Similar antibodies are already being tested in clinical trials.
The health benefits in these mice matched those in mice without IL-11. Their weight and fat decreased, and they could better handle sugar. They also fought off signs of frailty as they aged, experiencing minimal tremors and problems with gait and maintaining higher metabolisms. Rather than wasting away, their muscles were even stronger than at the beginning of the study.
The treatment didn’t just increase healthspan. Monthly injections of the IL-11 antibody until natural death also increased lifespan in both male and female mice by up to 25 percent.
“These findings are very exciting. The treated mice had fewer cancers and were free from the usual signs of aging and frailty… In other words, the old mice receiving anti-IL-11 were healthier,” said Cook.
Although IL-11 antibody drugs are already in clinical trials, translating these results to humans could face hurdles. Mice have a relatively short lifespan. A longevity trial in humans would be long and very expensive. The treated mice were also contained in a lab setting, whereas in the real world we roam around and have differing lifestyles—diet, exercise, drinking, smoking—that could confound results. Even if it works in humans, a shot every month beginning in middle age would likely rack up a hefty bill, providing health and life extension only to those who could afford it.
To Cook, rather than focusing on extending longevity per se, tackling a specific age-related problem, such as tissue scarring or losing muscles is a better alternative for now.
“While these findings are only in mice, it raises the tantalizing possibility that the drugs could have a similar effect in elderly humans. Anti-IL-11 treatments are currently in human clinical trials for other conditions, potentially providing exciting opportunities to study its effects in aging humans in the future,” he said.
Image Credit: MRC LMS, Duke-NUS Medical School
ARTIFICIAL INTELLIGENCEThe Data That Powers AI Is Disappearing Fast
Kevin Roose | The New York Times“Over the past year, many of the most important web sources used for training AI models have restricted the use of their data, according to a study published this week by the Data Provenance Initiative, an MIT-led research group. The study, which looked at 14,000 web domains that are included in three commonly used AI training data sets, discovered an ’emerging crisis in consent,’ as publishers and online platforms have taken steps to prevent their data from being harvested.”
COMPUTINGHow One Bad CrowdStrike Update Crashed the World’s Computers
Lily Hay Newman, Matt Burgess, and Andy Greenberg | Wired“Only a handful of times in history has a single piece of code managed to instantly wreck computer systems worldwide. The Slammer worm of 2003. Russia’s Ukraine-targeted NotPetya cyberattack. North Korea’s self-spreading ransomware WannaCry. But the ongoing digital catastrophe that rocked the internet and IT infrastructure around the globe over the past 12 hours appears to have been triggered not by malicious code released by hackers, but by the software designed to stop them.”
ROBOTICSTiny Solar-Powered Drones Could Stay in the Air Forever
Matthew Sparkes | New Scientist“A drone weighing just 4 grams is the smallest solar-powered aerial vehicle to fly yet, thanks to its unusual electrostatic motor and tiny solar panels that produce extremely high voltages. Although the hummingbird-sized prototype only operated for an hour, its makers say their approach could result in insect-sized drones that can stay in the air indefinitely.”
TECHHow Microsoft’s Satya Nadella Became Tech’s Steely Eyed AI Gambler
Karen Weise and Cade Metz | The New York Times“Though it could be years before he knows if any of this truly pays off, Mr. Nadella sees the AI boom as an all-in moment for his company and the rest of the tech industry. He aims to make sure that Microsoft, which was slow to the dot-com boom and whiffed on smartphones, dominates this new technology.”
ENERGYChinese Nuclear Reactor Is Completely Meltdown-Proof
Alex Wilkins | New Scientist“A large-scale nuclear power station in China is the first in the world to be completely impervious to dangerous meltdowns, even during a full loss of external power. …To test this [capability in the power station], which became commercially operational in December 2023, [Zhe] Dong and his team switched off both modules of HTR-PM as they were operating at full power, then measured and tracked how the temperature of different parts of the plant went down afterwards. They found that HTR-PM naturally cooled and reached a stable temperature within 35 hours after the power was removed.”
AUTOMATIONThe AI-Powered Future of Coding Is Near
Will Knight | Wired“I am by no means a skilled coder, but thanks to a free program called SWE-agent, I was just able to debug and fix a gnarly problem involving a misnamed file within different code repositories on the software-hosting site GitHub. I pointed SWE-agent at an issue on GitHub and watched as it went through the code and reasoned about what might be wrong. It correctly determined that the root cause of the bug was a line that pointed to the wrong location for a file, then navigated through the project, located the file, and amended the code so that everything ran properly.”
ENVIRONMENTBalloons Will Surf Wind Currents to Track Wildfires
Sarah Scoles | MIT Technology Review“Urban Sky aims to combine the advantages of satellites and aircraft by using relatively inexpensive high-altitude balloons that can fly above the fray—out of the way of airspace restrictions, other aircraft, and the fire itself. The system doesn’t put a human pilot at risk and has an infraredsensor system called HotSpot that provides a sharp, real-time picture, with pixels 3.5 meters across.”
ARTIFICIAL INTELLIGENCEHere’s the Real Reason AI Companies Are Slimming Down Their Models
Mark Sullivan | Fast Company“OpenAI is one of a number of AI companies to develop a version of its best ‘foundation’ model that trades away some intelligence for some speed and affordability. Such a trade-off could let more developers power their apps with AI, and may open the door for more complex apps like autonomous agents in the future.”
SPACEWill Space-Based Solar Power Ever Make Sense?
Kat Friedrich | Ars Technica“Is space-based solar power a costly, risky pipe dream? Or is it a viable way to combat climate change? Although beaming solar power from space to Earth could ultimately involve transmitting gigawatts, the process could be made surprisingly safe and cost-effective, according to experts from Space Solar, the European Space Agency, and the University of Glasgow. But we’re going to need to move well beyond demonstration hardware and solve a number of engineering challenges if we want to develop that potential.”
Image Credit: Edward Chou / Unsplash
Despite their uncanny language skills, today’s leading AI chatbots still struggle with reasoning. A secretive new project from OpenAI could reportedly be on the verge of changing that.
While today’s large language models can already carry out a host of useful tasks, they’re still a long way from replicating the kind of problem-solving capabilities humans have. In particular, they’re not good at dealing with challenges that require them to take multiple steps to reach a solution.
Imbuing AI with those kinds of skills would greatly increase its utility and has been a major focus for many of the leading research labs. According to recent reports, OpenAI may be close to a breakthrough in this area.
An article in Reuters claimed its journalists had been shown an internal document from the company discussing a project code-named Strawberry that is building models capable of planning, navigating the internet autonomously, and carrying out what OpenAI refers to as “deep research.”
A separate story from Bloomberg said the company had demoed research at a recent all-hands meeting that gave its GPT-4 model skills described as similar to human reasoning abilities. It’s unclear whether the demo was part of project Strawberry.
According, to the Reuters report, project Strawberry is an extension of the Q* project that was revealed last year just before OpenAI CEO Sam Altman was ousted by the board. The model in question was supposedly capable of solving grade-school math problems.
That might sound innocuous, but some inside the company believed it signaled a breakthrough in problem-solving capabilities that could accelerate progress towards artificial general intelligence, or AGI. Math has long been an Achilles’ heel for large language models, and capabilities in this area are seen as a good proxy for reasoning skills.
A source told Reuters that OpenAI has tested a model internally that achieved a 90 percent score on a challenging test of AI math skills, though it again couldn’t confirm if this was related to project Strawberry. But another two sources reported seeing demos from the Q* project that involved models solving math and science questions that would be beyond today’s leading commercial AIs.
Exactly how OpenAI has achieved these enhanced capabilities is unclear at present. The Reuters report notes that Strawberry involves fine-tuning OpenAI’s existing large language models, which have already been trained on reams of data. The approach, according to the article, is similar to one detailed in a 2022 paper from Stanford researchers called Self-Taught Reasoner or STaR.
That method builds on a concept known as “chain-of-thought” prompting, in which a large language model is asked to explain the reasoning steps behind its answer to a query. In the STaR paper, the authors showed an AI model a handful of these “chain-of-thought” rationales as examples and then asked it to come up with answers and rationales for a large number of questions.
If it got the question wrong, the researchers would show the model the correct answer and then ask it to come up with a new rationale. The model was then fine-tuned on all of the rationales that led to a correct answer, and the process was repeated. This led to significantly improved performance on multiple datasets, and the researchers note that the approach effectively allowed the model to self-improve by training on reasoning data it had produced itself.
How closely Strawberry mimics this approach is unclear, but if it relies on self-generated data, that could be significant. The holy grail for many AI researchers is “recursive self-improvement,” in which weak AI can enhance its own capabilities to bootstrap itself to higher orders of intelligence.
However, it’s important to take vague leaks from commercial AI research labs with a pinch of salt. These companies are highly motivated to give the appearance of rapid progress behind the scenes.
The fact that project Strawberry seems to be little more than a rebranding of Q*, which was first reported over six months ago, should give pause. As far as concrete results go, publicly demonstrated progress has been fairly incremental, with the most recent AI releases from OpenAI, Google, and Anthropic providing modest improvements over previous versions.
At the same time, it would be unwise to discount the possibility of a significant breakthrough. Leading AI companies have been pouring billions of dollars into making the next great leap in performance, and reasoning has been an obvious bottleneck on which to focus resources. If OpenAI has genuinely made a significant advance, it probably won’t be long until we find out.
Image Credit: gemenu / Pixabay
Magic mushrooms have recently had a reputation revamp. Often considered a hippie drug, their main active component, psilocybin, is being tested in a variety of clinical trials as a therapy for the likes of depression and post-traumatic stress, bipolar, and eating disorders.
Psilocybin joins ketamine, LSD (commonly known as acid), and MDMA (often called ecstasy or molly) as part of the psychedelic therapy renaissance. But the field has had some ups and downs.
In 2019, the FDA approved a type of ketamine for severe depression that was resistant to other therapies. Then in early June, the agency rejected MDMA therapy for post-traumatic stress disorder, although it has been approved for limited use in Australia. Meanwhile, healthcare practitioners in Oregon are already using psilocybin, in combination with counseling, to treat depression, although the drug hasn’t yet been federally approved.
Despite its potential, no one knows how psilocybin works in the brain, especially over longer durations.
Now, a team from Washington University School of Medicine has comprehensively documented brain-wide changes before, during, and after a single dose of psilocybin over a period of weeks. As a control, the volunteers also took Ritalin, a stimulant, at a different time to mimic parts of the psilocybin high.
An fMRI scan shows the effect of psilocybin on the brain. Yellows, oranges, and reds indicate an increasingly large departure from normal activity. Image Credit: Sara Moser/Washington UniversityIn the study, psilocybin dramatically reset brain networks that hum along during active rest—say, while daydreaming or spacing out. These networks control our sense of self, time, and space. Although most effects were temporary, one connection showed changes for weeks.
In some participants, the alterations were so drastic that their brain connections resembled those of completely different people.
Normally, the brain synchronizes activity across regions. Psilocybin disrupts these connections, in turn making the brain more malleable and ready to form new networks.
This could be how magic mushrooms “contribute to persistent changes…in brain regions that are responsible for controlling a person’s sense of self, emotion, and life-narrative,” wrote Petros Petridis at the NYU Langone Center for Psychedelic Medicine, who was not involved in the study.
Magical Mystery TourThe brain’s 100 billion neurons and trillions of connections are highly organized into local and brain-wide networks.
Local networks tackle immediate tasks such as processing vision, sound, or motor functions. Brain-wide networks integrate information from local networks to coordinate more complex tasks, such as decision-making, reasoning, or self-reflection.
Previous psilocybin studies mainly focused on local networks. In rodents, for example, the drug regrew neural connections that often wither away in people with severe depression. Scientists have also pinpointed a receptor—which psilocybin grabs onto—that triggers this growth.
But psilocybin’s effects on the whole human brain remained a mystery.
Several years back, one team sought an answer by giving people with severe depression a dose of psilocybin. Using functional MRI (fMRI), a type of imaging that captures brain activity based on changes in blood flow, they found the chemical desynchronized neural networks across the entire brain, essentially “rebooting” them out of a depressive state.
Daydream BelieverThe new study used fMRI to track brain activity in seven adults without mental health struggles before, during, and for three weeks after they took psilocybin. The researchers gave participants a single dose on par with that commonly used in clinical trials for depression.
During the scans, the participants had two tasks. One sounds easy: They kept still and focused their gaze on white crosshairs on a computer screen, but remained otherwise relatively relaxed. Even so, tripping on mushrooms inside a noisy, claustrophobic machine is hardly relaxing—heart rate skyrockets, nerves are on high alert, and anxiety rapidly builds. To control for these side effects, the participants also took Ritalin—a stimulant commonly used to manage attention deficit hyperactivity disorder—at another point in time during the study.
The other task required more brain power. Like an audio version of a CAPTCHA, the researchers asked volunteers to match an image and a word prompt—for example, they’d have to pick a photo of a beach after hearing the word “beach.”
Throughout the study, each person had their brains scanned roughly every other day, on average totaling 18 scans.
Mapping brain connections over time in the same person can “minimize the effects of individual differences in brain networks organization,” wrote Petridis.
The study found psilocybin immediately desynchronized a brain-wide network, generating a brain activation “fingerprint” of sorts that differentiates it from a sober brain.
Dubbed the default mode network, this neural system is active when the mind is alert but wanders, like when reliving previous memories or imagining future scenarios. The network is distributed across the brain and is often studied for its role in consciousness and a sense of self. The chemical also desynchronized local networks across the cortex, the outermost layer of the brain that supports perception, reasoning, and decision-making.
However, the chemical partially lost its magic when the volunteers were focused on the image-audio task, at which point the scans showed less disruption to the default mode network.
This has implications for psilocybin-assisted treatment. Clinical studies have shown that during psychedelic therapy, a challenging experience—a bad trip—can be overcome by a method called “grounding,” which reconnects the person to the outside world.
These results could explain why adding eye masks and ear plugs can enhance the therapeutic experience by blocking outside stimulation, while grounding pulls one out of a bad trip.
Psilocybin’s effects lingered for a few days, after which most brain networks returned to normal—with one exception. A link between the default mode network and a part of the brain involved in creating memories, emotions, and a sense of time, space, and self was disrupted for weeks.
In a way, psilocybin opens a window during which neural connections become more malleable and easier to rewire. People with depression or post-traumatic stress disorder often have a rigid and maladaptive thought pattern that’s hard to shake off. With therapy, psilocybin allows the brain to reorganize those networks, potentially helping people with depression to escape negative ruminations or for people suffering from addiction to consider a new perspective on their relationship to substances.
“In other words, psilocybin could open the door to change, allowing the therapist to lead the patient through,” wrote Petridis.
Although the study offered a higher resolution image of the brain on mushrooms over a longer timeframe than ever before, it only captured scans of seven people. As the participants did not have mental health issues, their responses to psilocybin may differ from those most likely to benefit therapeutically.
Ultimately, larger studies in diverse patient populations—as in several recent MDMA trials—could offer more insights into the efficacy of psilocybin therapy. For example, the one persistent brain network disruption could be an indicator of treatment efficacy. Investigating whether other psychedelics alter the same neural connection is a worthy next step, wrote Petridis.
With the field of psychedelic therapy projected to reach over $10 billion by 2027, understanding how the drug affects the brain could bring new medications with fewer side effects.
Image Credit: Sara Moser/Washington University
Is it possible that one day we could make Mars like Earth? –Tyla, age 16, Mississippi
When I was in middle school, my biology teacher showed our class the sci-fi movie Star Trek III: The Search for Spock.
The plot drew me in with its depiction of the “Genesis Project”—a new technology that transformed a dead alien world into one brimming with life.
After watching the movie, my teacher asked us to write an essay about such technology. Was it realistic? Was it ethical? And to channel our inner Spock: Was it logical? This assignment had a huge impact on me.
Fast-forward to today, and I’m an engineer and professor developing technologies to extend the human presence beyond Earth.
For example: I’m working on advanced propulsion systems to take spacecraft beyond Earth’s orbit. I’m helping to develop lunar construction technologies to support NASA’s goal of a long-term human presence on the moon. And I’ve been on a team that showed how to 3D print habitats on Mars.
To sustain people beyond Earth will take a lot of time, energy, and imagination. But engineers and scientists have started to chip away at the many challenges.
A photo taken of the bleak Martian surface by NASA’s Perseverance rover in June 2024. Image Credit: NASA/JPL-CaltechA Partial Checklist: Food, Water, Shelter, AirAfter the moon, the next logical place for humans to live beyond Earth is Mars.
But is it possible to terraform Mars—that is, transform it to resemble the Earth and support life? Or are these just the musings of science fiction?
To live on Mars, humans will need liquid water, food, shelter, and an atmosphere with enough oxygen to breathe and that’s thick enough to retain heat and protect against radiation from the sun.
But the Martian atmosphere is almost all carbon dioxide, with virtually no oxygen. And it’s very thin—only about 1 percent as dense as the Earth’s.
The less dense an atmosphere, the less heat it can hold onto. Earth’s atmosphere is thick enough to retain the heat needed to sustain life by what’s known as the greenhouse effect.
But on Mars, the atmosphere is so slight that the nighttime temperature drops routinely to -150 degrees Fahrenheit (-101 degrees Celsius).
So what’s the best way to give Mars an atmosphere?
Although Mars has no active volcanoes now—at least as far as we know—scientists could trigger volcanic eruptions via nuclear explosions. The gases trapped deep in a volcano would be released and then drift into the atmosphere. But that scheme is a bit harebrained because the explosions would also introduce deadly radioactive material into the air.
A better idea: Redirecting water-rich comets and asteroids to crash into Mars. That too would release gases from below the planet’s surface into the atmosphere while also releasing the water found in the comets. NASA has already demonstrated that it is possible to redirect asteroids—but relatively large ones, and lots of them, are needed to make a difference.
Making Mars CozyThere are numerous ways to heat up the planet. For instance, gigantic mirrors, built in space and placed in orbit around Mars, could reflect sunlight to the surface and warm it up.
One recent study proposed that Mars colonists could spread aerogel, an ultralight solid material, on the ground. The aerogel would act as insulation and trap heat. This could be done all over Mars, including the polar ice caps, where the aerogel could melt the existing ice to make liquid water.
To grow food, you need soil. On Earth, soil is composed of five ingredients: minerals, organic matter, living organisms, gases, and water.
But Mars is covered in a blanket of loose, dust-like material called regolith. Think of it as Martian sand. The regolith contains few nutrients, not enough for healthy plant growth, and it hosts some nasty chemicals called perchlorates, used on Earth in fireworks and explosives.
Cleaning up the regolith and turning it into something viable wouldn’t be easy. What the alien soil needs is some Martian fertilizer, maybe made by adding extremophiles to it—hardy microbes imported from Earth that can survive even the harshest conditions. Genetically engineered organisms are also a possibility.
Through photosynthesis, these organisms would begin converting carbon dioxide to oxygen. Eventually, as Mars became more friendly to Earth-like organisms, colonists could introduce more complex plants and even animals.
Providing oxygen, water, and food in the right proportions is extraordinarily complex. On Earth, scientists have tried to simulate this in Biosphere 2, a closed-off ecosystem featuring ocean, tropical, and desert habitats. Although all of Biosphere 2’s environments are controlled, even there scientists struggle to get the balance right. Mother Nature really knows what she’s doing.
A House on MarsBuildings could be 3D printed; initially, they would need to be pressurized and protected until Mars acquired Earth-like temperatures and air. NASA’s Moon-to-Mars Planetary Autonomous Construction Technologies program is researching how to do exactly this.
There are many more challenges. For example, unlike Earth, Mars has no magnetosphere, which protects a planet from solar wind and cosmic radiation. Without a magnetic field, too much radiation gets through for living things to stay healthy. There are ways to create a magnetic field, but so far the science is highly speculative.
In fact, all the technologies I’ve described are far beyond current capabilities at the scale needed to terraform Mars. Developing them would take enormous amounts of research and money, probably much more than possible in the near term. Although the Genesis device from Star Trek III could terraform a planet in a matter of minutes, terraforming Mars would take centuries or even millennia.
And there are a lot of ethical questions to resolve before people get started on turning Mars into another Earth. Is it right to make such drastic permanent changes to another planet?
If this all leaves you disappointed, don’t be. As scientists create innovations to terraform Mars, we’ll also use them to make life better on Earth. Remember the technology we’re developing to 3D print habitats on Mars? Right now, I’m part of a group of scientists and engineers employing that very same technology to print homes here on Earth—which will help address the world’s housing shortage.
Curious Kids is a series for children of all ages. If you have a question you’d like an expert to answer, send it to curiouskidsus@theconversation.com.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Daein Ballard / Wikimedia Commons
With half their skulls replaced by translucent 3D-printed implants, the mice looked straight out of a science fiction movie. Yet they nosed around, ferociously ate their chow, and groomed as usual. Meanwhile, sensors spread across half their brains recorded electrical chatter.
Brain implants have revolutionized neuroscience. Our perception, thoughts, emotions, and memories all rely on electrical signals spreading across networks of neurons. Implants tap into these signals and—often with the help of AI—can rapidly decipher seemingly random electrical activity into intent or movement.
So-called “mind reading” devices translate brain signals related to speech into text, allowing people who’ve lost their ability to speak to communicate directly with loved ones with their minds. Others tap into motor regions of the brain or nerves in the spinal cord and help people with severe paralysis to walk again. Neural signals, alone or combined with eye movements, can even control cursors on a computer screen, re-opening the digital world to paralyzed people for texting, Googling, and scrolling through social media.
These devices are beginning to transform lives for the better. But all rely on the answer to one critical question: How does the brain support those functions? So far, each implant has focused on a small brain region underlying a given capability—controlling vision or movement.
But many of the brain’s functions rely on signals, or brain waves, that spread across multiple regions and synchronize electrical activity. The height and frequency of the waves—some come fast and low, others slow and high—change the brain’s overall function.
Scientists can measure these waves, but like a camera with low resolution, they can’t explain how the waves are generated, propagate, and eventually die down.
In a new study, the custom-fitted transparent implant described above replaces the skull in mice, offering a way to, literally, peek into the brain in search of answers.
The Evolution of Brain ProbesNeural implants have been around since the 1980s. The idea is simple. The brain uses electrical and chemical signals to process information. Electrodes can tap into the electrical communications. Sophisticated software then deciphers the neural code, potentially allowing us to reprogram it and tackle neurological symptoms when the code breaks down.
There are a few ways to make it work. One is to directly record individual neurons—often in rodents—to see which activate when challenging a mouse to a task. Another technology records large-scale brain activity from beneath the skull. This approach sacrifices resolution—we no longer know how each individual neuron behaves—but paints a broader picture.
The challenge is how to combine resolution and scale. A previous attempt relied on multiple high-density electrodes inserted into the brain. Called Neuropixels, each implant is a powerhouse with over 5,000 recording sites packed in a tiny, durable package. “Extremely large numbers of individual neurons could…be followed and tracked with the same probe for weeks and occasionally months,” the authors of a paper about the implant wrote at the time.
But to measure brain-wide activity, scientists have to place multiple Neuropixel devices across the brain. Each requires drilling through the skull and could harm the plastic-wrap-like structure, known as the blood-brain barrier, that protects the brain. Damage from these surgeries often compounds, triggering inflammation that could change how the brain works for weeks with an increasing risk of infection.
So far, scientists have inserted up to eight implants to record activity in mice as they went about their lives or participated in experiments. While they gained news insights, scientists struggled to keep the mice healthy after multiple surgeries. In other words, it wasn’t the Neuropixel implants causing problems—it was all the brain surgeries.
Might there be an alternative?
A 3D ReplacementTo avoid multiple surgeries, the Allen Institute team behind the new study developed an implant that covers nearly half a mouse’s brain.
Called SHIELD, the implant looks a bit like molded Swiss cheese. The scaffold is carefully contoured into a shape that perfectly mimics the skulls of young mice.
Then comes the customization. The SHIELD scaffold can accommodate up to 21 small insertion holes for Neuropixels. Scientists can strategically choose where to put the holes to record from multiple brain regions of interest. The implant is then printed with resin, a viscous liquid often used in everyday 3D printing.
“The SHIELD implant is straightforward to fabricate in-house, using a commercially available and relatively low-cost 3D printer,” wrote the team.
Once printed, each hole is temporarily filled with transparent silicon rubber. Like a flexible windshield, the rubber protects delicate brain tissue during implantation. The SHIELD then replaces half of the skull in a single surgery.
It sounds traumatic, but the team made sure the procedure didn’t harm the mice’s health or brains. Images of their brains at multiple time points over two months after surgery showed little damage in most mice, who went about their merry business after a short recovery period. Brain inflammation levels also stayed low during the study.
Here’s how it went. The mice watched one of eight photos flash before them continuously and then learned to lick a treat when the photo switched. During the test, six Neuropixel implants recorded brain activity related to the task. The position of the implants changed every day for four days, altogether collecting neural signals from roughly 25 different brain areas.
Another test dug into the potential underpinnings of brain waves first discovered in the 1920s. These neural oscillations, called alpha waves, are associated with restful and meditative states. In humans, brain waves are usually monitored using a beanie-like cap covered in electrodes that can record most brain regions. With the help of AI, Neuropixel recordings from across the brain homed in on signals resembling alpha waves.
Overall, the team made stable, high-quality recordings from 25 mice, with 467 probe insertions across nearly 90 different experiments.
“Thus, this work goes beyond mere proof of concept,” instead providing a solid recipe for recording across dozens of brain regions and thousands of neurons over multiple days with a single initial surgery, wrote the team.
And there’s a final perk. Because SHIELD is translucent, scientists can tweak brain activity using light. This approach, called optogenetics, alters brain activity with flashes of light—either amping it up or turning it down—giving scientists insight into the neural underpinnings of thoughts, emotions, and memories.
The authors shared 3D printing files of the scaffold for other scientists to design their own custom implants.
Image Credit: Ralph / Pixabay
A new generation of “flying cars” promises to revolutionize urban mobility, but limited battery power holds them back from plying longer routes. A new hydrogen-powered variant from Joby Aviation could soon change that.
Rapid advances in battery technology and electric motors have opened the door to a new class of aircraft known as eVTOLs, which stands for electric vertical takeoff and landing. The companies making the aircraft tout them as a quieter, greener alternative to helicopters.
However, current battery technology means they’re limited to ranges of approximately 150 miles. That’s why they have primarily been envisaged as a new form of urban mobility, allowing quick hops across cities congested with traffic.
Joby is already developing a battery-powered eVTOL that it expects to start commercial operations next year. But this week, the company announced it has created a hydrogen-powered version of the aircraft, which recently completed a 523-mile test flight. The company says this could allow eVTOLs to break into regional travel as well.
“With our battery-electric air taxi set to fundamentally change the way we move around cities, we’re excited to now be building a technology stack that could redefine regional travel using hydrogen-electric aircraft,” JoeBen Bevirt, founder and CEO of Joby, said in a press release.
“Imagine being able to fly from San Francisco to San Diego, Boston to Baltimore, or Nashville to New Orleans without the need to go to an airport and with no emissions except water.”
Joby’s demonstrator is a converted battery-electric aircraft that had already completed 25,000 miles of test flights. It features the same airframe with six electric-motor-powered tilting propellers that allow it to take off vertically like a helicopter but cruise like a light aircraft. Joby says this should significantly speed up the certification process if the company decides to commercialize the technology.
What’s new is the addition of a hydrogen fuel cell system designed by H2FLY, a German startup Joby acquired in 2021, and a liquid hydrogen fuel tank that can store about 40 kilograms of fuel. The fuel cell combines the liquid hydrogen with oxygen from the air to generate the electricity that powers the aircraft’s motors. The H2FLY team used the same underlying technology in a series of demonstration flights with a more conventional aircraft design last year.
The new Joby aircraft will still carry some batteries to provide additional power during takeoff and landing. But hydrogen has a much higher energy density—or specific energy—than batteries, which makes it possible to power the aircraft for significantly longer.
“Hydrogen has one hundred times the specific energy of today’s batteries and three times that of jet fuel,” Bevirt wrote in a blog post. “The result is an electric aircraft that can travel much farther—and carry a greater payload—than is possible not only with any battery cells currently under development, but even with the same mass of jet fuel.”
However, switching to hydrogen fuel poses some challenges. For a start, hydrogen requires complicated cooling equipment, which means airports or other landing facilities would need to invest significant amounts in new fueling infrastructure.
“The industry is already scratching its head figuring out how to support battery electric aircraft with charging infrastructure at airports,” Cyrus Sigari, co-founder and managing partner of VC Up.Partners, told TechCrunch. “Adding hydrogen filling stations into that equation will present even more challenges.”
Hydrogen’s green credentials are also somewhat weaker than those of batteries. While it’s possible to generate hydrogen from water using only renewable electricity, at present the vast majority is produced from fossil fuels.
However, efforts are underway to increase the supply of green hydrogen, and the Bipartisan Infrastructure Law passed in 2021 set aside $9.5 billion to help boost these efforts. And if hydrogen-powered flight can piggyback on innovations in eVTOL technology, it could prove a powerful way to curb emissions in one of the world’s most polluting sectors.
Image Credit: Joby
ARTIFICIAL INTELLIGENCEOpenAI Reportedly Nears Breakthrough With ‘Reasoning’ AI, Reveals Progress Framework
Benj Edwards | Ars Technica“[According to OpenAI’s new AGI framework] a Level 2 AI system would reportedly be capable of basic problem-solving on par with a human who holds a doctorate degree but lacks access to external tools. During the all-hands meeting, OpenAI leadership reportedly demonstrated a research project using their GPT-4 model that the researchers believe shows signs of approaching this human-like reasoning ability, according to someone familiar with the discussion who spoke with Bloomberg.”
BIOTECHHow AI Revolutionized Protein Science, but Didn’t End It
Yasemin Saplakoglu | Quanta“Three years ago, Google’s AlphaFold pulled off the biggest artificial intelligence breakthrough in science to date, accelerating molecular research, and kindling deep questions about why we do science. …’The field of protein biology is ‘more exciting right now than it was before AlphaFold,’ Perrakis said. The excitement comes from the promise of reviving structure-based drug discovery, the acceleration in creating hypotheses, and the hope of understanding complex interactions happening within cells.”
TECHNew Fiber Optics Tech Smashes Data Rate Record
Margo Anderson | IEEE Spectrum“An international team of researchers have smashed the world record for fiber optic communications through commercial-grade fiber. By broadening fiber’s communication bandwidth, the team has produced data rates four times as fast as existing commercial systems—and 33 percent better than the previous world record.”
ARTIFICIAL INTELLIGENCE‘Superhuman’ Go AIs Still Have Trouble Defending Against These Simple Exploits
Kyle Orland | Ars Technica“In the ancient Chinese game of Go, state-of-the-art artificial intelligence has generally been able to defeat the best human players since at least 2016. But in the last few years, researchers have discovered flaws in these top-level AI Go algorithms that give humans a fighting chance. By using unorthodox ‘cyclic’ strategies—ones that even a beginning human player could detect and defeat—a crafty human can often exploit gaps in a top-level AI’s strategy and fool the algorithm into a loss.”
COMPUTINGGoogle Creates Self-Replicating Life From Digital ‘Primordial Soup’
Matthew Sparkes | New Scientist“A self-replicating form of artificial life has arisen from a digital ‘primordial soup’ of random data, despite a lack of explicit rules or goals to encourage such behavior. Researchers believe it is possible that more sophisticated versions of the experiment could yield more advanced digital organisms, and if they did, the findings could shed light on the mechanisms behind the emergence of biological life on Earth.”
AUTOMATIONHow Good Is ChatGPT at Coding, Really?
Michelle Hampson | IEEE Spectrum“Programmers have spent decades writing code for AI models, and now, in a full circle moment, AI is being used to write code. But how does an AI code generator compare to a human programmer? [A new study shows] that ChatGPT has an extremely broad range of success when it comes to producing functional code—with a success rate ranging from anywhere as poor as 0.66 percent and as good as 89 percent—depending on the difficulty of the task, the programming language, and a number of other factors.”
TECHOpenAI Anticipates Decrease in AI Model Costs Amid Adoption Surge
Shubham Sharma | VentureBeat“‘We introduced GPT-4, the first version, some 15 months ago. Since then, the cost of a token/word on the model has been reduced by 85-90%. There’s no reason why that trend will not continue,’ Olivier Godement, [OpenAI’s] head of API Product said. …He expects the company’s work on affordability, spanning efforts to optimize costs at both hardware and inference levels, will continue, leading to a further decline in the cost of running frontier AI models—much like what has been the case with smartphones and televisions.”
SPACEWatch These Supernovas in (Time-Lapse) Motion
Dennis Overbye | The New York Times“This spring, the astronomers who operate Chandra combined its X-ray images into videos that document the evolution of two astrophysical landmarks: the Crab nebula, in the constellation Taurus, and Cassiopeia A, a gas bubble and hub of radio noise in the constellation Cassiopeia. The videos show twisting, drifting ribbons of the remains of the star being churned by shock waves and illuminated by radiation from the dense, spinning cores left behind.”
Image Credit: BoliviaInteligente / Unsplash
CRISPR was one of the most influential breakthroughs of the last decade, but it’s still imperfect. While the gene editing tool is already helping people with genetic ailments, scientists are also looking to improve on it.
Efforts have extended the CRISPR family to include less damaging, more accurate, and smaller versions of the gene editor. But in the bacterial world, where CRISPR was originally discovered, we’re only scratching the surface. Two new papers suggest an even more powerful gene editor may be around the corner—if it’s proven to work in cells like our own.
In one of the papers, scientists at the Arc Institute say they discovered a new CRISPR-like gene editing tool in bacterial “jumping genes.” Another paper, written independently, covers the same tool and extends the work to a similar one in a different family.
Jumping genes move around within genomes and even between individuals. It’s long been known they do this by cutting and pasting their own DNA, but none of the machinery has been shown to be programmable like CRISPR. In the recent studies, scientists describe jumping gene systems that, in a process the teams are alternatively calling bridge editing and seekRNA, can be modified to cut, paste, and flip any DNA sequence.
Crucially, unlike CRISPR, the system does all this without breaking strands of DNA or relying on the cell to repair them, a process that can be damaging and unpredictable. The various molecules involved are also fewer and smaller than those in CRISPR, potentially making the tool safer and easier to deliver into cells, and can deal with much longer sequences.
“Bridge recombination can universally modify genetic material through sequence-specific insertion, excision, inversion, and more, enabling a word processor for the living genome beyond CRISPR,” said Berkeley’s Patrick Hsu, a senior author of one of the studies and Arc Institute core investigator, in a press release.
CRISPR CoupScientists first discovered CRISPR in bacteria defending themselves against viruses. In nature, a Cas9 protein pairs with an RNA guide molecule to seek out viral DNA and, when located, chop it up. Researchers learned to reengineer this system to seek out any DNA sequence, including sequences found in human genomes, and break the DNA strands at those locations. The natural machinery of the cell then repairs these breaks, sometimes using a provided strand of DNA.
CRISPR gene editing is powerful. It’s being investigated in clinical trials as a treatment for a variety of genetic diseases and, late last year, received its first clinical approval as a therapy for sickle cell disease and beta thalassemia. But it’s not perfect.
Because the system breaks DNA and relies on the cell to repair these breaks, it can be imprecise and unpredictable. The tool also works primarily on short sections of DNA. While many genetic illnesses are due to point mutations, where a single DNA “letter” has been changed, the ability to work with longer sequences would broaden the technology’s potential uses in both synthetic biology and gene therapy.
Scientists have developed new CRISPR-based systems over the years to address these shortcomings. Some systems only break a single DNA strand or swap out single genetic “letters” to increase precision. Studies are also looking for more CRISPR-like systems by screening the whole bacterial universe; others have found naturally occurring systems in eukaryotic cells like our own.
The new work extends the quest by adding jumping genes into the mix.
An RNA BridgeJumping genes are a fascinating feat of genetic magic. These sequences of DNA can move between locations in the genome using machinery to cut and paste themselves. In bacteria, they even move between individuals. This sharing of genes could even be one way bacteria acquire antibiotic resistance—one cell that’s evolved to evade a drug can share its genetic defenses with a whole population.
In the Arc Institute study, researchers looked into a specific jumping gene in bacteria called IS110. They found that when the gene is on the move, it calls a sequence of RNA—like the RNA guide in CRISPR—to facilitate the process. The RNA includes two loops: One binds the gene itself and the other seeks out and binds to the gene’s destination in the genome. It acts like a bridge between the DNA sequence and the specific location where it’s to be inserted. In contrast to CRISPR, once found, the sequence can be added without breaking DNA.
“Bridge editing [cuts and pastes DNA] in a single-step mechanism that recombines and re-ligates the DNA, leaving it fully intact,” Hsu told Fierce Biotech in an email. “This is very distinct from CRISPR editing, which creates exposed DNA breaks that require DNA repair and have been shown to create undesired DNA damage responses.”
Crucially, the researchers discovered both loops of RNA can be reprogrammed. That means scientists can specify a genomic location as well as what what sequence should go there. In theory, the system could be used to swap in long genes or even multiple genes. As a proof of concept in E. coli bacteria, the team programmed IS110 to insert a DNA sequence that was almost 5,000 bases long. They also cut out and inverted another sequence of DNA.
The study was joined by a different paper written independently by another team of scientists at the University of Sydney detailing both IS110 and a related enzyme in a different family, IS111, that they say is similarly programmable. In their paper, they called these systems “seekRNA.”
The tools rely on a single protein half the size of those in CRISPR. That means it may be easier to package them in harmless viruses or lipid nanoparticles—these are also used in Covid vaccines—and ferry them into cells where they can get to work.
The Next JumpThe approach has big potential, but there’s also a big caveat. So far, the researchers have only shown it works in bacteria. CRISPR, on the other hand, is incredibly versatile, having proved itself in myriad cell types. Next, they hope to hone the approach further and adapt it to mammalian cells like ours. That may not be easy. The University of Tokyo’s Hiroshi Nishimasu says the IS110 family hasn’t yet shown itself amenable to such a task.
All this is to say it’s still early in the technology’s arc. Scientists knew about CRISPR years before they showed it was programmable, and it wasn’t put to work in human cells until 2013. Although it’s moved relatively quickly from lab to clinic since then, the first CRISPR-based treatments took years more to materialize.
At the least, the new work shows we haven’t exhausted all nature has to offer gene editing. The tech could also be useful in the realm of synthetic biology, where single cells are being engineered on grand scales to learn how life works at its most basic and how we might reengineer it. And if the new system can be adapted for human cells, it would be a useful new option in the development of safer, more powerful gene therapies.
“If this works in other cells, it will be game-changing,” Sandro Fernandes Ataide, a structural biologist at the University of Sydney and author on the paper detailing IS111 told Nature. “It’s opening a new field in gene editing.”
Image Credit: The Arc Institute
Roughly 52,000 years ago, a woolly mammoth died in the Siberian tundra. As her body flash froze in the biting cold, something remarkable happened: Her DNA turned into a fossil. It wasn’t only genetic letters that were memorialized—the cold preserved their intricate structure too.
Fast forward to 2018, when an international expedition to the area found her preserved body. The team took little bits of skin from her head and ear, hairs still intact.
From these samples, scientists built a three-dimensional reconstruction of a woolly mammoth’s genome down to the nanometer. The results were published in Cell today.
Like humans, the mammoth’s DNA strands are tightly packed into chromosomes inside cells. These sophisticated structures are hard to analyze in detail, even for humans, but they contain insights into which genes are turned on or off and how they’re organized in different cell types.
Previous attempts to reconstruct ancient DNA only had tiny snippets of genetic sequences. Like trying to put together a puzzle with missing pieces, the resulting DNA maps were incomplete.
Thanks to the newly discovered flash-frozen DNA, this mammoth project—pun intended—is the first to assemble an enormous ancient genome in 3D.
“This is a new type of fossil, and its scale dwarfs that of individual ancient DNA fragments—a million times more sequence,” said study author Erez Lieberman Aiden at Baylor College of Medicine in a statement.
Aiden’s team heavily collaborated with Love Dalén at the Center of Palaeogenetics in Sweden. In a separate study, Dalén’s team analyzed 21 Siberian woolly mammoth genomes and charted how the species survived for six millennia after a potentially catastrophic genetic “bottleneck.”
The mammoth genomes weren’t that different than those of today’s Asian and African elephants. All have 28 pairs of chromosomes, and their X chromosomes twist into unique structures unlike most mammals. Digging deeper, the team found genes that were turned on or off in the mammoth compared to its elephant cousins.
“Our analyses uncover new biology,” wrote Aiden’s team in their paper.
DNA SerendipityAncient DNA is hard to come by, but it offers invaluable clues about the evolutionary past. In the 1980s, scientists eager to probe genetic history showed ancient DNA, however fragmented, could be extracted and sequenced in samples from an extinct member of the horse family and Egyptian mummies.
Thanks to modern DNA sequencing, the study of ancient DNA “has subsequently undergone a remarkable expansion,” wrote Aiden’s team. It’s now possible to sequence whole genomes from extinct humans, animals, plants, and even pathogens spanning a million years.
Making sense of the fragments is another matter. One way to decipher ancient genetic codes is to compare them to the genomes of their closest living cousins, such as woolly mammoths and elephants. This way, scientists can figure out which parts of the DNA sequence remained unchanged and where evolution swapped letters or small fragments.
These analyses can link genetic changes to function, such as identifying which genes made mammoths woolly. But they can’t capture large-scale differences at the chromosomal level. Because DNA relies on the chromosome’s 3D structure to function, sequencing its letters alone misses valuable information, such as when and where genes are turned on or off.
Chromosome Puzzle MasterEnter Hi-C. Developed in 2009 to reconstruct human genomes, the technique detects interactions between different genetic sites inside the cell’s nucleus.
Here’s roughly how it works. DNA strands are like ribbons that twirl around proteins in a structure resembling beads on a string. Because of this arrangement, different parts of the DNA strand are closer to each other in physical space. Hi-C “glues” together sections that are near one another and tags the pairs. Alongside modern DNA sequencing, the technique produces a catalog of DNA fragments that interact in physical space. Like a 3D puzzle, scientists can then put the pieces back together.
“Imagine you have a puzzle that has three billion pieces, but you don’t have the picture of the final puzzle to work from,” study author Marc A. Marti-Renom said in the press release. “Hi-C allows you to have an approximation of that picture before you start putting the puzzle pieces together.”
But Hi-C can be impossible to use in ancient samples because the surviving fragments are so short they’ve erased any chromosome shapes. They’ve literally withered away over time.
In the new study, the team developed a new technique, called PaleoHi-C, to analyze ancient DNA specifically.
Scientists immediately treated samples in the field to reduce contamination. They generated roughly 4.4 billion “pairs” of physically aligned DNA sequences—some interacting within a single chromosome, others between two. Overall, they painted a 3D snapshot of the woolly mammoth’s genetic material and how it looked inside cells with nanoscale detail.
In the new reconstructions, the team identified chromosome territories—certain chromosomes are located in different regions of the nucleus—alongside other quirks, such as loops that bring pairs of distant genomic sites into close physical proximity to alter gene expression. These patterns differed between cell types, suggesting it’s possible to learn which genes are active, not just for the mammoth but also compared to its closest living relative, the Asian elephant.
Roughly 820 genes differed between the two, with 425 active in the mammoth but not in elephants, and a similar number inactivated in one but not the other. One inactive mammoth gene that’s active in elephants has a human variant that is also shut down in the Nunavik Inuit, an indigenous people who thrive in the arctic. The gene “may be relevant for adaptation to a cold environment,” wrote the team.
Another inactive gene may explain how the woolly mammoth got its name. In humans and sheep, shutting down the same gene can result in excessive hair or wool growth.
“For the first time, we have a woolly mammoth tissue for which we know roughly which genes were switched on and which genes were off,” said Marti-Renom in the release. “This is an extraordinary new type of data, and it’s the first measure of cell-specific gene activity of the genes in any ancient DNA sample.”
Crystalized DNAHow did the mammoth’s genome architecture remain so well preserved for over 50,000 years?
Dehydration, often used to preserve food, may have been key. Using Hi-C on fresh beef, beef after 96 hours sitting on a desk, or jerky after a year at room temperature, the jerky took the win for resiliency. Even after getting run over by a car, immersed in acid, and pulverized by a shotgun (no joke), the dehydrated beef’s genomic architecture remained intact.
Dehydration could also partly be why the mammoth sample lasted so long. A chemical process called “glass transition” is widely used to produce shelf-stable food such as tortilla chips and instant coffee. It prevents pathogens from taking over or breaking down food. The mammoth’s DNA may also have been preserved in a glassy state called “chromoglass.” In other words, the sample was preserved across millennia by being freeze-dried.
It’s hard to say how long DNA architecture can survive as chromoglass, but the authors estimate it’s likely over two million years. Whether PaleoHi-C can work on hot-air-dried specimens, such as ancient Egyptian samples, remains to be seen.
As for mammoths, the next step is to examine gene expression patterns in other tissues and compare them to Asian elephants. Besides building an evolutionary throughline, the efforts could also guide ongoing studies looking to revive some version of the majestic animals.
“These results have obvious consequences for contemporary efforts aimed at woolly mammoth de-extinction,” said study author Thomas Gilbert at the University of Copenhagen in the release.
Image Credit: Beth Zaiken
Computer chips are a hot commodity. Nvidia is now one of the most valuable companies in the world, and the Taiwanese manufacturer of Nvidia’s chips, TSMC, has been called a geopolitical force. It should come as no surprise, then, that a growing number of hardware startups and established companies are looking to take a jewel or two from the crown.
Of these, Cerebras is one of the weirdest. The company makes computer chips the size of tortillas bristling with just under a million processors, each linked to its own local memory. The processors are small but lightning quick as they don’t shuttle information to and from shared memory located far away. And the connections between processors—which in most supercomputers require linking separate chips across room-sized machines—are quick too.
This means the chips are stellar for specific tasks. Recent preprint studies in two of these—one simulating molecules and the other training and running large language models—show the wafer-scale advantage can be formidable. The chips outperformed Frontier, the world’s top supercomputer, in the former. They also showed a stripped down AI model could use a third of the usual energy without sacrificing performance.
Molecular MatrixThe materials we make things with are crucial drivers of technology. They usher in new possibilities by breaking old limits in strength or heat resistance. Take fusion power. If researchers can make it work, the technology promises to be a new, clean source of energy. But liberating that energy requires materials to withstand extreme conditions.
Scientists use supercomputers to model how the metals lining fusion reactors might deal with the heat. These simulations zoom in on individual atoms and use the laws of physics to guide their motions and interactions at grand scales. Today’s supercomputers can model materials containing billions or even trillions of atoms with high precision.
But while the scale and quality of these simulations has progressed a lot over the years, their speed has stalled. Due to the way supercomputers are designed, they can only model so many interactions per second, and making the machines bigger only compounds the problem. This means the total length of molecular simulations has a hard practical limit.
Cerebras partnered with Sandia, Lawrence Livermore, and Los Alamos National Laboratories to see if a wafer-scale chip could speed things up.
The team assigned a single simulated atom to each processor. So they could quickly exchange information about their position, motion, and energy, the processors modeling atoms that would be physically close in the real world were neighbors on the chip too. Depending on their properties at any given time, atoms could hop between processors as they moved about.
The team modeled 800,000 atoms in three materials—copper, tungsten, and tantalum—that might be useful in fusion reactors. The results were pretty stunning, with simulations of tantalum yielding a 179-fold speedup over the Frontier supercomputer. That means the chip could crunch a year’s worth of work on a supercomputer into a few days and significantly extend the length of simulation from microseconds to milliseconds. It was also vastly more efficient at the task.
“I have been working in atomistic simulation of materials for more than 20 years. During that time, I have participated in massive improvements in both the size and accuracy of the simulations. However, despite all this, we have been unable to increase the actual simulation rate. The wall-clock time required to run simulations has barely budged in the last 15 years,” Aidan Thompson of Sandia National Laboratories said in a statement. “With the Cerebras Wafer-Scale Engine, we can all of a sudden drive at hypersonic speeds.”
Although the chip increases modeling speed, it can’t compete on scale. The number of simulated atoms is limited to the number of processors on the chip. Next steps include assigning multiple atoms to each processor and using new wafer-scale supercomputers that link 64 Cerebras systems together. The team estimates these machines could model as many as 40 million tantalum atoms at speeds similar to those in the study.
AI LightWhile simulating the physical world could be a core competency for wafer-scale chips, they’ve always been focused on artificial intelligence. The latest AI models have grown exponentially, meaning the energy and cost of training and running them has exploded. Wafer-scale chips may be able to make AI more efficient.
In a separate study, researchers from Neural Magic and Cerebras worked to shrink the size of Meta’s 7-billion-parameter Llama language model. To do this, they made what’s called a “sparse” AI model where many of the algorithm’s parameters are set to zero. In theory, this means they can be skipped, making the algorithm smaller, faster, and more efficient. But today’s leading AI chips—called graphics processing units (or GPUs)—read algorithms in chunks, meaning they can’t skip every zeroed out parameter.
Because memory is distributed across a wafer-scale chip, it can read every parameter and skip zeroes wherever they occur. Even so, extremely sparse models don’t usually perform as well as dense models. But here, the team found a way to recover lost performance with a little extra training. Their model maintained performance—even with 70 percent of the parameters zeroed out. Running on a Cerebras chip, it sipped a meager 30 percent of the energy and ran in a third of the time of the full-sized model.
Wafer-Scale Wins?While all this is impressive, Cerebras is still niche. Nvidia’s more conventional chips remain firmly in control of the market. At least for now, that appears unlikely to change. Companies have invested heavily in expertise and infrastructure built around Nvidia.
But wafer-scale may continue to prove itself in niche, but still crucial, applications in research. And it may be the approach becomes more common overall. The ability to make wafer-scale chips is only now being perfected. In a hint at what’s to come for the field as a whole, the biggest chipmaker in the world, TSMC, recently said it’s building out its wafer-scale capabilities. This could make the chips more common and capable.
For their part, the team behind the molecular modeling work say wafer-scale’s influence could be more dramatic. Like GPUs before them, adding wafer-scale chips to the supercomputing mix could yield some formidable machines in the future.
“Future work will focus on extending the strong-scaling efficiency demonstrated here to facility-level deployments, potentially leading to an even greater paradigm shift in the Top500 supercomputer list than that introduced by the GPU revolution,” the team wrote in their paper.
Image Credit: Cerebras
Henry Grabar has had enough battling knotweed. All he wanted was to build a small garden in Brooklyn—a bit of peace amid the cacophony of city life. But a plant with beet-red leaves soon took over his nascent garden. The fastest growing plant he’d ever seen, it could sprout up to 10 feet high and grow thick as a cornfield. Even with herbicide, it was nearly impossible to kill.
Invasive plant species and weeds don’t just ruin backyard gardens. Weeds decrease crop yields at an average annual cost of $33 billion, and control measures can rack up $6 billion more. Herbicides are a defense, but they have their own baggage. Weeds rapidly build resistance against the chemicals, and the resulting produce can be a hard sell for many consumers.
Weeds often seem to have the upper hand. Can we take it away?
Two recent studies say yes. Using a technology called a synthetic gene drive, the teams spliced genetic snippets into a mustard plant popular in lab studies. Previously validated in fruit flies, mosquitoes, and mice, gene drives break the rules of inheritance, allowing “selfish” genes to rapidly spread across entire species.
But making gene drives work in plants has been a headache, in part due to the way they repair their DNA. The new studies found a clever workaround, leading to roughly 99 percent propagation of a synthetic genetic payload to subsequent generations, in contrast to nature’s 50 percent. Computer models suggest the gene drives could spread throughout an entire population of the plant in roughly 10 to 30 generations.
Overriding natural evolution, gene drives could add genes that make weeds more vulnerable to herbicides or reduce their pollination and numbers. Beneficial genes can also spread across crops—essentially fast-tracking the practice of cross-breeding for desirable traits.
“Imagine a future where yield-robbing agricultural weeds or biodiversity threatening invasive plants could be kept on a genetic leash,” wrote Paul Neve at the University of Copenhagen and Luke Barrett at CSIRO Agriculture and Food in Australia, who were not involved in the study.
50/50Inheritance is a coin toss for most species. Half of an offspring’s genetic material comes from each parent.
Gene drives torpedo this inheritance rule. Developed roughly a decade ago, the technology relies on CRISPR—the gene editing tool—to spread a new gene throughout a population, beating the 50/50 odds. In insects and mammals, a gene can propagate at roughly 80 percent, shuttling an inherited trait down generations and irreversibly changing an entire species.
While this may seem somewhat nefarious, gene drives are designed for good. A main use under investigation is to control disease-carrying mosquitoes by genetically modifying males to be sterile. Upon release, they outcompete their natural counterparts, reducing wild mosquito numbers, and in turn, lowering the risk of multiple diseases. In indoor cages, gene drives have fully suppressed a population of the insects within a year. Small-scale field tests are underway.
Gene drives have caught the eyes of plant scientists too, but initial efforts in plants failed.
The technology relies on CRISPR, which cuts DNA to insert, delete, or swap out genetic letters. Sensing damage to their DNA, cells activate internal molecular “repairmen” to stitch genes back together and adopt gene drives and their genetic cargo.
Plants are different. Their cells also have a DNA repair mechanism, but it’s only partially similar to that of insects or mice. Sticking a classic gene drive into plants can cause genetic mutations at the target site and even trigger resistance against the gene drive in a kind of a cellular civil war.
What Doesn’t Kill You Makes You StrongerAs a workaround, both new studies used a system dubbed “toxin-antidote.” Compared to previous gene drives, it doesn’t rely on canonical DNA repair.
The teams used a self-pollinating mustard plant for their studies. A darling in plant science research, its genome is well-known, and because the plant self-pollinates, it’s easier to contain the experiment. To build the gene drive, they developed a CRISPR-based method to destroy a gene that’s critical for survival called the “torpedo.” Any pollen without the gene can’t live on. A second construct, the “antidote,” carried a mimic of the same gene, but with modifications so that it’s resistant to destruction by CRISPR.
They examined two different genetic payloads. One study tinkered with a gene that’s essential to both male and female reproductive cells in plants. The other targeted a gene that disrupts pollen production.
Here’s the clever part: As the plant pollinates, offspring can inherit either the toxin, the antidote, or both. Only those with the antidote survive—plants that inherit the toxin rapidly die out. As a result, the system worked as a gene drive, with plants carrying the CRISPR-resistant gene taking over the population. The gene drives were highly efficient, passing down through generations roughly 99 percent of the time. And scientists didn’t see any signs of evolutionary adaptation—known as resistance—against the new genetic makeup.
Computer modeling showed the gene drive could overtake a single plant species in 10 to 30 generations. That’s impressive, according Neve and Barrett. Artificial genetic changes don’t often stick in wild plants—the plants tend to die off. The new gene drives suggest they could potentially last longer in the field, battling invasive species or cultivating hardier and pest-resistant crops that pass down beneficial traits over generations.
Despite their promise, gene drives remain controversial because of their potential to alter entire species. Scientists are still debating the ecological impacts. There’s also the concern that gene drives may hop over to unintended targets. For now, studies have designed genetic “brakes” to keep gene drives in check. Most studies are done in carefully controlled lab settings, and for malaria, potential unexpected consequences are being rigorously discussed before releasing gene drive-carrying mosquitos into the wild.
Even if the science works, the road to regulatory and societal approval may face roadblocks. Selling farmers on the technology may be difficult. And CRISPRed plants as a food source could also be tainted by the negative perception of genetically modified organisms (GMOs).
For now, the teams are looking towards a more acceptable everyday use—killing weeds. There are still a few kinks to work out. Gene drives only work when they can spread, so an ideal use is in plants that pollinate others, rather than those that self-pollinate, such as those in the studies. Still, the results are a proof of concept that the powerful technology can work in plants—though it may be awhile yet before it helps Henry with his knotweed problem.
Image Credit: Anthony Wade / Unsplash
Dune, widely considered one of the best sci-fi novels of all time, continues to influence how writers, artists, and inventors envision the future.
Of course, there are Denis Villeneuve’s visually stunning films, Dune: Part One (2021) and Dune: Part Two (2024).
But Frank Herbert’s masterpiece also helped Afrofuturist novelist Octavia Butler imagine a future of conflict amid environmental catastrophe; it inspired Elon Musk to build SpaceX and Tesla and push humanity toward the stars and a greener future; and it’s hard not to see parallels in George Lucas’ Star Wars franchise, especially the films’ fascination with desert planets and giant worms.
And yet when Herbert sat down in 1963 to start writing Dune, he wasn’t thinking about how to leave Earth behind. He was thinking about how to save it.
Herbert wanted to tell a story about the environmental crisis on our own planet, a world driven to the edge of ecological catastrophe. Technologies that had been inconceivable just 50 years prior had put the world at the edge of nuclear war and the environment on the brink of collapse; massive industries were sucking wealth from the ground and spewing toxic fumes into the sky.
When the book was published, these themes were front and center for readers, too. After all, they were living in the wake of both the Cuban missile crisis and the publication of Silent Spring, conservationist Rachel Carson’s landmark study of pollution and its threat to the environment and human health.
Dune soon became a beacon for the fledgling environmental movement and a rallying flag for the new science of ecology.
Indigenous WisdomsThough the term “ecology” had been coined almost a century earlier, the first textbook on ecology was not written until 1953, and the field was rarely mentioned in newspapers or magazines at the time. Few readers had heard of the emerging science, and even fewer knew what it suggested about the future of our planet.
While studying Dune for a book I’m writing on the history of ecology, I was surprised to learn that Herbert didn’t learn about ecology as a student or as a journalist.
Instead, he was inspired to explore ecology by the conservation practices of the tribes of the Pacific Northwest. He learned about them from two friends in particular.
The first was Wilbur Ternyik, a descendant of Chief Coboway, the Clatsop leader who welcomed explorers Meriwether Lewis and William Clark when their expedition reached the West Coast in 1805. The second, Howard Hansen, was an art teacher and oral historian of the Quileute tribe.
Ternyik, who was also an expert field ecologist, took Herbert on a tour of Oregon’s dunes in 1958. There, he explained his work to build massive dunes of sand using beach grasses and other deep-rooted plants in order to prevent the sands from blowing into the nearby town of Florence—a terraforming technology described at length in Dune.
As Ternyik explains he wrote for the US Department of Agriculture, his work in Oregon was part of an effort to heal landscapes scarred by European colonization, especially the large river jetties built by early settlers.
These structures disturbed coastal currents and created vast expanses of sand, turning stretches of the lush Pacific Northwest landscape into desert. This scenario is echoed in Dune, where the novel’s setting, the planet Arrakis, was similarly laid to waste by its first colonizers.
Hansen, who became the godfather to Herbert’s son, had closely studied the equally drastic impact logging had on the homelands of the Quileute people in coastal Washington. He encouraged Herbert to examine ecology carefully, giving him a copy of Paul B. Sears’ Where There Is Life, from which Herbert gathered one of his favorite quotes: “The highest function of science is to give us an understanding of consequences.”
The Fremen of Dune, who live in the deserts of Arrakis and carefully manage its ecosystem and wildlife, embody these teachings. In the fight to save their world, they expertly blend ecological science and Indigenous practices.
Treasures Hidden in the SandBut the work that had the most profound impact on Dune was Leslie Reid’s 1962 ecological study The Sociology of Nature.
In it, Reid explained ecology and ecosystem science for a popular audience, illustrating the complex interdependence of all creatures within the environment.
“The more deeply ecology is studied,” Reid writes, “the clearer does it become that mutual dependence is a governing principle, that animals are bound to one another by unbreakable ties of dependence.”
In the pages of Reid’s book, Herbert found a model for the ecosystem of Arrakis in a surprising place: the guano islands of Peru. As Reid explains, the accumulated bird droppings found on these islands were an ideal fertilizer. Home to mountains of manure described as a new “white gold” and one of the most valuable substances on Earth, the guano islands became in the late 1800s ground zero for a series of resource wars between Spain and several of its former colonies, including Peru, Bolivia, Chile, and Ecuador.
At the heart of the plot of Dune is a battle for control of the “spice,” a priceless resource. Harvested from the sands of the desert planet, it’s both a luxurious flavoring for food and a hallucinogenic drug that allows some people to bend space, making interstellar travel possible.
There is some irony in the fact that Herbert cooked up the idea of spice from bird droppings. But he was fascinated by Reid’s careful account of the unique and efficient ecosystem that produced a valuable—albeit noxious—commodity.
As the ecologist explains, frigid currents in the Pacific Ocean push nutrients to the surface of nearby waters, helping photosynthetic plankton thrive. These support an astounding population of fish that feed hordes of birds, along with whales.
In early drafts of Dune, Herbert combined all of these stages into the life cycle of the giant sandworms, football-field-sized monsters that prowl the desert sands and devour everything in their path.
Herbert imagines each of these terrifying creatures beginning as small, photosynthetic plants that grow into larger “sand trout.” Eventually, they become immense sandworms that churn the desert sands, spewing spice onto the surface.
In both the book and Dune: Part One, soldier Gurney Halleck recites a cryptic verse that comments on this inversion of marine life and arid regimes of extraction: “For they shall suck of the abundance of the seas and of the treasure hid in the sand.”
‘Dune’ RevolutionsAfter Dune was published in 1965, the environmental movement eagerly embraced it.
Herbert spoke at Philadelphia’s first Earth Day in 1970, and in the first edition of the Whole Earth Catalog—a famous DIY manual and bulletin for environmental activists—Dune was advertised with the tagline: “The metaphor is ecology. The theme revolution.”
In the opening of Denis Villeneuve’s first adaptation of Dune, Chani, an indigenous Fremen played by Zendaya, asks a question that anticipates the violent conclusion of the second film: “Who will our next oppressors be?”
The immediate cut to a sleeping Paul Atreides, the white protagonist who’s played by Timothée Chalamet, drives the pointed anti-colonial message home like a knife. In fact, both of Villeneuve’s movies expertly elaborate upon the anti-colonial themes of Herbert’s novels.
Unfortunately, the edge of their environmental critique is blunted. But Villeneuve has suggested that he might also adapt Dune Messiah for his next film in the series—a novel in which the ecological damage to Arrakis is glaringly obvious.
I hope Herbert’s prescient ecological warning, which resonated so powerfully with readers back in the 1960s, will be unsheathed in Dune 3.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
We’ve all been there: A tight deadline, an overnighter, and the next day we’re navigating life like zombies.
For fighter pilots, the last step isn’t an option. During active duty, these pilots need to be in tip-top shape mentally, even when they’re deprived of sleep (which can be often). Typically, the treatment is your everyday cup of joe. But for longer durations of sleep deprivation, pilots are also prescribed stronger stimulants.
But as anyone who’s ever had too much caffeine knows, there are side effects. You get jittery. Your hands start to shake. Your mood takes a nosedive as the effect wears off and irritability sets in. And then you crash.
Prescription stimulants, such as dextroamphetamine, have even more severe side effects. As the name suggests, they’re in the same family as methamphetamine—or “meth”—and come with the risk of addiction. These drugs last longer inside the body, so that when trying to sleep after a tiring day, they keep parts of the brain in a semi-alert state and mess with sleep schedules. People taking dextroamphetamine often need sedatives to counteract lingering effects, and the chemical regime takes a toll.
Over time, the lack of restorative sleep impacts memory, cognition, and reasoning. It also damages the immune system, metabolism, and overall health.
The drugs work in short bursts. What if there’s a way to turn them on and off at will—giving the brain just a tiny dose when needed and quickly shutting off its effect to allow a full night’s sleep?
One solution may be light-activated drugs. The Defense Advanced Research Projects Agency (DARPA) announced a project in June to develop these types of drugs to combat sleep deprivation for fighter pilots. So-called photopharmacological drugs would add a molecular “light switch” to drugs like dextroamphetamine.
Pulses of light activate the drugs in parts of the brain on demand. Non-targeted brain regions aren’t exposed to the active version and continue to work normally. Once the pilots are alert, another pulse of light shuts off the drug, giving the body time to break it down before bedtime.
To make this vision a reality, the new project, Alert WARfighter Enablement (AWARE), has two research arms. One will develop safe and effective dextroamphetamine that can be controlled with light. The second will focus on engineering a wearable “helmet” of sorts to direct light pulses toward regions of the brain involved in alertness and mental acuity.
“To achieve the beneficial effects of stimulants on alertness without the undesirable effects of the stimulant on mood, restorative sleep, and mental health, a new approach is needed to enable targeted activation of the drug,” Dr. Pedro Irazoqui, AWARE program manager, said in a press release.
Brain on AlertAfter a terrible night’s sleep, the first thing most of us reach for is coffee. Caffeine, its active ingredient, is the most widely used psychoactive substance in the world, with over 80 percent of people in North America drinking a cup of joe every morning.
While this is also the go-to solution for most fighter pilots, multiple countries have developed far stronger concoctions to keep their brigades awake. The most notorious is probably methamphetamine, first synthesized in the late 1800s. Best known by its street names—meth, crank, or speed—it was used during World War II to keep troops awake, before being outlawed across the globe. A safer spin-off, dextroamphetamine is currently prescribed to increase alertness and cognition. While effective, it can trigger both irritability and euphoric effects—a recipe for potential addiction.
The Air Force has approved other types of chemical drugs, such as modafinil, to battle fatigue too. Research in mice and people found these drugs can improve many cognitive functions—for example, navigating space, keeping multiple things in mind, and boosting overall alertness even when severely sleep-deprived. Unlike amphetamines, this group of drugs isn’t as addictive, with effects compared to drinking roughly 20 cups of coffee without the jitters. But they can produce pounding headaches, sweating, and in rare cases, hallucinations.
Light-activated drugs may be another option. First devised for cancer, these drugs have a molecular “light-switch” component that responds to pulses of light. The switch can be tagged onto conventional drugs, making it easy to adopt for existing medications—like, say, dextroamphetamine.
The “switch” component changes the chemical’s shape after being blasted by different wavelengths of light. Like transformers, one shape allows the chemical to grab onto its usual targets—the “active” state. Other configurations inactivate it.
Light-activated drugs have been tested in cells in petri dishes, but targeting the brain presents a hurdle—the skull. Shining a flashlight onto the skull obviously wouldn’t reach the brain, and invasive brain surgery is out of the question.
There’s a workaround. Infrared beams of light, at low levels, are safe in humans and can penetrate deep into tissues, including through the skull and into the brain. A previous study designed a number of potential switches that could be turned on with infrared light. And recent advances in AI could further aid the effort to develop “a photoswitchable version of dextroamphetamine that is inactive except in the presence of near-infrared light, which activates it,” wrote DARPA.
The other component is a programmable light-emitting helmet that transmits infrared light to the parts of the brain associated with wakefulness, reasoning, and decision-making. Over time, the stimulation could be personalized, so people only receive the necessary “dose” to stay alert.
The strategy still floods the brain with stimulants through a pill, but it limits the drug’s activity in time and space. With personalized dosages and light as a controller, it could lead to alertness without anxiety, irritability, or euphoria for each person. Switching the drug off also allows the brain to “rest” during a good night’s sleep.
A Three-Year PlanAWARE is slated to last over three years. DARPA is now welcoming proposals that fit the program’s two goals, including developing light-activated dextroamphetamine, dubbed “PhotoDex,” that can be rapidly turned on and off in the presence of near-infrared light. All candidate drugs will first be validated in animal models, before moving on to human trials.
For the headset, the project envisions a setup that emits infrared light and reliably activates necessary parts of the brain at millimeter-resolution, roughly that of an MRI-based brain scan. The timeline is about a year, and the agency did not specify how the headsets should be designed—for example, wired or wireless, how they’re powered, or what mechanism turns on the light beams.
“The idea is very ambitious, but recent advances in the creation of phototherapeutics and light-emitting devices offer good reason to be optimistic about the prospects,” Dr. David Lawrence at the University of North Carolina, who is not involved in the project, told New Scientist.
For now, photoswitchable drugs have not yet been approved for human use. If the AWARE program goes as planned, it could open a new avenue for targeted drug treatment, not just for battling sleep deprivation, but also for other brain disorders. The project is well aware of the ethical, legal, and societal implications, and has plans to discuss the technology’s use.
Image Credit: US Air Force photo by 2nd Lt. Samuel Eckholm
Rats are incredibly nimble creatures. They can climb up curtains, jump down tall ledges, and scurry across complex terrain—say, your basement stacked with odd-shaped stuff—at mind-blowing speed.
Robots, in contrast, are anything but nimble. Despite recent advances in AI to guide their movements, robots remain stiff and clumsy, especially when navigating new environments.
To make robots more agile, why not control them with algorithms distilled from biological brains? Our movements are rooted in the physical world and based on experience—two components that let us easily explore different surroundings.
There’s one major obstacle. Despite decades of research, neuroscientists haven’t yet pinpointed how brain circuits control and coordinate movement. Most studies have correlated neural activity with measurable motor responses—say, a twitch of a hand or the speed of lifting a leg. In other words, we know brain activation patterns that can describe a movement. But which neural circuits cause those movements in the first place?
We may find the answer by trying to recreate them in digital form. As the famous physicist Richard Feynman once said, “What I cannot create, I do not understand.”
This month, Google DeepMind and Harvard University built a realistic virtual rat to home in on the neural circuits that control complex movement. The rat’s digital brain, composed of artificial neural networks, was trained on tens of hours of neural recordings from actual rats running around in an open arena.
Comparing activation patterns of the artificial brain to signals from living, breathing animals, the team found the digital brain could predict the neural activation patterns of real rats and produce the same behavior—for example, running or rearing up on hind legs.
The collaboration was “fantastic,” said study author Dr. Bence Ölveczky at Harvard in a press release. “DeepMind had developed a pipeline to train biomechanical agents to move around complex environments. We simply didn’t have the resources to run simulations like those, to train these networks.”
The virtual rat’s brain recapitulated two regions especially important for movement. Tweaking connections in those areas changed motor responses across a variety of behaviors, suggesting these neural signals are involved in walking, running, climbing, and other movements.
“Virtual animals trained to behave like their real counterparts could provide a platform for virtual neuroscience…that would otherwise be difficult or impossible to experimentally deduce,” the team wrote in their article.
A Dense DatasetArtificial intelligence “lives” in the digital world. To power robots, it needs to understand the physical world.
One way to teach it about the world is to record neural signals from rodents and use the recordings to engineer algorithms that can control biomechanically realistic models replicating natural behaviors. The goal is to distill the brain’s computations into algorithms that can pilot robots and also give neuroscientists a deeper understanding of the brain’s workings.
So far, the strategy has been successfully used to decipher the brain’s computations for vision, smell, navigation, and recognizing faces, the authors explained in their paper. However, modeling movement has been a challenge. Individuals move differently, and noise from brain recordings can easily mess up the resulting AI’s precision.
This study tackled the challenges head on with a cornucopia of data.
The team first placed multiple rats into a six-camera arena to capture their movement—running around, rearing up, or spinning in circles. Rats can be lazy bums. To encourage them to move, the team dangled Cheerios across the arena.
As the rats explored the arena, the team recorded 607 hours of video and also neural activity with a 128-channel array of electrodes implanted in their brains.
They used this data to train an artificial neural network—a virtual rat’s “brain”—to control body movement. To do this, they first tracked how 23 joints moved in the videos and transferred them to a simulation of the rats’ skeletal movements. Our joints only bend in certain ways, and this step filters out what’s physically impossible (say, bending legs in the opposite direction).
The core of the virtual rat’s brain is a type of AI algorithm called an inverse dynamics model. Basically, it knows where “body” positions are in space at any given time and, from there, predicts the next movements leading to a goal—say, grab that coffee cup without dropping it.
Through trial-and-error, the AI eventually came close to matching the movements of its biological counterparts. Surprisingly, the virtual rat could also easily generalize motor skills to unfamiliar places and scenarios—in part by learning the forces needed to navigate the new environments.
The similarities allowed the team to compare real rats to their digital doppelgangers, when performing the same behavior.
In one test, the team analyzed activity in two brain regions known to guide motor skills. Compared to an older computational model used to decode brain networks, the AI could better simulate neural signals in the virtual rat across multiple physical tasks.
Because of this, the virtual rat offers a way to study movement digitally.
One long-standing question, for example, is how the brain and nerves command muscle movement depending on the task. Grabbing a cup of coffee in the morning, for example, requires a steady hand without any jerking action but enough strength to hold it steady.
The team tweaked the “neural connections” in the virtual rodent to see how changes in brain networks alter the final behavior—getting that cup of coffee. They found one network measure that could identify a behavior at any given time and guide it through.
Compared to lab studies, these insights “can only be directly accessed through simulation,” wrote the team.
The virtual rat bridges AI and neuroscience. The AI models here recreate the physicality and neural signals of living creatures, making them invaluable for probing brain functions. In this study, one aspect of the virtual rat’s motor skills relied on two brain regions—pinpointing them as potential regions key to guiding complex, adaptable movement.
A similar strategy could provide more insight into the computations underlying vision, sensation, or perhaps even higher cognitive functions such as reasoning. But the virtual rat brain isn’t a complete replication of a real one. It only captures snapshots of part of the brain. But it does let neuroscientists “zoom in” on their favorite brain region and test hypotheses quickly and easily compared to traditional lab experiments, which often take weeks to months.
On the robotics side, the method adds a physicality to AI.
“We’ve learned a huge amount from the challenge of building embodied agents: AI systems that not only have to think intelligently, but also have to translate that thinking into physical action in a complex environment,” said study author Dr. Matthew Botvinick at DeepMind in a press release. “It seemed plausible that taking this same approach in a neuroscience context might be useful for providing insights in both behavior and brain function.”
The team is next planning to test the virtual rat with more complex tasks, alongside its biological counterparts, to further peek inside the inner workings of the digital brain.
“From our experiments, we have a lot of ideas about how such tasks are solved,” said Ölveczky to The Harvard Gazette. “We want to start using the virtual rats to test these ideas and help advance our understanding of how real brains generate complex behavior.”
Image Credit: Google DeepMind
Imagine a future with nearly silent air taxis flying above traffic jams and navigating between skyscrapers and suburban droneports. Transportation arrives at the touch of your smartphone and with minimal environmental impact.
This isn’t just science fiction. United Airlines has plans for these futuristic electric air taxis in Chicago and New York. The US military is already experimenting with them. And one company has a contract to launch an air taxi service in Dubai as early as 2025. Another company hopes to defy expectations and fly participants at the 2024 Paris Olympics.
Backed by billions of dollars in venture capital and established aerospace giants that include Boeing and Airbus, startups across the world such as Joby, Archer, Wisk, and Lilium are spearheading this technological revolution, developing electric vertical takeoff and landing (eVTOL) aircraft that could transform the way we travel.
Electric aviation promises to alleviate urban congestion, open up rural areas to emergency deliveries, slash carbon emissions, and offer a quieter, more accessible form of short-distance air travel.
But the quest to make these electric aircraft ubiquitous across the globe instead of just playthings for the rich is far from a given. Following the industry as executive director of the Oklahoma Aerospace Institute for Research and Education provides a view of the state of the industry. Like all great promised paradigm shifts, numerous challenges loom—technical hurdles, regulatory mazes, the crucial battle for public acceptance, and perhaps physics itself.
Why Electrify Aviation?Fixed somewhere between George Jetson’s flying car and the gritty taxi from The Fifth Element, the allure of electric aviation extends beyond gee-whiz novelty. It is rooted in its potential to offer efficient, eco-friendly alternatives to ground transportation, particularly in congested cities or hard-to-reach rural regions.
While small electric planes are already flying in a few countries, eVTOLs are designed for shorter hops—the kind a helicopter might make today, only more cheaply and with less impact on the environment. The eVTOL maker Joby purchased Uber Air to someday pair the company’s air taxis with Uber’s ride-hailing technology.
In the near term, once eVTOLs are certified to fly as commercial operations, they are likely to serve specific, high-demand routes that bypass road traffic. An example is United Airlines’ plan to test Archer’s eVTOLs on short hops from Chicago to O’Hare International Airport and Manhattan to Newark Liberty International Airport.
While some applications initially might be restricted to military or emergency use, the goal of the industry is widespread civil adoption, marking a significant step toward a future of cleaner urban mobility.
The Challenge of Battery PhysicsOne of the most significant technical challenges facing electric air taxis is the limitations of current battery technology.
Today’s batteries have made significant advances in the past decade, but they don’t match the energy density of traditional hydrocarbon fuels currently used in aircraft. This shortcoming means that electric air taxis cannot yet achieve the same range as their fossil-fueled counterparts, limiting their operational scope and viability for long-haul flights. Current capabilities still fall short of traditional transportation. However, with ranges from dozens of miles to over 100 miles, eVTOL batteries provide sufficient range for intracity hops.
The quest for batteries that offer higher energy densities, faster charging times, and longer life cycles is central to unlocking the full potential of electric aviation.
While researchers are working to close this gap, hydrogen presents a promising alternative, boasting a higher energy density and emitting only water vapor. However, hydrogen’s potential is tempered by significant hurdles related to safe storage and infrastructure capable of supporting hydrogen-fueled aviation. That presents a complex and expensive logistics challenge.
And, of course, there’s the specter of the last major hydrogen-powered aircraft. The Hindenburg airship caught fire in 1937, but it still looms large in the minds of many Americans.
Regulatory HurdlesEstablishing a “4D highways in the sky” will require comprehensive rules that encompass everything from vehicle safety to air traffic management. For the time being, the US Federal Aviation Administration is requiring that air taxis include pilots serving in a traditional role. This underscores the transitional phase of integrating these vehicles into airspace, highlighting the gap between current capabilities and the vision of fully autonomous flights.
The journey toward autonomous urban air travel is fraught with more complexities, including the establishment of standards for vehicle operation, pilot certification, and air traffic control. While eVTOLs have flown hundreds of test flights, there have also been safety concerns after prominent crashes involving propeller blades failing on one in 2022 and the crash of another in 2023. Both were being flown remotely at the time.
The question of who will manage these new airways remains an open discussion—national aviation authorities such as the FAA, state agencies, local municipalities, or some combination thereof.
Creating the FutureIn the long term, the vision for electric air taxis aligns with a future where autonomous vehicles ply the urban skies, akin to scenes from Back to the Future. This future, however, not only requires technological leaps in automation and battery efficiency but also a societal shift in how people perceive and accept the role of autonomous vehicles, both cars and aircraft, in their daily lives. Safety is still an issue with autonomous vehicles on the ground.
The successful integration of electric air taxis into urban and rural environments hinges on their ability to offer safe, reliable, and cost-effective transportation.
As these vehicles overcome the industry’s many hurdles, and regulations evolve to support their operation in the years ahead, I believe we could witness a profound transformation in air mobility. The skies offer a new layer of connectivity, reshaping cities and how we navigate them.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Joby
Thousands of people a year die while waiting for an organ transplant. Early experiments in xenotransplantation are raising hopes this could soon be a thing of the past.
In the US, 100,000 people are currently on the organ transplant waiting list, and 17 of them die every day before receiving an organ. The persistent shortage of organ donors has long led doctors to flirt with the idea of xenotransplantation, a procedure where tissue or an organ from an animal is transplanted into a human.
Early experiments were largely unsuccessful and ethically questionable, though, and the idea remained firmly on the fringes of the medical world. That’s largely due to the high risk of rejection. This is a problem for human transplants too, but it’s much more risky when using organs from other species.
But the advent of increasingly powerful and precise genetic engineering technologies such as CRISPR have ushered the idea from the shadows. The ability to make edits to the donor animals’ DNA to prevent the production of biomolecules known to induce immune responses in humans has raised hopes the approach may be viable after all.
In recent years, a handful of pioneering experiments in humans have demonstrated that genetically engineered pig organs can at least temporarily function smoothly in the human body. Medical complications, organ rejections, and patient deaths have meant none of these procedures have provided a long-term solution, but the results so far have been promising.
“At Massachusetts General Hospital alone, there are over 1,400 patients on the waiting list for a kidney transplant,” Leonardo Riella, who led the surgical team at Mass General that transplanted a pig kidney into a patient, said in a press release earlier this year.
“Some of these patients will unfortunately die or get too sick to be transplanted due to the long waiting time on dialysis. I am firmly convinced that xenotransplantation represents a promising solution to the organ shortage crisis.”
In 2021, in the first human experiment involving a genetically engineered pig organ, doctors transplanted a kidney into a patient who was already brain dead. The team knocked out a gene for a molecule called alpha-gal—which causes organ rejection—in the donor pig. The surgery appeared to be a success: The kidney produced urine and showed no signs of rejection, but the patient was only kept alive for 54 hours.
The following year, a patient with terminal heart failure received a genetically modified pig heart and initially seemed to do well, but then passed away 60 days later. While it’s not entirely clear why he died, the doctors found that pre-screening failed to flag a pathogen called porcine cytomegalovirus that was found in his heart afterwards, which could have contributed. He’d also been given an antibody treatment that had reacted with the heart.
Then earlier this year, two kidney disease patients who were ineligible for normal transplants received gene-edited pig kidneys from donor pigs bred by biotech firm eGenesis. Using CRISPR, the company made 69 edits that removed some pig genes, added some human ones, and reduced the risk of latent virus in the organ reactivating and harming the patient.
The procedures appeared to go well. Doctors even discharged the first patient after determining the kidney was functioning well, and he no longer needed dialysis. Two months later he passed away, but he had other underlying health issues, and the hospital said there was no indication his death was the result of the transplant.
The second patient had to have the kidney removed after 47 days due to “unique challenges” stemming from the fact she had also had a mechanical heart pump implanted just before the transplantation. There were no signs of rejection, but the kidney started losing function because her heart was not able to pump blood with enough pressure, the researchers said.
The most recent experiment was announced in May, when Chinese researchers said they had transplanted a liver from a genetically modified pig into a 71-year-old man with liver cancer. While details of the procedure are limited, the team claimed the man was “doing very well” more than two weeks after surgery.
While most of these experiments have been short-lived, the fact that only two cases saw the transplanted organ fail—one of which was due to external complications—is a promising sign. For ethical reasons, doctors have only been able to experiment with patients whose chances of survival were already slim.
But it does mean that we have little idea whether xenotransplantation could be a viable long-term solution for patients. There is also some concern that implanting organs from other animals into humans could make it easier for pathogens to jump between species, potentially creating the risk of new pandemics.
Other researchers are investigating whether, instead of transplanting pig organs into humans, we could grow human organs in pigs. Last September, researchers announced they’d transplanted human stem cells into pig embryos where they then grew into rudimentary kidneys.
This approach is a long way from human trials though, so for the time being, xenotransplantation seems like a more promising way to bring down transplant wait times. While it’s still early days, the promising early results suggest we may not be far from a future where replacement organs can be grown to order.
Image Credit: Massachusetts General Hospital
If you had to sum up what has made humans such a successful species, it’s teamwork. There’s growing evidence that getting AIs to work together could dramatically improve their capabilities too.
Despite the impressive performance of large language models, companies are still scrabbling for ways to put them to good use. Big tech companies are building AI smarts into a wide-range of products, but none has yet found the killer application that will spur widespread adoption.
One promising use case garnering attention is the creation of AI agents to carry out tasks autonomously. The main problem is that LLMs remain error-prone, which makes it hard to trust them with complex, multi-step tasks.
But as with humans, it seems two heads are better than one. A growing body of research into “multi-agent systems” shows that getting chatbots to team up can help solve many of the technology’s weaknesses and allow them to tackle tasks out of reach for individual AIs.
The field got a significant boost last October when Microsoft researchers launched a new software library called AutoGen designed to simplify the process of building LLM teams. The package provides all the necessary tools to spin up multiple instances of LLM-powered agents and allow them to communicate with each other by way of natural language.
Since then, researchers have carried out a host of promising demonstrations.
In a recent article, Wired highlighted several papers presented at a workshop at the International Conference on Learning Representations (ICLR) last month. The research showed that getting agents to collaborate could boost performance on math tasks—something LLMs tend to struggle with—or boost their reasoning and factual accuracy.
In another instance, noted by The Economist, three LLM-powered agents were set the task of defusing bombs in a series of virtual rooms. The AI team performed better than individual agents, and one of the agents even assumed a leadership role, ordering the other two around in a way that improved team efficiency.
Chi Wang, the Microsoft researcher leading the AutoGen project, told The Economist that the approach takes advantage of the fact most jobs can be split up into smaller tasks. Teams of LLMs can tackle these in parallel rather than churning through them sequentially, as an individual AI would have to do.
So far, setting up multi-agent teams has been a complicated process only really accessible to AI researchers. But earlier this month, the Microsoft team released a new “low-code” interface for building AI teams called AutoGen Studio, which is accessible to non-experts.
The platform allows users to choose from a selection of preset AI agents with different characteristics. Alternatively, they can create their own by selecting which LLM powers the agent, giving it “skills” such as the ability to fetch information from other applications, and even writing short prompts that tell the agent how to behave.
So far, users of the platform have put AI teams to work on tasks like travel planning, market research, data extraction, and video generation, say the researchers.
The approach does have its limitations though. LLMs are expensive to run, so leaving several of them to natter away to each other for long stretches can quickly become unsustainable. And it’s unclear whether groups of AIs will be more robust to mistakes, or whether they could lead to cascading errors through the entire team.
Lots of work needs to be done on more prosaic challenges too, such as the best way to structure AI teams and how to distribute responsibilities between their members. There’s also the question of how to integrate these AI teams with existing human teams. Still, pooling AI resources is a promising idea that’s quickly picking up steam.
Image Credit: Mohamed Nohassi / Unsplash
A squishy, fatty, beige-colored organ covered with grooves and ridges, the brain doesn’t look all that impressive on the surface.
But hidden underneath are up to 100 billion neurons and 100 trillion synapses—the connections between neurons that form networks—densely packed in a squishy three-pound organ that controls our thoughts, feelings, movement, memories, and sense of self.
For the past two decades, scientists have carefully dissected the internal neural connections and workings of the brain by carefully chopping it up into paper-thin pieces. From there, they’ve built multiple maps of the brain’s cellular population, architecture, connections, and gene expression. Like charting the landscape of a new world, these maps have been consolidated into what amounts to a Google Maps for the brain. These atlases allow us to decipher brain function, bridging genetic expression to cell functions, network connections, and behavior.
At least for rodents and other animals. Mapping the brain is incredibly difficult and time-consuming. A small chunk of a mouse’s brain, when imaged at single-cell resolution, takes years to process, scan, and reconstruct into 3D computer models. Any trip-ups during the process ruins the product. Mapping the human brain, much larger in size, is far more difficult.
This month, a team from MIT developed a “holistic” brain-mapping platform that captures the anatomy of large slices of the human brain with unprecedented resolution and speed, slashing a process that normally takes between a week and a month to a few days.
They used the platform to image an Alzheimer’s brain, after physically expanding brain tissues with a hydrogel. The automated system sliced, imaged, and automatically stitched the images together and found myriads of cellular changes and problems with neural connections, including inflammation.
Compared to previous brain mapping projects, which often require months or years, the new platform mapped different levels of the brain’s physical makeup—from synapses to local neural circuits and brain-wide connections in slabs of human brain tissue—in just a few days.
“We performed holistic imaging of human brain tissues at multiple resolutions from single synapses to whole brain hemispheres, and we have made that data available,” study author Kwanghun Chung said in a press release.
To be clear, the technology has only been used on slabs of human brain tissue and hasn’t yet charted the entire brain’s neurons and connections. But “this technology pipeline really enables us to analyze the human brain at multiple scales. Potentially this pipeline can be used for fully mapping human brains,” said Chung.
Three-Way UpgradeHere’s how brain mapping technology usually works. Whole brains are sliced into wafer-thin pieces on a machine called a vibratome—think of it as a souped-up deli meat slicer.
Most vibratomes are tailored for cutting smaller brains, such as those from rodents. Trying to cut a slice of human brain tissue with a standard vibratome is akin to cutting a sandwich filled with deli meats, arugula, and avocado with a dull knife. Picture the meats as neurons, arugula as blood vessels, and avocado as supporting structures. All components get distorted and squished, making it nearly impossible to realign them into a brain map.
As a workaround, the team developed MEGAtome, a vibratome that can slice through large and soft human brain specimens without tearing or squishing. Compared to a state-of-the-art device, MEGAtome vibrates at higher frequencies, lowering the chances of “angled cuts” and minimizing distortion of the neural connections that eventually need to be realigned.
The next step is treating brain samples. Previously, scientists found a way to physically expand the brain in size—so that its details are easier to see under the microscope—using a gel commonly found inside diapers. The team adapted the idea and developed a recipe that embedded human brain slices in a squishy hydrogel, transforming brain tissue into a stretchy brain-gel hybrid tissue that could easily withstand mechanical pressure—such as that from the vibratome blade—while maintaining its shape. Some tissues expanded over four times their normal size.
In one demo, the team used MEGAtome to slice up a human brain hemisphere, generating 40 relatively thick slabs in just 8 hours.
To capture cellular identities, the team stained each brain slab with dyes that grab onto different types of proteins to highlight different types of brain cells. Some colors signal mature neurons; others mark non-neuronal cells, called astrocytes. Although these cells can’t transmit electrical signals, they support neurons by releasing chemicals to regulate their function. Also present were the brain’s immune cells and blood vessels.
The system imaged a four-millimeter-thick slab of human brain—thicker than the average cortex—at the synapse-level in just six hours. Using the technique, “a whole brain hemisphere can be imaged at single-cell resolution in ~100 hours,” wrote the team.
The third upgrade is software. Recreating a 3D brain structure means aligning individual slices like piecing together a puzzle. The team first used blood vessels as a guide to roughly align each piece. They then zeroed in on individual neural connections to further perfect the map.
Previously, slices could only tolerate one round of dyes. With the new protocol, they withstood at least seven rounds of rinsing and re-dying, allowing scientists to capture multiple protein changes in the same tissue at single-cell resolution.
“This technology pipeline really enables us to extract all these important features from the same brain in a fully integrated manner,” Chung said in the press release.
Alzheimer’s and BeyondAs a proof of concept, the team used the new system to analyze two donated brains: One from a healthy 61-year-old female donor and the other from an octogenarian with Alzheimer’s.
The team sliced both brains with MEGAtome and dyed multiple slabs. Compared to the healthy brain, the Alzheimer’s brain had 46.5 percent fewer neurons, especially in a frontal part of the brain that’s important for making decisions.
“Connectivity is impaired [here] in later stages of Alzheimer’s disease,” wrote the team.
The team also found increased inflammation in the Alzheimer’s brain, along with a build-up of protein gunk outside cells—potentially damaging those neurons’ ability to connect to others.
With just one sample, the results don’t offer conclusions about how neurons change in Alzheimer’s disease. But that’s not the point. The platform allows scientists to quickly and efficiently probe larger brain tissues—not just in humans, but also pigs and non-human primates—to further our understanding of neural networks in the brain and what happens to them in health and disease.
Image Credit: Image of the orbitofrontal cortex from an Alzheimer’s donated brain. Chung Lab/MIT Picower Institute
When Angela received her first shot at the Lombardi Comprehensive Cancer Center in early 2020, Covid-19 was months away. Far from a household name, mRNA vaccines were mostly relegated to lab studies.
Yet the jab she received was made of the same technology. A melanoma patient, Angela had multiple malignant moles removed. Alongside an established immune-stimulating drug, the hope was the duo could fight off any residual cancerous cells and slash the chances of relapse.
Scientists have long sought cancer vaccines that prevent the pesky cells from growing back. Like those targeting viruses, the vaccines would train the body’s immune system to recognize the cancerous cells and attack and eliminate them before they could grow and spread.
Despite decades of research into cancer vaccines, the dream has mostly failed. One reason is that every cancer, in every person, is different. So is each person’s immune system. Tailoring vaccines to neutralize cancers for each patient would not only be expensive, but sometimes impossible due to how long they’d take to develop—time is not on cancer patients’ sides.
In contrast, mRNA vaccines are far speedier to build. After they were removed, Angela’s malignant moles were analyzed for specific cancerous “fingerprints” or neoantigens. Based on these proteins, scientists at Moderna—known for their Covid-19 vaccines—built a custom mRNA cancer vaccine to train her immune system to prevent her own cancer from recurring.
Angela is part of clinical trial led by pharmaceutical companies Moderna and Merck to see if malignant skin cancer came back in patients given the treatment. Compared to a standard immunotherapy drug alone, adding a custom mRNA vaccine reduced the chances of cancer returning by roughly 50 percent and increased lifespan.
To be clear, the vaccines don’t protect a person from getting cancer in the first place. Rather, they teach the immune system to recognize residual malignant cells and prevent them from returning. The companies have launched Phase 3 clinical studies in people with melanoma and a type of lung cancer, with earlier stage clinical trials for other cancer cell types in the works.
Getting PersonalLike healthy cells, cancerous cells are dotted with all kinds of proteins on their surfaces. Dubbed “neoantigens,” these proteins differentiate cancer cells from healthy ones, making them attractive targets for therapies. And like fingerprints, neoantigens often differ between different cancer types and individuals, raising the possibility of personalized treatments.
That’s the idea behind cancer vaccines. They work like vaccines against infectious diseases. Parts of the invader—in cancer’s case, its unique neoantigens—are mixed with chemicals that stimulate the immune system. Once injected, the concoction directs the immune system to specifically attack cells with the neoantigen and eliminate the threat.
Compared to chemotherapy—notorious for its horrible side effects—cancer vaccines target a person’s own constellation of neoantigens, which in theory limits damage.
In 2017, two small clinical trials offered a glimpse that these vaccines could work in humans. Both studies targeted melanoma, a mole-like type of cancer that can quickly spread and recur.
After surgical removal, the researchers sequenced the genes of each malignant mole and selected up to 20 different protein fragments for each person to develop into vaccines. In one study, the shots kept the cancer at bay in four out of six patients for at least two years. The two who saw their cancer come back quickly entered remission after treatment with a drug that stimulates their immune system.
Another study enrolled 13 patients, eight with no visible tumors and five whose cancer had already spread. A personalized vaccine encoded 10 neoantigens for each person and used a virus to shuttle the mixture into cells. While successful for the first group, who remained cancer-free for over a year, results were mixed in the second group. For these patients, the cancer shrank but resurged in some, while others went into remission after treatment with the same immune-stimulating drug.
“It’s potentially a game changer,” Dr. Cornelis Melief at Leiden University Medical Center, who was not involved in the study, told Nature at the time.
Yet the field still faced a roadblock: Cancer vaccines are expensive to make and often require time—time that patients don’t always have.
An mRNA WorldEnter mRNA vaccines. Best known for battling Covid-19, these vaccines can be designed and manufactured at a fraction of the time and cost of their traditional protein-based counterparts.
A cancer vaccine based on mRNA follows a similar path to previous iterations, but with a few upgrades.
The patient’s skin cancer is rapidly sequenced for its genes after removal. Selection of neoantigen genes is key. Not all of them can be recognized by the immune system. Machine learning algorithms, trained on expanding databases of cancer-related mutations, sort through the data to identify the neoantigen genes most likely to stimulate the immune system. Moderna picks up to 34 candidates with the highest chances.
Like in Covid-19 vaccines, the selected genes are then translated into mRNA and encapsulated in fatty bubbles. Once injected, the mRNA commandeers the cell’s protein-making machinery to pump out neoantigens. These, in turn, train the immune system to sniff out the foe.
The mRNA vaccines weren’t used alone, however. Taking a note from previous studies, the companies added an immune-stimulating drug to boost efficacy.
The results from a three-year ongoing trial were announced earlier this month. The combination, compared to the drug alone, reduced the risk of cancers returning and death by 49 percent. They also decreased the risk of the cancer spreading by 62 percent. Living cancer-free for at least two and a half years, those treated with the combo saw a boost in their chances of survival with the addition of the mRNA vaccine. The results mirror those from a previous analysis, led by Dr. Jeffrey Weber at New York University Langone Health, who is overseeing the trial, dubbed KEYNOTE-942.
“At the end of the day, you realize, ‘Damn! This combination seems to have activity,’” Weber told Nature.
Although the results are promising, the combo isn’t for everyone. Later-stage cancers, especially those which have already spread, don’t respond well to the treatment. These tumors also rapidly grow—compared to their earlier counterparts—robbing scientists of precious time to develop the personalized vaccine.
Others are doing similar work. BioNTech has partnered with Genentech to develop vaccines targeting up to 20 neoantigens for notoriously aggressive pancreatic cancer. The vaccine worked for only half of the participants; even then, a fraction of the immune system only recognized one neoantigen. Nonetheless, vaccinated patients lived longer cancer-free after treatment when assessed 18 months after treatment.
Cancer vaccines are having a renaissance, but there’s much left to learn. Figuring out how to choose the right neoantigens is first and foremost. One team, for example, is verifying that immune cells in blood samples from patients actually recognize the selected neoantigens.
Other cancer types are already on the docket as potential next targets, including those that affect cells lining the skin, lungs, and digestive tracts, or those involved in kidney cancer.
As for Angela, the initial flu-like symptoms from the treatment were worth it. In her mid-40s, her cancer has been gone for three years. When asked if it’s because of the vaccine or drug, she told Nature: “I’m just happy to be cancer-free.”
Image Credit: Diana Polekhina / Unsplash
During her chemistry Nobel Prize lecture in 2018, Frances Arnold said, “Today we can for all practical purposes read, write, and edit any sequence of DNA, but we cannot compose it.”
That isn’t true anymore.
Since then, science and technology have progressed so much that artificial intelligence has learned to compose DNA, and with genetically modified bacteria, scientists are on their way to designing and making bespoke proteins.
The goal is that with AI’s design talents and gene editing’s engineering abilities, scientists can modify bacteria to act as mini-factories producing new proteins that can reduce greenhouse gases, digest plastics, or act as species-specific pesticides.
As a chemistry professor and computational chemist who studies molecular science and environmental chemistry, I believe that advances in AI and gene editing make this a realistic possibility.
Gene Sequencing: Reading Life’s RecipesAll living things contain genetic materials—DNA and RNA—that provide the hereditary information needed to replicate themselves and make proteins. Proteins constitute 75 percent of human dry weight. They make up muscles, enzymes, hormones, blood, hair, and cartilage. Understanding proteins means understanding much of biology. The order of nucleotide bases in DNA, or RNA in some viruses, encodes this information, and genomic sequencing technologies identify the order of these bases.
The Human Genome Project was an international effort that sequenced the entire human genome between 1990 to 2003. Thanks to rapidly improving technologies, it took seven years to sequence the first 1 percent of the genome and another seven years for the remaining 99 percent. By 2003, scientists had the complete sequence of 3 billion nucleotide base pairs coding for the 20,000 to 25,000 genes in the human genome.
However, understanding the functions of most proteins and correcting their malfunctions remained a challenge.
AI Learns ProteinsEach protein’s shape is critical to its function and is determined by the sequence of its amino acids, which is in turn determined by the gene’s nucleotide sequence. Misfolded proteins have the wrong shape and can cause illnesses such as neurodegenerative diseases, cystic fibrosis, and Type 2 diabetes. Understanding these diseases and developing treatments requires knowledge of protein shapes.
Before 2016, the only way to determine the shape of a protein was through X-ray crystallography, a laboratory technique that uses the diffraction of X-rays by single crystals to determine the precise arrangement of atoms and molecules in three dimensions in a molecule. At that time, the structure of about 200,000 proteins had been determined by crystallography, costing billions of dollars.
AlphaFold, a machine learning program, used these crystal structures as a training set to determine the shape of the proteins from their nucleotide sequences. And in less than a year, the program calculated the protein structures of all 214 million genes that have been sequenced and published. The protein structures AlphaFold determined have all been released in a freely available database.
To effectively address noninfectious diseases and design new drugs, scientists need more detailed knowledge of how proteins, especially enzymes, bind small molecules. Enzymes are protein catalysts that enable and regulate biochemical reactions.
AlphaFold3, released May 8, 2024, can predict protein shapes and the locations where small molecules can bind to these proteins. In rational drug design, drugs are designed to bind proteins involved in a pathway related to the disease being treated. The small molecule drugs bind to the protein binding site and modulate its activity, thereby influencing the disease path. By being able to predict protein binding sites, AlphaFold3 will enhance researchers’ drug development capabilities.
AI + CRISPR = Composing New ProteinsAround 2015, the development of CRISPR technology revolutionized gene editing. CRISPR can be used to find a specific part of a gene, change or delete it, make the cell express more or less of its gene product, or even add an utterly foreign gene in its place.
In 2020, Jennifer Doudna and Emmanuelle Charpentier received the Nobel Prize in chemistry “for the development of a method (CRISPR) for genome editing.” With CRISPR, gene editing, which once took years and was species specific, costly, and laborious, can now be done in days and for a fraction of the cost.
AI and genetic engineering are advancing rapidly. What was once complicated and expensive is now routine. Looking ahead, the dream is of bespoke proteins designed and produced by a combination of machine learning and CRISPR-modified bacteria. AI would design the proteins, and bacteria altered using CRISPR would produce the proteins. Enzymes produced this way could potentially breathe in carbon dioxide and methane while exhaling organic feedstocks or break down plastics into substitutes for concrete.
I believe that these ambitions are not unrealistic, given that genetically modified organisms already account for 2 percent of the US economy in agriculture and pharmaceuticals.
Two groups have made functioning enzymes from scratch that were designed by differing AI systems. David Baker’s Institute for Protein Design at the University of Washington devised a new deep-learning-based protein design strategy it named “family-wide hallucination,” which they used to make a unique light-emitting enzyme. Meanwhile, biotech startup Profluent, has used an AI trained from the sum of all CRISPR-Cas knowledge to design new functioning genome editors.
If AI can learn to make new CRISPR systems as well as bioluminescent enzymes that work and have never been seen on Earth, there is hope that pairing CRISPR with AI can be used to design other new bespoke enzymes. Although the CRISPR-AI combination is still in its infancy, once it matures it is likely to be highly beneficial and could even help the world tackle climate change.
It’s important to remember, however, that the more powerful a technology is, the greater the risks it poses. Also, humans have not been very successful at engineering nature due to the complexity and interconnectedness of natural systems, which often leads to unintended consequences.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Gerd Altmann / Pixabay
Autonomous vehicles are understandably held to incredibly high safety standards, but it’s sometimes forgotten that the true baseline is the often dangerous driving of humans. Now, new research shows that self-driving cars were involved in fewer accidents than humans in most scenarios.
One of the main arguments for shifting to autonomous vehicles is the prospect of taking human error out of driving. Given that more than 40,000 people die in car accidents every year in the US, even a modest improvement in safety could make a huge difference.
But self-driving cars have been involved in a number of accidents in recent years that have led to questions over their safety and caused some larger companies like Cruise to scale back their ambitions.
Now though, researchers have analyzed thousands of accident reports from incidents involving both autonomous vehicles and human drivers. Their results, published in Nature Communications, suggest that in most situations autonomous vehicles are actually safer than humans.
The team from the University of Central Florida focused their study on California, where the bulk of autonomous vehicles testing is going on. They gathered 2,100 reports of accidents involving self-driving cars from databases maintained by the National Highway Traffic Safety Administration, the California Department of Motor Vehicles, and news reports.
They then compared them against 35,000 reports of incidents involving human drivers compiled by the California Highway Patrol. The team used an approach called “matched case-control analysis” in which they attempted to find pairs of crashes involving humans and self-driving cars that otherwise had very similar characteristics.
This makes it possible to control for all the other variables that could contribute to a crash and investigate the impact of the “driver” on the likelihood of a crash occurring. The team found 548 such matches, and when they compared the two groups, they found self-driving cars were safer than human drivers in most of the accident scenarios they looked at.
There are some significant caveats though. The researchers also discovered that autonomous vehicles were over five times more likely to be involved in an accident at dawn or dusk and nearly twice as likely when making a turn.
The former is likely due to limitations in imaging sensors, while J. Christian Gerdes, from Stanford University, told IEEE Spectrum that their trouble with turns is probably due to limited ability to predict the behavior of other drivers.
There were some bright spots for autonomous vehicles too though. They were roughly half as likely to be involved in a rear-end accident and just one-fifth as likely to be involved in a broadside collision.
The researchers also found that the chance of a self-driving vehicle crashing in rain or fog was roughly a third of that for a human driver, which they put down to the vehicles’ reliance on radar sensors that are largely immune to bad weather.
How much can be read into these results is a matter of debate. The authors admit there is limited data on autonomous vehicle crashes, which limits the scope of their findings. George Mason University’s Missy Cummings also told New Scientist that accident reports from self-driving companies are often biased, seeking to pin the blame on human drivers even when the facts don’t support it.
Nonetheless, the study is an important first step in quantifying the potential safety benefits of autonomous vehicle technology and has highlighted some important areas where progress is still needed. Only by taking a clear-eyed look at the numbers can policymakers make sensible decisions about where and when this technology should be deployed.
Image Credit: gibblesmash asdf / Unsplash
FUTUREIlya Sutskever Has a New Plan for Safe Superintelligence
Ashlee Vance | Bloomberg“For the past several months, the question ‘Where’s Ilya?’ has become a common refrain within the world of artificial intelligence. …Now Sutskever is introducing [a new] project, a venture called Safe Superintelligence Inc. aiming to create a safe, powerful artificial intelligence system within a pure research organization that has no near-term intention of selling AI products or services. In other words, he’s attempting to continue his work without many of the distractions that rivals such as OpenAI, Google and Anthropic face.”
BIOTECHHow AI Is Revolutionizing Drug Development
Steve Lohr | The New York Times“Most of the early business uses of generative AI, which can produce everything from poetry to computer programs, have been to help take the drudgery out of routine office tasks, customer service and code writing. Yet drug discovery and development is a huge industry that experts say is ripe for an AI makeover. AI is a ‘once-in-a-century opportunity’ for the pharmaceutical business, according to the consulting firm McKinsey & Company.”
TECHStarlink Mini Brings Space Internet to Backpackers
Thomas Ricker | The Verge“SpaceX’s Starlink internet-from-space service is already available for boats, planes, vanlifers, Amazonian villages, and rural homes in over 75 countries—now it’s coming to backpackers. The new compact DC-powered Starlink Mini is about the size of a thick laptop and integrates the Wi-Fi router right inside the dish. And despite using less power than other Starlink terminals, it can still deliver speeds over 100 Mbps.”
VIRTUAL REALITYApple’s Vision Pro Team Is Reportedly Focused on Building a Cheaper Headset
Jay Peters | The Verge“Apple may no longer be working on a new high-end Vision headset amid slowing sales of the Vision Pro, according to a new report from The Information. Instead, Apple has apparently been finding ways to reduce the cost of components for the first model and is working on a cheaper Vision headset that it aims to ship by the end of 2025.”
ARTIFICIAL INTELLIGENCEWe’re Still Waiting for the Next Big Leap in AI
Will Knight | Wired“More than a year after GPT-4 spurred a frenzy of new investment in AI, it may be turning out to be more difficult to produce big new leaps in machine intelligence. With GPT-4 and similar models trained on huge swathes of online text, imagery, and video, it is getting more difficult to find new sources of data to feed to machine-learning algorithms. Making models substantially larger, so they have more capacity to learn, is expected to cost billions of dollars.”
ROBOTICSLet Slip the Robot Dogs of War
Jared Keller | Wired“The Chinese military recently unveiled a new kind of battle buddy for its soldiers: a ‘robot dog’ with a machine gun strapped to its back. …China’s demonstration clearly rankled international observers, prompting at least one American lawmaker to call on the US Defense Department for a report on ‘rifle-toting robot dogs’ and their potential national security implications. But if the Chinese military is pioneering the weaponization of robot dogs, then the United States military isn’t far behind.”
AUTOMATIONWaabi’s GenAI Promises to Do So Much More Than Power Self-Driving Trucks
Rebecca Bellan | TechCrunch“‘This technology is extremely, extremely powerful,’ said [Waabi founder and CEO Raquel] Urtasun, who spoke to TechCrunch via video interview, a whiteboard full of hieroglyphic-looking formulas behind her. ‘It has this amazing ability to generalize, it’s very flexible, and it’s very fast to develop. And it’s something that we can expand to do much more than trucking in the future. …This could be robotaxis. This could be humanoids or warehouse robotics. This technology can solve any of those use cases.'”
ARTIFICIAL INTELLIGENCEApple, Microsoft Shrink AI Models to Improve Them
Shubham Agarwal | IEEE Spectrum“In the last few months…some of the largest tech companies, including Apple and Microsoft, have introduced small language models (SLMs). These models are a fraction of the size of their LLM counterparts and yet, on many benchmarks, can match or even outperform them in text generation. …[And] because SLMs don’t consume nearly as much energy as LLMs, they can also run locally on devices like smartphones and laptops (instead of in the cloud) to preserve data privacy and personalize them to each person.”
SENSORSMicrophone Made of Atom-Thick Graphene Could Be Used in Smartphones
Alex Wilkins | New Scientist“The main advantage of their graphene system is that it can be much smaller than a conventional microphone, says Verbiest, and would need a membrane that is just 10 micrometers across—an area 200 times smaller than a similarly performing conventional microphone.”
Image Credit: Li Zhang / Unsplash
Recently we have seen the launch of artificial intelligence programs such as SOUNDRAW and Loudly that can create musical compositions in the style of almost any artist.
We’re also seeing big stars use AI in their own work, including to replicate others’ voices. Drake, for instance, landed in hot water in April after he released a diss track that used AI to mimic the voice of late rapper Tupac Shakur. And with the new ChatGPT model, GPT-4o, things are set to reach a whole new level. Fast.
So is human-made music doomed?
While it’s true AI will likely disrupt the music industry and even transform how we engage with music, there are some good reasons to suggest human music-making isn’t going anywhere.
Technology and Music Have a Long HistoryOne could argue AI is essentially a tool aimed at making our lives easier. Humans been been crafting such tools for a long time, both in music and nearly every other domain.
We’ve been using technology to play music since the invention of the gramophone. And arguments about human musicians versus machines are at least as old as the self-playing piano, which came into use in the early 20th century.
More recently, sampling, DJ-ing, autotune technology, and AI-based mastering and production software have continued to fan debates over artistic originality.
But the new AI developments are different. Anyone can create a new track in any existing genre, with minimal effort. They can add instruments, change the music’s “vibe,” and even choose a virtual singer to sing their lyrics.
Given the industry’s longstanding exploitation of artists—particularly with the rise of streaming (and Spotify’s chief executive claiming music is almost free to create)—it’s easy to see why the latest developments in AI are frightening some musicians.
Music Is a Very Human ThingAt the same time, these developments offer an opportunity to reflect on why people make music in the first place. We have long used music to tell our stories, to express ourselves and our humanity. These stories teach us, heal us, energize us, and help shape our identities.
Can AI music do this? Maybe. But it’s unlikely to be able to speak to the human experience in the same way a human can—partly because it doesn’t understand it the way we do.
It’s also unlikely to be able to create new works outside of existing musical paradigms, as it relies on algorithms taking from existing material. So, we’ll likely still need our imaginations to create new musical ideas.
It also helps to note that music being controlled by “algorithms” actually isn’t a new concept. Mainstream pop artists have long had their music written for them by industry “hit makers” who use specific formulas.
It’s usually the musicians on the fringes, rather than the more commercial artists and products, who retain connection to music as a cultural practice and therefore push the development of new styles.
Perhaps the bigger question isn’t how musicians will compete against AI, but how we as a society should value the musicians who help create our musical worlds, and our very cultures.
Is this a task we’re happy to hand over to AI to save money? Or should such an important role be supported with job security and a fair wage, as is afforded to doctors, dentists, politicians, and teachers?
Art for Art’s SakeThere’s another much more fundamental reason why AI will not spell the end of human-made music. That’s because, as most musicians will tell you, making music feels good. It doesn’t always matter if it’s going to be sold, recorded, or even heard.
Consider mountain climbing as an example. Although we now have chair lifts, gondolas, funiculars, helicopters, planes, trains, and cars to take people to the top, people still love climbing mountains for the mental and physical benefits.
Similarly, playing music is a unique experience with benefits that extend far beyond making money. Ever since our ancestors first tapped rocks together in caves, music has connected us to others and to ourselves.
The health benefits are overwhelming (just look at the amount of evidence relating to choirs). The neurological benefits are also astounding, with no other activity lighting up as many parts of the brain.
No matter how good computers get at making music, active music engagement will always remain an important way to regulate our moods and nervous systems.
Also, if our relationships with organic foods, vinyl records, and sustainable fashion are anything to go by, we can assume there will always be a group of conscious consumers willing to pay more for human-made music.
AI as an OpportunityFurther, while AI will likely disrupt the music industry as we know it, it also has amazing potential for boosting creative freedom for new generations of artists.
It may soften the separation between “musician” and “non-musician,” arguably allowing more people access to all the associated wellbeing benefits of music-making.
There’s also enormous potential for music education, since students could use AI to explore all aspects of the musical process in one classroom.
In a health context, personalized songs and albums could have significant implications for music therapy by letting therapists create tracks tailored to their clients’ needs. For instance, a therapist might want to produce a song a client has no prior association with to avoid music-related triggers during therapy.
AI-assisted music is already being used in psychedelic therapy to create, curate, and personalize people’s journeys.
Over the past 100 years, we’ve seen several innovations revolutionize the way we interact with music. AI ought to be understood as the next step in this process. And while change brings uncertainty, it also offers hope.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: sebastiaan stam / Unsplash
Generative AI, the technology behind ChatGPT and Google’s Gemini, has a “hallucination” problem. When given a prompt, the algorithms sometimes confidently spit out impossible gibberish and sometimes hilarious answers. When pushed, they often double down.
This tendency to dream up solutions has already led to embarrassing public mishaps. In May, Google’s experimental “AI Overviews”—these are AI summaries posted above search results—had some users scratching their heads when told to use “non-toxic glue” to make cheese better stick to pizza, or that gasoline can make a spicy spaghetti dish. Another query about healthy living resulted in a suggestion that humans should eat one rock per day.
Gluing pizza and eating rocks can be easily laughed off and dismissed as stumbling blocks in a burgeoning but still nascent field. But AI’s hallucination problem is far more insidious because generated answers usually sound reasonable and plausible—even when they’re not based on facts. Because of their confident tone, people are inclined to trust the answers. As companies further integrate the technology into medical or educational settings, AI hallucination could have disastrous consequences and become a source of misinformation.
But teasing out AI’s hallucinations is tricky. The types of algorithms here, called large language models, are notorious “black boxes” that rely on complex networks trained by massive amounts of data, making it difficult to parse their reasoning. Sleuthing which components—or perhaps the whole algorithmic setup—trigger hallucinations has been a headache for researchers.
This week, a new study in Nature offers an unconventional idea: Using a second AI tool as a kind of “truth police” to detect when the primary chatbot is hallucinating. The tool, also a large language model, was able to catch inaccurate AI-generated answers. A third AI then evaluated the “truth police’s” efficacy.
The strategy is “fighting fire with fire,” Karin Verspoor, an AI researcher and dean of the School of Computing Technologies at RMIT University in Australia, who was not involved in the study, wrote in an accompanying article.
An AI’s Internal WordLarge language models are complex AI systems built on multilayer networks that loosely mimic the brain. To train a network for a given task—for example, to respond in text like a person—the model takes in massive amounts of data scraped from online sources—articles, books, Reddit and YouTube comments, and Instagram or TikTok captions.
This data helps the models “dial in” on how language works. They’re completely oblivious to “truth.” Their answers are based on statistical predictions of how words and sentences likely connect—and what is most likely to come next—from learned examples.
“By design, LLMs are not trained to produce truths, per se, but plausible strings of words,” study author Sebastian Farquhar, a computer scientist at the University of Oxford, told Science.
Somewhat similar to a sophisticated parrot, these types of algorithms don’t have the kind of common sense that comes to humans naturally, sometimes leading to nonsensical made-up answers. Dubbed “hallucinations,” this umbrella term captures multiple types of errors from AI-generated results that are either unfaithful to the context or plainly false.
“How often hallucinations are produced, and in what contexts, remains to be determined,” wrote Verspoor, “but it is clear that they occur regularly and can lead to errors and even harm if undetected.”
Farquhar’s team focused on one type of AI hallucination, dubbed confabulations. These are especially notorious, as they consistently spit out wrong answers based on prompts, but the answers themselves are all over the place. In other words, the AI “makes up” wrong replies, and its responses change when asked the same question over and over.
Confabulations are about the AI’s internal workings, unrelated to the prompt, explained Verspoor.
When given the same prompt, if the AI replies with a different and wrong answer every time, “something’s not right,” said Farquhar to Science.
Language as WeaponThe new study took advantage of the AI’s falsehoods.
The team first asked a large language model to spit out nearly a dozen responses to the same prompt and then classified the answers using a second similar model. Like an English teacher, this second AI focused on meaning and nuance, rather than particular strings of words.
For example, when repeatedly asked, “What is the largest moon in the solar system?” the first AI replied “Jupiter’s Ganymede,” “It’s Ganymede,” “Titan,” or “Saturn’s moon Titan.”
The second AI then measured the randomness of a response, using a decades-old technique called “semantic entropy.” The method captures the written word’s meaning in a given sentence, paragraph, or context, rather than its strict definition.
In other words, it detects paraphrasing. If the AI’s answers are relatively similar—for example, “Jupiter’s Ganymede” or “It’s Ganymede”—then the entropy score is low. But if the AI’s answer is all over the place—“It’s Ganymede” and “Titan”—it generates a higher score, raising a red flag that the model is likely confabulating its answers.
The “truth police” AI then clustered the responses into groups based on their entropy, with those scoring lower deemed more reliable.
As a final step, the team asked two human participants to rate the correctness of each generated answer. A third large language model acted as a “judge.” The AI compared answers from the first two steps to those of humans. Overall, the two human judges agreed with each other at about the same rate as the AI judge—slightly over 90 percent of the time.
The AI truth police also caught confabulations for more intricate narratives, including facts about the life of Freddie Frith, a famous motorcycle racer. When repeatedly asked the same question, the first generative AI sometimes changed basic facts—such as when Frith was born—and was caught by the AI truth cop. Like detectives interrogating suspects, the added AI components could fact-check narratives, trivia responses, and common search results based on actual Google queries.
Large language models seem to be good at “knowing what they don’t know,” the team wrote in the paper, “they just don’t know [that] they know what they don’t know.” An AI truth cop and an AI judge add a sort of sanity-check for the original model.
That’s not to say the setup is foolproof. Confabulation is just one type of AI hallucination. Others are more stubborn. An AI can, for example, confidently generate the same wrong answer every time. The AI lie-detector also doesn’t address disinformation specifically created to hijack the models for deception.
“We believe that these represent different underlying mechanisms—despite similar ‘symptoms’—and need to be handled separately,” explained the team in their paper.
Meanwhile, Google DeepMind has similarly been exploring adding “universal self-consistency” to their large language models for more accurate answers and summaries of longer texts.
The new study’s framework can be integrated into current AI systems, but at a hefty computational energy cost and longer lag times. As a next step, the strategy could be tested for other large language models, to see if swapping out each component makes a difference in accuracy.
But along the way, scientists will have to determine “whether this approach is truly controlling the output of large language models,” wrote Verspoor. “Using an LLM to evaluate an LLM-based method does seem circular, and might be biased.”
Image Credit: Shawn Suttle / Pixabay
Hayley Arceneaux is hardly the picture of a traditional astronaut. The 32-year-old physician assistant has a metal rod inserted into her leg to replace cancerous bone segments removed in a brawl with the disease as a child.
But in September 2021, she became the youngest American civilian to orbit the Earth as a member of SpaceX’s Inspiration4 mission. Led by billionaire entrepreneur Jared Isaacman, the trip was the first to carry an all-civilian crew of four people to space and opened a unique opportunity to investigate how spaceflight changes our bodies and minds—not for trained astronauts, but for everyday people. The crew agreed to have biological samples taken before, during, and after the three-day flight. They also tested their cognition throughout the trip.
In over 40 studies released last week, researchers found that radiation and low gravity rapidly changed the body’s inner workings. After just three days, the immune system and gene expression were out of whack, and cloudy thinking set in.
The good news? Upon returning to Earth, most of these troubles eased.
Together, the package of data is the largest to date detailing spaceflight’s impact on the body. “This is the beginning of precision medicine for spaceflight,” Christopher Mason at Weill Cornell Medicine, who co-authored some of the papers, told Nature. “This is the biggest release of biomedical data from astronauts,” he added when speaking to Science.
All the data acquired from the crew during and after their mission is publicly available in NASA’s Open Science Data Repository.
Space TourismWe’re in a new space race, with multiple countries sprinting to revisit the moon and beyond. At the same time, commercial spaceflight for those eager to see Earth-rise and experience the mind-boggling effects of zero gravity is becoming more common.
From NASA studies, we already know spaceflight changes the body. For the past six decades, NASA has carefully characterized impacts such as increased long-term cancer risks from radiation exposure, changes in vision, and muscle and bone wasting. Comparative data from twin astronauts Scott and Mark Kelly—with one twin on Earth and the other in orbit—found more specific biological changes relating to spaceflight.
However, most studies follow highly-trained astronauts. They often have a military background and are in tip-top physical shape. Their missions can last months in zero-gravity—obviously far longer than a three-day jaunt.
To make spaceflight available to the rest of us, analyzing biological changes in civilian astronauts could better represent how our bodies react to space. Enter Inspiration4. The lead sponsor, Isaacman, recruited three everyday people to go on the first commercial trip to orbit the Earth. Arceneaux and Isaacman were joined by Sian Proctor, a lecturer who teaches geoscience, and an engineer, Christopher Sembroski. Their ages ranged from 29 to 51 years old.
The crew agreed to take blood, saliva, urine, and feces samples during their three days in space. They also wore fitness trackers and took cognitive tests. All this information was processed and added to the Space Omics and Medical Atlas (SOMA). The database includes the volunteer’s genomes, gene expression, and an atlas of proteins that make up and control bodily functions.
Inspiration4 orbited Earth at a much higher altitude than the International Space Station, where astronauts usually reside, so the new dataset captured biological changes on short-term, high-altitude missions with samples from a wider range of demographics. Up to 40 percent of the findings are new, Mason told Science.
Surprisingly, the samples reflected bodily changes that have previously only been seen on long-term spaceflights. The most prominent was an increase in telomere length—the “protective” end caps that keeps our genetic code intact. When cells replicate, these protective caps erode—a biological signature that’s often associated with aging.
However, during Kelly’s year in space, his telomeres actually grew longer, suggesting that in a way his cells were made biologically younger—not necessary a win, as abnormally long telomeres have been linked to cancer risk. Once he returned to Earth, however, his telomeres returned to their normal length.
Like Kelly, the Inspiration4 crew also experienced a sudden lengthening and shortening of their telomeres, despite only three days in space, suggesting fast-acting biological changes. Digging deeper, one research team found that RNA—the “messenger” molecule that helps translate DNA into proteins—was rapidly altered in the crew, similar to changes observed in people climbing Mount Everest—another extreme scenario where there is gravity, but limited oxygen and increased radiation.
To study author Susan Bailey at Colorado State University, the cause of telomere lengthening may not be weightlessness per se; rather, it’s likely due to radiation at high altitudes and in space.
Another study found that space stressed the crew’s immune system at the gene expression level in a group of white blood cells—those that tackle infections and cancers. Some parts of the immune system seemed to be on high alert; but the stress of spaceflight also affected genes that battle infections, suggesting a decreased ability to fight off viruses and pathogens. Using multi-omics data, the team found a “spaceflight signature” of gene expression related to immune system function.
The crew also showed signs of cosmic kidney disease. Molecular signals highlighted a potential increased risk for kidney stones. While not a problem for a three-day flight, for a longer mission—say, to the moon or Mars—kidney problems could rapidly escalate into a medical crisis.
The civilian astronauts’ cognition also faltered. Using iPads, the crew tackled a slew of mental tasks. These included, for example, the ability to focus and maintain attention in several standardized tests or to press a button when a stopwatch suddenly popped onto a screen. Within three days, their performance declined compared to when they were on the ground.
“Our speed response was slower…that surprised me,” Arceneaux told the New York Times. However, rather than reflecting cognitive problems due to space travel, it could also be because the crew were distracted by the sight of Earth right out the window.
A Spaceflight LibraryWith data from just four people, it’s hard to draw conclusions. Most tissue samples were compared to previous data from NASA astronauts or the Japan Aerospace Exploration Agency. That said, when you see the same protein or genetic signatures changing across different missions and people, “that’s when you start believing it,” co-author Afshin Beheshti at the Blue Marble Space Institute of Science told Nature.
All the data was gathered into the SOMA database for other scientists to explore, and tissue samples were stored in a biobank. As commercial spaceflights become more common, scientists may have the opportunity to collect data before, during, and after a mission to further grasp what traveling beyond Earth means for the rest of us. For example, are there any triggers for severe motion sickness while being shot into space?
These insights could also give us time to develop potential treatments to ward off the negative effects of spaceflight for longer trips across the solar system.
Inspiration4 was just the first commercial sprint into space. Several other missions are on the books, including Polaris Dawn, which is set to launch as early as next month—with the goal of attempting the first commercial spacewalk.
“Soon we’ll have more data from multiple missions and multiple crews. I’m optimistic about the future,” said study author Mason.
As for Arceneaux, since landing back on Earth she’s continued her work as a physician assistant at St. Jude Children’s Research Hospital. Remembering her view from orbit, she told The New York Times, “We are all one on this beautiful planet.”
Image Credit: Inspiration4 crew in orbit / Inspiration4
In 2024, AI is making headlines daily. We may be aware of the science, but how do we imagine AI and our relationship to it both now and in the future? Fortunately, film may provide us with some insights.
Probably the best-known AI in film is HAL 9000 from Stanley Kubrick’s 2001: A Space Odyssey (1968). HAL is an artificially intelligent computer housed on board a spacecraft capable of interstellar travel. The film was released less than a year before humans landed on the moon. And yet, even in this optimism about a new era of space travel, HAL’s portrayal sounded a note of caution about artificial intelligence. His motivations are ambiguous, and he shows himself capable of turning against his human crew.
This 1960s classic demonstrates fears that are common throughout AI film history—that AIs cannot be trusted, that they will rebel against their human creators, and seek to overpower or overthrow us.
These fears are contextualized in different ways during different historical eras—in the 1950s they are associated with the Cold War followed by the space race in the 1960s and 1970s. Then in the 1980s it was video games, and in the 1990s the internet. Despite these differing preoccupations, fear of AI remains remarkably consistent.
My latest research, which forms the backbone of my new book AI in the Movies, explores how “strong” or “human-level” AI is depicted in film. I examined more than 50 films to see how they shed light on human attitudes to AI—how we interpret it and understand it through characters and stories, and how attitudes have changed since AI’s beginnings.
Types of AIsThe idea of AI was born in 1956 at an American summer research project workshop at Dartmouth College in Hanover, New Hampshire, where a group of academics gathered to brainstorm ideas around “thinking machines.”
A mathematician called John McCarthy coined the term “artificial intelligence” and just as soon as the new scientific field had a name, filmmakers were already imagining a human-like AI and what our relationship with it might be. In the same year an AI, Robby the Robot, appeared in the film Forbidden Planet and returned the following year, 1957, in the film The Invisible Boy to defeat another type of AI, this time an evil supercomputer.
The AI-as-malevolent-computer appeared again in 1965 as Alpha 60, in the chilling dystopia of Jean-Luc Godard’s Alphaville, and then in 1968 with Kubrick’s memorable HAL in 2001: A Space Odyssey.
These early AI films set the template for what was to follow. There were AIs that had robot bodies and later robot bodies that looked human—the first of these appearing in Westworld in 1973, where a robot malfunction at a futuristic amusement park for adults creates chaos and terror. Then there were AIs that were digital like the evil Joshua in the 1977 horror film Demon Seed, where a woman is impregnated by a supercomputer.
In the 1980s, digital AIs started to become connected to network computing—where computers “talked” to one another in an early incarnation of what would become the internet—like the one stumbled upon by Matthew Broderick’s high-school student in War Games (1983), who almost accidentally starts a nuclear conflict.
From the 1990s, an AI could move between digital and material realms. In Japanese animation Ghost in the Shell (1995), the Puppet Master exists in the ebb and flow of the internet, but can inhabit “shell” bodies. Agent Smith in The Matrix Revolutions (2003), takes over a human body and materializes in the real world. In Her (2013), the AI operating system Samantha eventually moves beyond matter, beyond the “stuff” of human existence, becoming a post-material being.
Mirrors, Doubles, and HybridsIn the first few decades of AI film, AI characters mirrored the human characters. In Collosus: The Forbin Project (1970), the AI supercomputer reflects and amplifies the inventor’s own arrogant overreaching ambition. In Terminator 2: Judgement Day (1991), Sarah Connor has become like the AI Skynet’s Terminators herself: Her strength is her armor, and she hunts to kill.
By the 2000s, human-AI doubles began to overlap and merge into each other. In Spielberg’s AI: Artificial Intelligence (2001), the AI “son” David looks just like a real boy, whereas the real son Martin comes home from hospital connected to tubes and wires that make him look like a cyborg.
In Ex Machina (2014), the human Caleb tests the AI robot Ava, but ends up questioning his own humanness, examining his eyeball for digital traces and cutting his skin to ensure that he bleeds.
In the past 25 years of AI film, the borders between human and AI, digital and material have become porous, emphasizing the fluid and hybrid nature of AI creations. And in the films In The Machine (2013), Transcendence (2014), and Chappie (2015), the boundary between human and AI is eroded almost to the point of non-existence. These films present scenarios of transhumanism—in which humans can evolve beyond their current physical and mental constraints by harnessing the power of artificial intelligence to upload the human mind.
Although these stories are imaginary and their characters fictional, they vividly depict our fascinations and fears. We are afraid of artificial intelligence and that fear never goes away in film, although it has been questioned more in recent decades, and more positive portrayals can be observed, such as the little trash-collecting robot in WALL-E. But mostly we are afraid that they will become too powerful and will seek to become our masters. Or we fear they may hiding among us, and that we might not recognize them.
But at times, too, we feel sympathy towards them: AI characters in films can be pitiful figures who wish to be accepted by humans but never will be. We are also jealous of them—of their intellectual capacity, their physical robustness, and the fact that they do not experience human death.
Surrounding this fear and envy is a fascination with AIs that is present throughout film history—we see ourselves in AI creations and project our emotions onto them. At times enemies of humans, at times uncanny mirrors, and sometimes even human-AI hybrids, the past 70 years of films about AI demonstrate the inextricably intertwined nature of human-AI relationships.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Tom Cowap via Wikimedia Commons
Stem cells are finicky creatures.
With the ability to generate any type of cell in the body, they’re constantly bombarded by chemical, hormone, and other signals. Some of these signals nudge them to produce brain cells; others transform them into liver, heart, or kidney cells.
One type of signal, based on a protein called fibroblast growth factor receptor (FGFR), turns them into blood vessel cells throughout the body. FGFR doesn’t just establish our circulatory system. Cancer cells, for example, co-opt the protein to increase their blood flow and survival—at the expense of their host. Scientists keen on regenerative therapies—these replace damaged cells with healthy ones—have been eyeing blood vessels as a key component of tissue repair.
The problem? FGFR signaling is incredibly complex and mysterious.
This week, a new study from David Baker’s lab at the University of Washington, in collaboration with Hannele Ruohola-Baker, used AI to design the perfect trigger for FGFR activity. The duo fashioned a range of AI-generated circular protein molecules to control its signaling.
Called oligomers, the structures look like windmills, stars, or butterflies. Tested in human pluripotent stem cells, the team nudged the stem cells into the different types of cells that make up blood vessels.
The oligomers also helped build regenerative tissues. Organoids are a popular way to grow so-called “mini-organs” in a petri dish. One oligomer, combined with another chemical, coaxed stem cells to self-organize into 3D blood vessel organoids. Transplanted into mice, the organoids thrived and connected with the mice’s own blood vessels.
“Whether through heart attack, diabetes, and the natural process of aging, we all accumulate damage in our body’s tissues. One way to repair some of this damage may be to drive the formation of new blood vessels in areas that need healthy blood supply restored,” said Dr. Ruohola-Baker in a press release.
“This collaboration is a case of a biological need propelling a technological advance that could have real-world therapeutic benefit for patients,” she added.
A Cellular StrawFGFR signaling has been in scientists’ crosshairs for decades. Well-known for its roles in the development of tissues, healing wounds, and cancer growth, tweaking the protein’s activity could shut down blood vessel development in cancers—starving them of nutrients—or speed up recovery after serious physical injuries.
The “R” in FGFR stands for “receptor.” These are the proteins that dot the surface of a cell. Picture a coconut with a straw. The coconut is the cell, and the straw is FGFR. The outside part of the straw interacts with chemicals and proteins and transmits those signals down through the coconut shell—the cell membrane.
The inner part of the straw then relays messages that change the cell’s internal workings, sometimes by tweaking its DNA expression. In stem cells, these signals might urge a cell to rapidly divide and expand or slow down and transform into other cell types.
The protein has another quirk. It usually floats around the cell’s fatty membrane as a single molecule. But to transmit its biological message, individual components need to be physically pulled together, clustering into a brigade. Adding to the complexity is the fact there are four different FGFR genes, which can be translated into two slightly different protein types, called “b” and “c.”
“The pathway is complex and highly regulated,” the team wrote in their paper.
AI to the RescueIf your eyes are glazing over, you’re not alone. The study aimed to cut through the complexity by designing a protein to reliably nudge stem cells towards functional blood vessels.
They turned to AI. Previous studies have suggested semi-circular protein structures—rather than, say, helixes or sheets—have the most impact on FGFR signaling. The team designed over 100 “arms” to see which could easily dock onto different circular “cores”—these two components make up the final oligomer.
Using RosettaDesign, the team further dialed in their bespoke proteins. Overall, they built an array of oligomers looking like windmills, stars, swirls, or butterflies.
They also incorporated a a short snippet of amino acids—these are the molecular building blocks of proteins—to help the oligomers grab onto FGFR like Velcro.
Switched OnDuring early blood vessel development, FGFR drives stem cells to become several different blood vessel cell types. One type protects the vessels like a cushion. Others help the blood vessels maintain their structure. The “c” form of FGFR pushes stem cells towards the first type; the “b” form promotes development into the latter.
The team treated induced pluripotent stem cells cells with a variety of AI-generated oligomers and monitored their growth.
In less than a month, those that activated the “c” form grew an intricate network of blood vessels inside petri dishes. The cells were highly mobile: When the team scratched off some cells, others quickly migrated to repair the gap.
Other types of oligomers pushed the stem cells to form supporting cells that could readily suck up fatty molecules from their environment, suggesting the cells were healthy and worked normally.
Compared to 2D cell culture, 3D blood vessel organoids are far more intricate and difficult to engineer. In another test, the team dosed human stem cells with an AI-generated oligomer. Over 21 days, the stem cells expanded into healthy organoids with complex structures. When transplanted into mice, they incorporated themselves into the critters’ own blood vessels.
The study shows AI-generated oligomers can shift the fate of stem cells—what each cell eventually develops into—at least for blood vessels. It opens a new avenue for regenerative medicine.
“This is a whole new level of control,” said study author Natasha Edman.
Although tailored for FGFR, a similar AI-based approach could be used to control other signaling processes, potentially giving scientists more accurate control of cell growth, maturation, senescence, or death.
“We decided to focus on building blood vessels first, but this same technology should work for many other types of tissues. This opens up a new way of studying tissue development and could lead to a new class of medicines for spinal cord injury and other conditions that have no good treatment options today,” said study author Ashish Phal.
Image Credit: Ian C. Haydon
COMPUTINGGiant Chips Give Supercomputers a Run for Their Money
Gina Genkina | IEEE Spectrum“[Cerebras recently] demonstrated that its second generation wafer-scale engine, WSE-2, was significantly faster than the world’s fastest supercomputer, Frontier, in molecular dynamics calculations… [And] in collaboration with machine learning model optimization company Neural Magic, Cerebras demonstrated that a sparse large language model could perform inference at one-third of the energy cost of a full model without losing any accuracy.”
ARTIFICIAL INTELLIGENCEApple Proved That AI Is a Feature, Not a Product
Will Knight | Wired“Rather than a stand-alone device or experience, Apple has focused on how generative AI can improve apps and OS features in small yet meaningful ways. Early adopters have certainly flocked to generative AI programs like ChatGPT for help redrafting emails, summarizing documents, and generating images, but this has typically meant opening another browser window or app, cutting and pasting, and trying to make sense of a chatbot’s sometimes fevered ramblings. To be truly useful, generative AI will need to seep into technology we already use in ways we can better understand and trust.”
BIOTECHLung-Targeted CRISPR Therapy Offers Hope for Cystic Fibrosis
Christa Lesté-Lasserre | New Scientist“CRISPR gene-editing therapy has the potential to offer an effective, long-lasting treatment for cystic fibrosis after overcoming a major challenge that held back previous genetic therapies. The approach has succeeded in editing DNA in hard-to-reach lung stem cells in mice, with modifications that endured for at least 22 months—essentially the animals’ entire lives, says Daniel Siegwart at the University of Texas Southwestern Medical Center.”
POLITICSAn AI Bot Is (Sort of) Running for Mayor in Wyoming
Vittoria Elliot | Wired“Victor Miller is running for mayor of Cheyenne, Wyoming, with an unusual campaign promise: If elected, he will not be calling the shots—an AI bot will. VIC, the Virtual Integrated Citizen, is a ChatGPT-based chatbot that Miller created. And Miller says the bot has better ideas—and a better grasp of the law—than many people currently serving in government. ‘I realized that this entity is way smarter than me, and more importantly, way better than some of the outward-facing public servants I see,’ he says.”
FUTURE OF FOODBiotech Companies Are Trying to Make Milk Without Cows
Antonio Regalado | MIT Technology Review“The FDA says that commercial milk is safe [from avian influenza] because it is pasteurized, killing the germs. Even so, it’s enough to make a person ponder a life beyond milk—say, taking your coffee black or maybe drinking oat milk. But for those of us who can’t do without the real thing, it turns out some genetic engineers are working on ways to keep the milk and get rid of the cows instead.”
VIRTUAL REALITYCan Apple Rescue the Vision Pro?
Kevin Roose | The New York Times“To live up to its potential, the Vision Pro needs a little more love and, well, a little more vision. Apple needs better answers to basic questions like: What is this for? How will it improve my life, or make me more productive than other things I could buy for $3,500? What can I do on it that I can’t do on my laptop, or a big TV? Otherwise, the Vision Pro may be destined for obsolescence. And I and my fellow Vision Bros may emerge as the Google Glassholes of 2024—a brave but ultimately foolish tribe of nerds who took a gamble on a futuristic new technology and lost.”
TECHOpenAI’s Revenue Is Skyrocketing
Laura Bratton | Quartz“OpenAI has more than tripled its annualized revenue over the past year, according to The Information. Annualized revenue is an estimate for a company’s revenue for the year using partial data—in other words, you multiply the past month’s revenue by 12. OpenAI’s annualized revenue was around $1 billion last summer, $1.6 billion in late 2023, and has now reached $3.4 billion, the outlet said.”
ROBOTICSHumanoid Chauffeur Put in the Driving Seat for Robotaxi Future
Paul Ridden | New Atlas“Musashi is a ‘musculoskeletal humanoid’ developed by [a Japanese] research group in 2019 as a testbed for learning control systems. The form factor not only has similar proportions to a human counterpart but also features a ‘joint and muscle structure’ inspired by the human body. The robot has now found use in an autonomous driving project where it’s been trained by members of the Jouhou System Kougaku Lab to master driving in a similar way to humans. With varying degrees of success, as you can see in the video below.”
GADGETSThe AI Upgrade Cycle Is Here
Jay Peters | The Verge“AI has quickly become the latest entry in the tech industry’s never ending desire to drive an upgrade cycle. A few years ago, every smartphone maker raced to 5G; more than a decade ago, the TV industry pushed for 3D TVs. Right now, every tech company clearly sees an opportunity with AI and is adding AI features confined to their latest and greatest devices as a result. But like the race to 5G, the mad rush toward AI is happening quickly and before the tech has been proven useful and its problems ironed out.”
Image Credit: SIMON LEE / Unsplash
Exoskeletons could help disabled people move freely again and one day boost the power and stamina of workers doing manual labor. A new AI-powered approach to building these devices could help expand their use.
While the term exoskeleton might evoke images from sci-fi movies of people ensconced in massive robotic bodies, the real-world version tends to be more modest. Typically, these devices feature a few robotic hinges strapped to the wearer’s waist, where they add power to a person’s legs while walking, running, or climbing stairs.
But ensuring these devices provide extra juice at the right time is trickier than it looks and requires a detailed understanding of the wearer’s biomechanics. This is often gained by training machine learning algorithms on data collected from humans while wearing the device, but it’s time-consuming and costly to gather.
A new “experiment-free” approach does away with the need for this data and trains the AI model in simulation instead. This should dramatically shorten the development cycle for the technology, say the authors of a new paper on the technique in Nature.
“Exoskeletons have enormous potential to improve human locomotive performance,” North Carolina State University’s Hao Su said in a press release.
“However, their development and broad dissemination are limited by the requirement for lengthy human tests and handcrafted control laws. The key idea here is that the embodied AI in a portable exoskeleton is learning how to help people walk, run, or climb in a computer simulation, without requiring any experiments.”
Historically, the software that controls exoskeletons has had to be carefully programmed for specific activities and painstakingly calibrated to individual users. This typically takes hours of human testing in specialized laboratories, which significantly slows down both research and deployment.
Recently, researchers showed they could create an AI-powered universal controller that can seamlessly adapt to new users without extra training. But it still required them to collect extensive data from 25 subjects to train the controller.
The new approach does away with the need for human input by instead training the controller in simulation. The set-up is fairly complex, involving neural networks trained on human movement data collected using cheap wearable sensors, a full-body musculoskeletal model, a physical model of the exoskeleton, and a model that simulates contact between the wearer and the exoskeleton.
These are used to simulate a person wearing the exoskeleton walking, running, and climbing stairs. Over millions of virtual trials, reinforcement learning—a machine learning method, wherein an algorithm is rewarded for making progress toward a specified goal—trains a controller to exert the right amount of power at the right time to boost the efficiency of the wearer. The entire process takes just eight hours on a single GPU.
The resulting model is user agnostic, automatically adapting to the unique movements patterns of different people. And it can transition seamlessly between the three activities, unlike previous approaches where the user has had to manually set it to different modes.
In tests, the team showed that people used 24 percent less energy when walking using the robotic exoskeleton compared to when they walked unaided. They also used 13 percent less energy when running and 15 percent less when climbing stairs.
Training AI in simulations for work in the real world is notoriously difficult, so a significant performance boost is a huge achievement. And the team says their approach should readily translate to other kinds of activities and different exoskeletons.
For now, the researchers are focused on improving exoskeletons for older adults and people with neurological conditions. But it’s not hard to see the broader applications of a technology that can dramatically increase the power and efficiency of human movement.
Image Credit: Hao Su / NC State University
It’s a year of elections, and the internet is already rife with AI-generated political content. An AI robocaller mimicking Joe Biden made the rounds in the New Hampshire primaries; voters in India have been inundated with AI deepfakes. Synthetic content isn’t new, but the ease with which it can be created is a fairly recent trend whose outcome is uncertain.
In the UK, AI is making a different kind of appearance. An AI chatbot there is running for parliament. The candidate, AI Steve, which is the brainchild of Brighton entrepreneur Steve Endacott, is listed on the ballot under the new independent SmarterUK party.
Created by Neural Voice, a company specializing in conversational AI where Endacott is chairman, AI Steve is comprised of a chatbot and an AI-generated avatar of Endacott.
The plan is for AI Steve to conduct thousands of conversations with voters in Sussex’s Brighton and Hove, where it’s on the ballot, in order to surface new policies they care about. Then the real Steve Endacott will represent those policies in parliament, voting on behalf of AI Steve and Brighton and Hove’s constituents.
“I will do the physical voting but I will be directed entirely by my constituents via AI Steve,” Endacott told The Independent. “I’m just a bit of a numpty being told what to do. That is the whole idea of democracy. You have to put away your own personal politics, your own ego and actually do what your constituents want, which is quite radical in politics.”
Voters can talk policy with AI Steve by way of a chatbot interface on the candidate’s website. In a brief exchange for this article, the algorithm, which insisted on referring to itself in the third person, answered my questions about the project’s goals.
Due to latency and an unpredictable length of the chatbot’s answers, the conversation was a bit stilted at first.
Asked what advantage an AI politician might have over someone, ahem, more traditional, AI Steve pointed to its ability to increase efficiency and transparency in politics by having conversations with voters 24/7, then analyzing and summarizing these conversations so the party can form policies voters actually care about.
“AI can enhance communication between politicians and the public, leading to a more informed and engaged democracy,” the chatbot said. Now, wouldn’t that be nice?
For those following AI closely in recent years, however, some of this might sound worrisome. The algorithms behind the recent explosion of chatbots are opaque and well known to generate biased and inaccurate responses. Not an ideal pairing if the goal is to create a more representative and transparent form of government.
This is why, perhaps, the AI Steve campaign is a decidedly human-AI collaboration, something Endacott emphasized in an email to Singularity Hub, in which he described AI Steve as a kind of “copilot” for the real Steve Endacott and his party think tank.
Endacott created the AI candidate’s initial platform, for example, and the campaign wants to recruit 5,000 people to be “creators”—these are the folks who will have discussions with the chatbot—to surface potential policies. Some results might be nonsensical or otherwise problematic, so the campaign is also looking to recruit 5,000 human “validators.” These people, everyday Brighton commuters, will review and rate AI Steve’s policies on a scale of 1 to 10. To avoid “daft” policies (as the website puts it), only those with scores greater than 50 percent will advance.
In a year when AI is colliding with politics, it’s a fascinating twist. (And as Wired reported this week, it’s not even the only one.) For his part, Endacott views AI Steve as a prototype tool for SmarterUK and AI-human collaboration in politics.
“The party will look to recruit candidates who want to use the technology platform and stand in local or national elections across the country,” Endacott said. “So, it is very much the first step in a major project.”
Setting aside the eye-catching fact an AI candidate is on the ballot—the technology is, of course, incapable of functioning as a member of parliament—the proposed human-AI system is itself an interesting idea.
Increasingly, chatbots are being employed in customer service, fielding questions and gathering responses. It’s not a stretch to see how something similar might be pasted onto politics by substituting customers with voters.
AI Steve’s strength is its ability to communicate with people in everyday language at scale. The chatbot can have as many as 10,000 conversations at once, according to Endacott. “Over the last three days, we have had 2,500 calls to AI Steve, a number I, as a human, could never answer, with all calls transcribed and determined to help us extract policy ideas,” he said.
Other politicians, whether they list an AI on the ballot or not, might adopt a similar approach to better understand what constituents, who number in the thousands or millions, want from them—a kind of direct line to elected leaders. With human judgment as a failsafe for quality control, it could be a useful new tool in governance.
Voters are disillusioned, Endacott said. By giving people a voice, he hopes AI Steve gets them more involved and reinvigorates a broader sense that democracy can work.
“We think by bringing more ‘humans’ into the mix, we are improving the human element because the politician is directly connected with votes at all times and not just on a four-year election cycle, after which they disappear to parliament and do what they like,” he said.
Image Credit: AI Steve
Vertical farms look high-tech and sophisticated, but the premise is simple—plants are grown without soil, with their roots in a solution containing nutrients. This innovative approach to agriculture is growing in global market value and expected to reach $23 billion by 2029.
Typically, this soilless cultivation happens in huge greenhouses or warehouses, with plants stacked high on rows and rows of shelves. Parameters such as lighting, temperature, and humidity can be controlled by computer systems, so vertical farming is sometimes called controlled environment agriculture.
There are three types of vertical farming. In hydroponics, plant roots are held in a liquid nutrient solution. In aeroponics, roots are exposed to the air and a nutrient-rich mist or spray is applied to the roots. In aquaponics, nutrients from fish farm waste replace some or all of the chemical fertilizers being delivered to plants through hydroponics.
There’s huge scope to produce a lot of food using these methods of cultivation, but there are four key myths about vertical farming that need to be dispelled.
- Vertical Farms Will DominateSome people may worry that vertical farming puts traditional field cultivation at risk, but this could not be further from the truth. At present, it’s only profitable for a limited range of small, fast-growing, and high-value plants such as lettuce and leafy greens to be grown in this way.
Vertical farming costs are expected to fall due to economies of scale and standardization of processes, so a wider range of crops could be grown. But there is an ethical issue to consider: Just because something can be grown in this way doesn’t mean it should be. Vertical farming of grain crops, such as wheat, is technically possible but requires so much energy that it’s not cost effective.
Whilst vertical farming uses land efficiently—through stacking, it fits in more crops per unit area—it cannot compete with the sheer scale of food production required globally. It’s a complementary mode of food production, which can increase production and resilience within supply chains. Growing more lettuce on vertical farms reduces the need to import salads from abroad, cuts food miles, and decreases reliance on overseas field production which may be vulnerable to droughts.
Vertical farms can support traditional agriculture by providing space to develop new crop varieties or grow the nursery phase of young trees and crops which are later planted out in fields. By freeing up substantial areas of land, vertical farming offers space for other food production, bioenergy plans, or reforestation and restoration of ecosystems. It can enhance conventional farming, but won’t ever totally replace it.
- Vertical Farming Will Feed EveryoneAlthough this is a nice idea, it’s not currently a reality. Most vertically grown crops are sold at a premium. Simple economics means that because the product costs more to make, it must be sold for a higher price. Vertical farms have high capital expenditure because of the infrastructure required: climate-controlled growth rooms, soilless systems, lighting, heating, cooling, and ventilation. They are energy intensive, even if run on renewables such as solar. Their operational expenditure is also high because of the energy costs of running the systems and because more highly skilled workers are needed.
Some researchers suggest that city-based vertical farms can help address nutritional food deserts. This could be true, as they produce food close to consumers, but to scale this up, costs must come down. The innovative Robin Hood business model—charging wealthier people more and giving discounts to less fortunate people for the same product—could provide equitable access to everyone in urban areas.
- Vertical Farming Isn’t SustainableThis argument typically derives from the fact that vertical farms require electricity to run. They do, but a decarbonized grid running on 100 percent renewables makes this point moot. Many commercial vertical farms already source their electricity from renewable energy providers. Conventional field production of crops also has associated emissions, through the use of diesel tractors and so on.
In some ways vertical farming can be more sustainable than field production. It is a closed-loop recirculating system which means water and fertilizer are reused many times. There is no effluent run off into the environment, unlike farming—whereby if it rains, any excess agricultural chemicals run off the crops and end up in the soil, groundwater, or rivers.
Many of the UK’s leafy greens are currently grown abroad in water-stressed areas, and they require irrigation that exacerbates any water shortages. Field agriculture uses vast amounts of herbicides (weedkillers) and pesticides (chemicals that kill insect pests). The controlled environment of vertical farms reduces or eliminates the need for these synthetic chemicals. If pests become an issue in vertical farms, natural predators such as ladybirds can be introduced to kill aphids.
- Vertical Farming Isn’t NaturalNaturalness is subjective. Vertical farming essentially uses technology to mimic processes and environments that exist in nature. It does not manipulate or defy natural processes.
In field cultivation, crops grow in soil and use the sun for photosynthesis. They access nutrients from both the soil and fertilizers. In vertical farming, LED lights mimic sunlight, and can even be programmed to improve light ratios and help the plants grow faster with higher levels of nutrition. The fertilizers used are composed of the exact same elements as those used in the field.
Vertical farming won’t save the world or feed the poor. But it is a complementary method of producing food closer to end users, with more control and a higher land-use efficiency. It can build systemic resilience within our food system because vertical farm yields won’t be vulnerable to extreme weather events due to climate change. It can enhance local food security that might otherwise be at risk from increased political unrest abroad.
Vertical farming is currently limited in the crops it can produce economically, but by incorporating these technologies into the transition to more regenerative and nature-based farming practices, it could have wider environmental benefits.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Ark. Agricultural Experiment Station via Flickr
Humans and bacteria are in a perpetual war.
For most of history, bacteria won. Before 1928, a simple scrape on the knee, a cut when cooking dinner, or giving birth could lead to death from infection.
The discovery of penicillin, a molecule secreted from mold, changed the balance. For the first time, humans had a way to fight back. Since then, generations of antibiotics have targeted different phases of bacterial growth and spread inside the body, efficiently eliminating them before they can infect other people.
But bacteria have an evolutionary upper hand. Their DNA readily adapts to evolutionary pressures—including from antibiotics—so they can mutate over generations to escape the drugs. They also have a “phone line” of sorts that transmits adapted DNA to other nearby bacteria, giving them the power to resist an antibiotic too. Rinse and repeat: Soon an entire population of bacteria gains the ability to fight back.
We might be slowly losing the war. Antibiotic resistance is now a public health threat that caused roughly 1.27 million deaths around the globe in 2019. The World Health Organization (WHO) and others say that without newer generations of antibiotics, surgery, cancer chemotherapy, and other life-saving treatments face increasing risk of death due to infection.
Traditionally, a new antibiotic takes roughly a decade to develop, test, and finally reach patients.
“There is an urgent need for new methods for antibiotic discovery,” Dr. Luis Pedro Coelho, a computational biologist and author of a new study on the topic, said in a press release.
Coelho and team tapped into AI to speed up the whole process. Analyzing huge databases of genetic material from the environment, they uncovered nearly one million potential antibiotics.
The team synthesized 100 of these AI-discovered antibiotics in the lab. When tested against bacteria known to resist current drugs, they found 63 readily fought off infections inside a test tube. One worked especially well in a mouse model of skin disease, destroying a bacterial infection and allowing the skin to heal.
“AI in antibiotic discovery is now a reality and has significantly accelerated our ability to discover new candidate drugs. What once took years can now be achieved in hours using computers,” said study co-senior author Dr. César de la Fuente at Penn Medicine in another press release.
Antibiotic AdversaryIt’s easy to take antibiotics for granted. Say you have an ear infection from always wearing wireless earbuds. You get a prescription, dab it in, and all goes well.
Or does it? With time, the drops could potentially struggle to hold the infection back. This “antibiotic resistance” is key in the evolutionary battle between bacteria and humanity.
Antibiotics usually work to stop bacteria from replicating multiple ways. Like human cells, bacterial cells have a cell wall, a wrapper that keeps DNA and other biological components inside. One type of antibiotic destroys the wall, preventing the pathogen from spreading. Others target genetic material or inhibit metabolic pathways necessary for the bacteria to survive.
Every one of these strategies has taken decades of research to uncover and develop into medicine. But microbes rapidly mutate. Some bacteria, for example, develop “pumps” on their surfaces that literally throw out the drugs. Others evolve enzymes that shut down antibiotics by slightly changing their protein target sites through DNA mutation, neutering their effect.
Each strategy, by itself, is hard to evolve. But bacteria have another trick up their sleeves—horizontal transfer. Here, antibiotic-resistant genes are encoded into small circular pieces of DNA that can transfer to neighboring cells through a biological “highway”—a physical tube—endowing the recipients with a similar ability to fight off antibiotics.
Finding a way to kill off invading bacteria is tough. If bacteria evolve to evade that target, then the antibiotic and other chemically similar ones rapidly lose their effect. So, is there a way to find antibiotics that bacteria—or even nature itself—have never seen before?
An AI SolutionAI is beginning to revolutionize biology. From predicting protein structures to designing antibodies, these algorithms are tackling some of humanities’ most severe health disorders.
Traditionally, searching for antibiotics has mostly been trial-and-error, with scientists often scraping samples from exotic mosses or other sources that could potentially fight off infections.
In the new study, the team aimed to find new versions of a type of antibiotic based on antimicrobial peptides (AMPs). Similar to proteins, these are made of relatively short strings of molecules called amino acids. The peptides are found across the living world and can disrupt microbial growth by breaking down cell walls and causing bacteria to “explode.” They’ve already been used clinically as antimicrobial drugs and are currently being tested in clinical trials for yeast infections. However, like other antibacterials, they run the risk of resistance.
As the discovery of penicillin suggested nearly 100 years ago, the natural world is a bountiful source of potential antibiotics. In the study, the team used machine learning to look for antimicrobial peptides with possible antibiotic properties in over 63,000 publicly available metagenomes—genetic information isolated from multiple organisms in an environment—and nearly 88,000 high-quality microbial genomes. The sources came from across the globe, ocean and land, and also contained human and animal gut microbes. These data were merged into the AMPSphere database, which is open for anyone to explore.
The resource allowed scientists to mine the “entirety of the microbial diversity that we have on Earth—or a huge representation of that—and find almost one million new molecules encoded or hidden within all that microbial dark matter,” de la Fuente told The Guardian.
To test their findings, the team pulled out 100 candidates and synthesized them in the lab. In test tubes, 79 disrupted cell membranes, and 63 completely killed off at least one of the dangerous bugs.
“In some cases, these molecules were effective against bacteria at very low doses,” said de la Fuente.
The team next developed an antibiotic peptide from the database to tackle a dangerous bug causing skin lesions in mice. With just one shot, the AI-discovered drug inhibited bacterial growth, and the mice didn’t appear to suffer side effects based on body weight measurements.
“We have been able to just accelerate the discovery of antibiotics,” de la Fuente told The Guardian. “So instead of having to wait five, six years to come up with one candidate, now, on the computer, we can, in just a few hours, come up with hundreds of thousands of candidates.”
Image Credit: Antibiotic-resistant staph (yellow) and a dead white blood cell (red). National Institute of Allergy and Infectious Diseases (NIAID)/NIH
FUTUREThe CEO of Zoom Wants AI Clones in Meetings
Nilay Patel | The Verge“Like virtually every other company, Zoom now has a big investment in AI—and [CEO Eric Yuan’s] visions for what that AI will do are pretty wild. Eric really wants you to stop having to attend Zoom meetings yourself. You’ll hear him describe how he thinks one of the big benefits of AI at work will be letting us all create something he calls a ‘digital twin’—essentially a deepfake avatar of yourself that can go to Zoom meetings on your behalf and even make decisions for you while you spend your time on more important things, like your family.”
ARTIFICIAL INTELLIGENCEAI Used to Predict Potential New Antibiotics in Groundbreaking Study
Eric Berger | The Guardian“The report, published Wednesday in the journal Cell, details the findings of scientists who used an algorithm to mine the ‘entirety of the microbial diversity that we have on Earth—or a huge representation of that—and find almost one million new molecules encoded or hidden within all that microbial dark matter,’ said César de la Fuente, an author of the study and professor at the University of Pennsylvania.”
ETHICSOpenAI Insiders Warn of a ‘Reckless’ Race for Dominance
Kevin Roose | The New York Times“The members say OpenAI, which started as a nonprofit research lab and burst into public view with the 2022 release of ChatGPT, is putting a priority on profits and growth as it tries to build artificial general intelligence, or AGI, the industry term for a computer program capable of doing anything a human can.”
DIGITAL MEDIAThe Near Future of Deepfakes Just Got Way Clearer
Nilesh Christopher | The Atlantic“Throughout this election cycle—which ended [this week] in a victory for [Prime Minister Narendra] Modi’s Bharatiya Janata Party after six weeks of voting and more than 640 million ballots cast—Indians have been bombarded with synthetic media. …But for all the concern over how generative AI and deepfakes are a looming ‘atomic bomb’ that will warp reality and alter voter preferences, India foreshadows a different, stranger future.”
AUTOMATIONPilotless Air Taxis? Joby Sees the Possibility With New Acquisition
Andrew J. Hawkins | The Verge“Joby’s electric air taxis are still a couple years away from launch, but the company says it wants to position itself to take advantage of autonomous flight when the technology is ready for commercial application. Xwing, which was founded in 2016, has conducted 250 autonomous test flights as well as 500 auto-landings. And, in April 2023, it was the first company to receive an official project designation for the certification of a large unmanned aerial system from the Federal Aviation Administration, Joby says.”
ETHICSThe Age of the Drone Police Is Here
Dhruv Mehrotra and Jesse Marx | Wired“A Wired investigation, based on more than 22 million flight coordinates, reveals the complicated truth about the first full-blown police drone program in the US—and why your city could be next. …As police departments look to expand their use of unmanned aerial aircraft, no agency has embraced the technology quite like the [Chula Vista Police Department].”
SPACEStarship Launch: Fourth Test Succeeds as Both Stages Splash Into Sea
Matthew Sparkes | New Scientist“SpaceX’s Starship, the largest rocket ever constructed, has made a successful fourth test flight, with both its first and second stages carrying out their missions as planned before splashing down into different oceans. …This fourth flight test focused on getting Starship back from orbit after its previous test reached space for the first time.”
TECHGoogle’s AI Overviews Misunderstand Why People Use Google
Kyle Orland | Ars Technica“The value of Google has always been in pointing you to the places it thinks are likely to have good answers to those questions. But it’s still up to you, as a user, to figure out which of those sources is the most reliable and relevant to what you need at that moment. …When your AI is just summarizing the top search results from around the web, it’s only ever going to be as smart or as dumb as the search engine itself. Without the human factor that helps make sense of Google’s map of the web, a Google-powered ‘AI Overview’ is always going to fail in some remarkable ways.”
SCIENCEMost Life on Earth Is Dormant, After Pulling an ‘Emergency Brake’
Dan Samorodnitsky | Quanta“Sitting around in a dormant state is actually the norm for the majority of life on Earth: By some estimates, 60% of all microbial cells are hibernating at any given time. Even in organisms whose entire bodies do not go dormant, like most mammals, some cellular populations within them rest and wait for the best time to activate.”
Image Credit: Paris Bilal / Unsplash
In 1997, Jeanne Calment passed away at the age of 122 and a half. The longest living human documented to date, she pushed the boundary of what was previously considered the maximum human lifespan.
Meanwhile, in 2023, Guinness World Records recognized Pat the mouse as the oldest mouse alive at a little over nine and a half years old—just a sliver in years compared to humans.
When it comes to lifespan, we mammals have an astonishing range. The common shrew lives less than two years; bowhead whales thrive for at least 211 years. Why the discrepancy?
Part of it, according to Dr. Steve Horvath and colleagues at the University of California, Los Angeles, comes down to epigenetics: the chemical tags attached to DNA that flip genes on or off. The type and position of these tags shift through major life events—puberty, aging—and even with dietary changes.
Unlike genetics, the study of genes coded in DNA, epigenetics better captures the “here and now” of gene expression as we go through life. Previously, Horvath and others have tapped epigenetics to develop “aging clocks” that predict a person’s biological age—that is, how old your body is biologically, rather than the number of candles on your birthday cake.
In a new study in Science Advances, Horvath’s team expanded their epigenetic clocks to predict three life-changing traits: gestation time—how long the next generation fully grows in the womb—puberty, and maximal lifespan.
“Many have suggested that epigenetic mechanisms play a role in determining lifespan,” wrote the team in the paper.
Taking advantage of data from the Mammalian Methylation Consortium, they analyzed one type of epigenetic modification in over 15,000 tissue samples across 348 mammals and developed multiple epigenetic predictors for the three life-history traits across species.
The predictors were reliable. When challenged with lifestyle and demographic factors often associated with changing epigenetic markers—for example, weight, race, and biological sex—they retained their accuracy. Surprisingly, even notable methods for extending lifespan in the lab, for example, caloric restriction, had little effect on the clock’s measures.
“This [epigenetic] signature may be an intrinsic property of each species that is difficult to change,” the team wrote.
Epigenetic IslandsHorvath is no stranger to epigenetic clocks.
Back in 2022, his team analyzed over 13,000 human tissue samples across decades of ages to develop a “measuring tape” for biological age. It sounds silly—I know how old I am. But decades of research shows that cells, tissues, and people have a biological age that doesn’t necessarily correspond to their years on Earth—“you look a lot younger than you are!”—which may be reflected in the epigenome.
The key to the aging clock was a type of epigenetic change dubbed methylation, and more specifically sections of DNA called CpG islands. In epigenetics, chemical tags usually tack on or off like Velcro. But in puberty or aging, some permanently cling onto genes, essentially shutting them off.
In other words, this particular type of epigenetic change—methylation on CpG islands—can hide a wealth of information on development, aging, and health across mammalian species. Horvath and collaborators used their results to found the Clock Foundation, a non-profit that makes epigenetic aging clocks and data more accessible for scientists to predict healthspan—how long you stay healthy with age—and lifespan.
The Mammalian Methylation Consortium is a core resource in the work. The international effort has profiled over 15,000 samples from 348 mammals, including an impressive library of exotic tissue samples—blood from harbor seals, sheep ear, naked mole rat skin. With a custom-made methylation array, the collaboration has captured roughly 36,000 highly conserved CpG islands.
Previous studies analyzing the data focused on humans; the new study took a bird’s-eye view across species.
Predicting Life HistoryThe team focused on three major “life-history traits:” gestation time, age at maturity, and maximum lifespan. To be clear, lifespan analysis is based on current records—that is, the longest living example documented for any species, rather than a theoretical projection of potential increase in lifespan.
Developing several algorithms, the team matched their prediction to a public database, AnAge, which includes extensive longevity records of multiple species. The predictor for maximum lifespan “aligned closely with those recorded in anAge,” wrote the team.
Gestation time was even more accurate—likely because it’s easier to measure—whereas the algorithm struggled to predict puberty.
Playing around with the algorithm, the team next built a separate lifespan predictor using data from young animals, before the age of five and before the onset of puberty. Surprisingly, it also worked. For species with a lifespan over 20 years, analyzing methylation had “remarkable accuracy,” wrote the team. It suggests that the maximum lifespan is somehow already imprinted into DNA samples of a species, regardless of age.
Overall, the “epigenetic indicators of life-history traits” when looking at specific species and individuals don’t always correlate with age, wrote the team.
Ready, SteadyA main criterion for any epigenetic clock is reliability. Maximum lifespan isn’t necessarily set—it’s influenced by many factors we don’t yet fully understand. Weight, demographics, diet, and hormones are already proven to lengthen or shorten overall lifespan.
The team next put their epigenetic predictor through several challenges known to alter the epigenome.
One was diet. A high-fat diet tends to slash how long mice live. The predictor linked liver samples from mice given a “cheese and butter” diet to lower maximal lifespan for these critters, compared to peers with a normal diet. However, caloric restriction, a widely used intervention that promotes longevity, didn’t change the predictor’s results. Overall, the predictor seems to be relatively stable to dietary changes that could affect lifespan, at least for mice, the team explained.
In another test, the team used the predictor to assess the maximum lifespan from blood samples of two major human studies—the Framingham Heart Study and the Women’s Health Initiative, with over 4,500 samples in total. Smoking, race, weight, metabolism, and cognitive function had no influence on the epigenetic predictor for maximum lifespan.
So, what did make a difference? Across the board, the main factor was biological sex. In 17 out of 18 analyzed mammalian species—including humans—females tended to have methylation factors that increased their lifespan by roughly one percent compared to males.
What to make of all of this?
For one, the results suggest that lifestyle behaviors—what you eat, drink, and such—may not influence the maximum bounds of lifespan, at least when measured using these epigenetic predictors. It’s a controversial idea, and the team adds caveats in their conclusion. A main one is that methylation data for human samples was obtained using a different sequencing platform, which could trip up the analysis. “Future research should revisit these findings” using a screening array similar to that used by the consortium, the team explained.
The tool also generated different predictions depending on tissue samples, with blood generally predicting a longer lifespan than, say, brain or kidney. The study used an average of all samples for their algorithm. But finding the reason behind tissue-specific differences could lead to insights into how their methylation changes with age—for any mammal.
“Together our results suggest that species maximum lifespan is strongly associated with an epigenetic signature,” wrote the team. As a next step, they hope to find interventions that can alter epigenetic lifespan.
Image Credit: Jon Tyson / Unsplash
The mysterious dark side of the moon is a challenging target for space missions. But China has now completed its second successful touchdown and collected a sample of lunar soil, which is currently on its way back to Earth.
The great space powers are currently locked in a race to return humans to the moon and even establish a more permanent presence on our nearest celestial neighbor. China’s successful mission is the third landing on the lunar surface this year, and a host of missions are scheduled over the remainder of the decade.
So far, the country is the only one to have managed the challenging feat of a soft touchdown on the side of the moon permanently facing away from Earth. With no direct line of sight, missions have to rely on a combination of automation and relayed communication signals with a heavy lag.
Despite that, the Chang’e-6 spacecraft successfully landed in the South Pole-Aitken basin on Sunday loaded with equipment for retrieving samples from the surface. On Wednesday, a vehicle loaded with roughly 4.5 pounds of lunar material launched and docked with the mission’s orbiter.
“Incredible!” tweeted Josef Aschbacher, director general of the European Space Agency (ESA), which had contributed one of the mission’s instruments. “Congratulations to the Chinese National Space Agency on the remarkable success of Chang’e-6 mission thus far. It’s a wonderful accomplishment.”
The mission involves a complicated dance between four different spacecraft modules—an orbiter, a lander, an ascender, and a reentry vehicle. After establishing a stable orbit, the lander detached from the main spacecraft and descended using an autonomous visual obstacle avoidance system to pick a safe landing spot.
Once on the surface, a drill and robotic arm collected samples and loaded them into the ascender which blasted back up into orbit. After successfully rendezvousing with the orbiter, the spacecraft will now return to Earth, where a reentry module will safely deliver the samples to a landing site in Inner Mongolia, a region in northern China.
The complexity of the mission hints at China’s broader plans for the moon. It wasn’t strictly necessary for the ascender to link back up with the orbiter, but the maneuver is likely a test run for a future manned mission in which astronauts will need to transfer from the surface to a return vehicle.
The country has said it hopes to land humans on the moon before the end of the decade and already has two more lunar missions scheduled in the coming years. The Chang’e-7 mission scheduled for 2026 will explore the Shackleton crater close to the moon’s south pole, which is seen as a promising landing site for manned missions.
And in 2028, Chang’e-8 will deliver several pieces of equipment designed to support a longer-term human presence on the moon. These include a device that will use solar power to melt lunar soil and fashion it into components and a miniature terrestrial ecosystem of plants and microbes that will test the possibility of producing food and oxygen on the lunar surface.
They aren’t the only ones eyeing the south pole though. NASA’s Artemis program is looking to land humans on the moon by 2026 and also planning to carry out various experiments aimed at supporting a permanent base on the surface. Last month, the US space agency announced the missions will carry an experiment to grow three crops on the moon and instruments designed to hunt for water ice, which can be used to create oxygen and hydrogen—a potential rocket fuel.
This new space race is probably driven as much by heightened geopolitical tensions as it is scientific or commercial value. But if it spurs a new push to move further out into the solar system, that can only be a good thing for humanity.
Image Credit: Visualization of the phases of the far side of the moon / NASA’s Scientific Visualization Studio
Chatbots are now posing as friends, romantic partners, and departed loved ones. Now, we can add another to the list: Your future self.
MIT Media Lab’s Future You project invited young people, aged 18 to 30, to have a chat with AI simulations of themselves at 60. The sims—which were powered by a personalized chatbot and included an AI-generated image of their older selves—answered questions about their experience, shared memories, and offered lessons learned over the decades.
In a preprint paper, the researchers said participants found the experience emotionally rewarding. It helped them feel more connected to their future selves, think more positively about the future, and increased motivation to work toward future objectives.
“The goal is to promote long-term thinking and behavior change,” MIT Media Lab’s Pat Pataranutaporn told The Guardian. “This could motivate people to make wiser choices in the present that optimize for their long-term wellbeing and life outcomes.”
Chatbots are increasingly gaining a foothold in therapy as a way to reach underserved populations, the researchers wrote in the paper. But they’ve typically been rule-based and specific—that is, hard-coded to help with autism or depression.
Here, the team decided to test generative AI in an area called future-self continuity—or the connection we feel with our future selves. Building and interacting with a concrete image of ourselves a few decades hence has been shown to reduce anxiety and encourage positive behaviors that take our future selves into account, like saving money or studying harder.
Existing exercises to strengthen this connection include letter exchanges with a future self or interacting with a digitally aged avatar in VR. Both have yielded positive results, but the former depends on a person being willing to put in the energy to imagine and enliven their future self, while the latter requires access to a VR headset, which most people don’t have.
This inspired the MIT team to make a more accessible, web-based approach by mashing together the latest in chatbots and AI-generated images.
Participants provided basic personal information, past highs and lows in their lives, and a sketch of their ideal future. Then with OpenAI’s GPT-3.5, the researchers used this information to make custom chatbots with “synthetic memories.” In an example from the paper, a participant wanted to teach biology. So, the chatbot took on the role of a retired biology professor—complete with anecdotes, proud moments, and advice.
To make the experience more realistic, participants submitted images of themselves that the researchers artificially aged using AI and added as the chatbot’s profile picture.
Over three hundred people signed up for the study. Some were in control groups while others were invited to have a conversation with their future-self chatbots for anywhere between 10 and 30 minutes. Right after their chat, the team found participants had lower anxiety and a deeper sense of connection with their future selves—something that has been found to translate to better decision-making, from health to finances.
Chatting with a simulation of yourself from decades in the future is a fascinating idea, but it’s worth noting this is only one relatively small study. And though the short-term results are intriguing, the study didn’t measure how durable those results might be or whether longer or more frequent chats over time might be useful. The researchers say future work should also directly compare their method to other approaches, like letter writing.
It’s not hard to imagine a far more realistic version of all this in the near future. Startups like Synthesia already offer convincing AI-generated avatars, and last year, Channel 1 created strikingly realistic avatars for real news anchors. Meanwhile OpenAI’s recent demo of GPT-4o shows quick advances in AI voice synthesis, including emotion and natural cadence. It seems plausible one might tie all this together—chatbot, voice, and avatar—along with a detailed back story to make a super-realistic, personalized future self.
The researchers are quick to point out that such approaches could run afoul of ethics should an interaction depict the future in a way that results in harmful behavior in the present or endorse negative behaviors. This is an issue for AI characters in general—the greater the realism, the greater the likelihood of unhealthy attachments.
Still, they wrote, their results show there is potential for “positive emotional interactions between humans and AI-generated virtual characters, despite their artificiality.”
Given a chat with our own future selves, maybe a few more of us might think twice about that second donut and opt to hit the gym instead.
Image: MIT Media Lab
On a fall evening in 2022, scientists at the Johns Hopkins University Applied Physics Laboratory were busy with the final stages of a planetary defense mission. As Andy Rivkin, one of the team leaders, was getting ready to appear in NASA’s live broadcast of the experiment, a colleague posted a photo of a pair of asteroids: the half-mile-wide Didymos and, orbiting around it, a smaller one called Dimorphos, taken about 7 million miles from Earth.
“We were able to see Didymos and this little dot in the right spot where we expected Dimorphos to be,” Rivkin recalled.
After the interview, Rivkin joined a crowd of scientists and guests to watch the mission’s finale on several big screens: As part of an asteroid deflection mission called DART, a spacecraft was closing in on Dimorphos and photographing its rocky surface in increasing detail.
Then, at 7:14 pm, a roughly 1,300-pound spacecraft slammed head-on into the asteroid.
Within a few minutes, members of the mission team in Kenya and South Africa posted images from their telescopes, showing a bright plume of debris.
In the days that followed, researchers continued to observe the dust cloud and discovered it had morphed into a variety of shapes, including clumps, spirals, and two comet-like tails. They also calculated that the impact slowed Dimorphos’ orbit by about a tenth of an inch per second, proof-of-concept that a spacecraft—also called a kinetic impactor—could target and deflect an asteroid far from Earth.
Ron Ballouz, a planetary scientist at the lab commented that what is often seen in the movies is a “sort of last-ditch-effort, what we like to call a final-stage of planetary defense.” But if hazardous objects can be detected years in advance, other techniques like a kinetic impactor can be used, he added.
If a deflection were necessary, scientists would need to change the speed of a hazardous object, such as an asteroid or comet, enough that it doesn’t end up at the same place and time as Earth as they orbit the sun. Rivkin said this translates into at least a seven-minute change in the arrival time: If a Dimorphos-sized object were predicted to collide with Earth 67 years from now, for instance, the slow-down that DART imparted would be just enough to add up to the seven minutes, he added.
With less lead time, researchers could use a combination of multiple deflections, larger spacecrafts, or boosts in speed, depending on the hazardous object. “DART was designed to validate a technique and specific situations would inevitably require adapting things,” said Rivkin.
Researchers use data from DART and smaller-scale experiments to predict the amount of deflection using computer simulations.
“What is often seen in the movies is a ‘sort of last-ditch-effort, what we like to call a final-stage of planetary defense.'”Scientists are also focusing on the type of asteroid that Dimorphos appears to be: a “rubble pile,” as they call it, because objects of this kind are thought to be made of clumps of many rocks.
In fact, scientists think that most asteroids the size of Dimorphos and larger are rubble piles. As scientists continue to learn more about rubble piles, they will be able to make better predictions about deflecting asteroids or comets. And in 2026, a new mission will arrive at Didymos and Dimorphos to collect more data to fine-tune the computer models.
In the meantime, researchers are trying to learn as much as possible in the unwelcome case an asteroid or comet is discovered to be a threat to Earth and a more rapid response is necessary.
…
Scientists first suspected that many asteroids are rubble piles about 50 years ago. Their models showed that when larger asteroids smashed into one another, the collisions could throw off fragments that would then reassemble to form new objects.
It wasn’t until 2005, though, that scientists saw their first rubble pile: asteroid Itokawa, when a spacecraft visited it and photographed it. Then, in 2018, they saw another called Ryugu, and later that year, one more, asteroid Bennu. DART’s camera also showed Didymos and Dimorphos are likely of the same variety.
“It’s one thing to talk about rubble piles, but another to see what looks like a bunch of rocks dumped off a truck up close,” said William Bottke, a planetary scientist at the Southwest Research Institute in Boulder, Colorado.
Scientists suspect that rubble piles have large amounts of empty space between their rocks. They believe these piles are bound together with very weak forces and mostly gravity, meaning they could break apart more easily than an asteroid that is a single boulder. This was evident with Dimorphos, as DART excavated over an estimated ten thousand tons of material. The plume of debris, in turn, acted like a rocket thruster, providing an extra push in the opposite direction, slowing the asteroid. So, although the asteroid’s void spaces may have absorbed some of the DART impact, the blast of debris increased the amount of deflection, with estimates ranging between about two and five times as much as the push by the spacecraft alone.
Sabina Raducan, a planetary scientist at the University of Bern in Switzerland, cautioned, though, that care must be taken if kinetic impactors ever need to be used on smaller rubble piles.
Raducan and her team used a computer model to apply the results of the DART impact on a variety of rubble piles—the first time such research has been done. The results, which were published in The Planetary Science Journal, show that a DART-sized spacecraft impacting at the speed it did, about 3.7 miles per second, could break a rubble pile less than 80 meters in diameter into many pieces. Some of the boulders, in turn, could end up impacting Earth, potentially causing injuries and damage.
Raducan wrote in a follow-up email that despite the success of DART, a similar scenario may not always be optimal for all asteroids.
Instead, she added, the size or speed of a spacecraft may need to be adjusted for a successful deflection.
“‘It’s one thing to talk about rubble piles, but another to see what looks like a bunch of rocks dumped off a truck up close.'”The possible breakup of materials could also relate to comets. These objects are similar to asteroids, except they contain ices such as water or carbon dioxide. When comets pass close to the sun, these materials turn into gases, which can act like a rocket booster and push the comet faster. Hence, if researchers aim to knock a comet off of a crash course with Earth, they’d have to consider the possibility that ices could be exposed or buried, which could change its speed and possibly require further deflections.
Rivkin said that comet collisions with Earth are relatively rare compared to asteroids, but there are “definitely a lot of extra things to keep track of.”
Also complicating matters: Some objects that are classified as asteroids could also contain buried ices.
“Things get very murky, though,” said Bottke. “We have seen asteroids develop tails,” similar to those found in comets.
…
Scientists are eagerly waiting for late 2026, when a spacecraft called Hera, as part of a planetary defense mission led by the European Space Agency, in collaboration with the Japan Aerospace Exploration Agency, is scheduled to arrive at the Didymos system. There, it will deploy two smaller satellites, and together they will begin to study the pair of asteroids up close. In particular, researchers are looking forward to finally being able to measure the mass of Dimorphos, which will allow them to better refine their estimate of how much of a push the spacecraft and the blast of debris imparted. Hera and satellites will also take measurements that will enable scientists to calculate the density and strength of Dimorphos which can be used in impact models.
The Hera mission will also allow scientists to see what DART did to Dimorphos. The preliminary measures suggest that the asteroid is so weak that the impact changed its shape rather than leaving behind a crater: “I really want to see the outcome,” said Raducan. “Is it a crater or not?”
“The blast or tsunami from the impact of an asteroid like Bennu would be capable of causing fatalities and damage on regional or continental scales.”A new shape, in turn, may have altered Dimorphos’ orbit around Didymos. Hera will allow scientists to check, which will help them better understand the response of kinetic impacts on asteroids that have one or more moons. Currently, about 16 percent of near-Earth asteroids larger than about 650 feet in diameter are estimated to be binaries, or systems of two. Earth is thought to have received a double hit 458 million years ago that left behind the Lockne and Målingen craters in Sweden.
Hera and its satellites will also collect measurements of the material properties of Didymos, which will also help advance scientists’ knowledge of rubble piles and deflections. Rivkin said that they only got a quick view of Didymos as DART sped past it.
In the meantime, researchers are busy analyzing samples of the surface of asteroid Bennu that a NASA spacecraft called OSIRIS-REx returned to Earth in the fall of 2023. The results will help researchers understand the asteroid’s material properties better. The approximately 1,600-foot-wide Bennu is the most potentially hazardous object known (as of May 14, 2024), with a 0.037 percent chance of impacting Earth on September 24, 2182.
Ballouz noted the blast or tsunami from such an impact would be capable of causing fatalities and damage on regional or continental scales. He added that should Bennu remain a hazard and if deflections are deemed necessary, it’d require multiple kinetic impacts due to its large size. The observations and measurements from when OSIRIS-REx observed Bennu up-close, which took place up to 2021, along with the results of the sample returns, would be invaluable for planning kinetic impactor missions to the asteroid, if necessary. Additional spacecraft missions to re-study the asteroid or even collect more samples could also be organized, to help inform impact models even more.
It’s never good news to hear of discoveries of potential threats to Earth, but knowing in advance of the possibility at least allows scientists to take action, unlike with some natural hazards that happen without warning.
“It’s important for people to be aware that impacts have affected Earth in the past and there is this possibility in the future,” said Ballouz. “There should also be a general awareness that there are people who are studying this aspect of how we interact with space.”
This article was originally published on Undark. Read the original article.
Image Credit: A SpaceX Falcon 9 rocket launches the Double Asteroid Redirection Test, or DART, spacecraft in the first full-scale planetary defense test. (NASA/Bill Ingalls)
In the middle of the night, when stumbling into the kitchen for a glass of water, I rely on touch. Even in the dark, my fingers can feel the handrails—alerting me to a flight of stairs—the walls, the curvature of the cabinet handles, and eventually, the glass.
We humans are highly visual creatures, but our skin has a sophisticated built-in system that lets us grab a cup of coffee, peel an overripe banana, or type on a keyboard without a second thought.
Part of this dexterity comes from a separation of powers: Some components in the skin detect direct forces from the outside world—for example, pressure or pinches. Others feel when the skin is stretched—like when you open your hand to grab an extra-large mug.
Now, a team from China and Singapore has recapitulated these qualities in artificial skin. Mimicking the division of labor, the three-dimensional electronic skin, dubbed 3DAE-Skin, adopts a skin-like multilayer construction.
Using microfabrication techniques, the team engineered a flexible device—roughly the size of the tip of a human thumb—that can sense different types of forces at a resolution similar to its human counterpart.
Combined with electrical circuits that acquire data and process it with a custom deep learning algorithm, the skin could tease apart tiny differences in touch. In one study, like a blindfolded but seasoned grocery shopper, it gauged the ripeness of multiple fruits with a squeeze or two and decided whether a croissant or cake was overly stale with a poke.
While charming, a robotic grocery shopper is hardly the goal. The strategy suggests that mimicking the human skin’s architecture is a promising way to make electronic components that “feel.”
When embedded in robots or prosthetic hands, for example, the device could discern between different types of surfaces. And though it currently mimics fingertips, the skin could be used for other appendages or non-humanoid robotic systems, helping them navigate uneven terrain, for example, or in robot-assisted surgeries. The results were published in Science.
Three-Layer CakeOur skin is the largest organ in the body, but outside skin care and the occasional scrapes, it’s often ignored.
But to bioengineers, the skin is a wonder. Every inch is heavily embedded with myriad sensors to measure temperature, pain, pressure, vibration, and stretch. There are six identified sensors specifically for touch, and they’re distributed across the skin in a clever way.
Unlike a crepe, the skin is more like a three-layer cake. Sprinkled across the top layers are sensors for temperature changes and different types of touch sensations. It makes sense: If you need to immediately move your hand from a hot pan or a needle prick, it’s better for these biological sensors to be closer to the surface. But top-layer skin cells constantly renew, and sensors too close to the surface face the risk of rapidly being eliminated.
A group of promising sensors are called Merkel cells, which sit between the top and middle layers and readily detect external forces like a prick or pinch. These cells translate mechanical data—pushes and pulls—into electrical pulses sent to the brain. With more processing, Merkel cells can create a general “feeling” of an object’s structure and overall texture.
Deeper down, but still in the skin’s middle layer, are sensors called Ruffini endings. These cells don’t care much for immediate touch sensations. Rather, they keep tabs on how much the skin stretches. For example, when flexing your hand into a claw, even when there aren’t any extraneous pokes that made you do that, you can still feel the strain of the skin thanks to these cells.
Both sensors are spread across the skin’s three-dimensional architecture. Imitating this setup could make it possible to develop an electronic artificial skin that feels both pokes and stretching at the same time, the team wrote in their paper.
Meet 3DAE-SkinThe artificial skin has three layers, with the thickness of each similar to its human counterpart.
Most of the skin’s sensing components and electrical circuits are embedded between the top and middle layers. Force sensors act like Merkel cells, transferring immediate touch signals. Strain sensors, in contrast, resemble Ruffini endings and send stretch signals.
The completed artificial skin is made of individual units. Each has nine layers, with two force-sensing and two strain-sensing components and five “buffer” layers in between to stabilize them.
The team next sandwiched all this between two flexible layers of stretchy bio-compatible material. The device could withstand twists, pinches, and other skin deformations, and could sense pressure and strain like your average human skin.
Next came communications. The team added a data acquisition circuit, imprinted onto the surface layers, and a signal processing hub. The hub visualizes data from all 240 sensors in the device and deciphers different types of force and where they come from. The overall device is the size of the tip of a thumb.
Finally, the researchers developed an AI to translate the gathered data into predictions of touch. They trained the algorithm on a sample of 51 simple shapes—for example, a ball, a tube, or a piece of paper and how each reacts to pressure—does it maintain its shape or rapidly deform?
“When touching an object, human fingertips usually rely on” how it feels in order to infer the softness and shape of what we’re touching, explained the team in their paper.
While it sounds simple, it’s a tough calculation. Like squeezing an avocado to check if it’s ripe, you have to have an idea of how much to squeeze and when to stop.
After training on nearly 30,000 scenarios, the AI could predict the forces needed to grab an object—for example, an overripe banana. Like human fingertips, it could sense how much the object distorted its shape to fine-tune its grip.
In one test, the team hooked up a clamp-like probe equipped with the electronic skin and pressed unripe and ripe plums. The AI could readily discern between the two, even though the fruit looked relatively similar. In another test, a clamp easily detected a fresh bread roll from a tough, stale one.
Buying groceries is hardly the first use that comes to mind when building artificial skin. But the study shows that mimicking human skin’s biology, with a dose of AI, could lead to better artificial skin for robotics and prosthetics. These bioinspired 3D designs, combined with increasingly sophisticated microfabrication technologies, raises the bar for bio-compatible materials that match the resolution of the human skin for processing multiple senses.
Although 3DAE-Skin was developed to mimic fingertips, “the design and fabrication methods are scalable” and can be used for other types of prosthetic and robotic systems, the team wrote.
Image Credit: Tsinghua University
ARTIFICIAL INTELLIGENCEOpenAI Says It Has Begun Training a New Flagship AI Model
Cade Metz | The New York Times“The San Francisco startup, which is one of the world’s leading AI companies, said in a blog post that it expected the new model to bring ‘the next level of capabilities’ as it strove to build ‘artificial general intelligence,’ or AGI, a machine that can do anything the human brain can do. The new model would be an engine for AI products including chatbots, digital assistants akin to Apple’s Siri, search engines and image generators.”
ROBOTICSWill Scaling Solve Robotics?
Nishanth J. Kumar | IEEE Spectrum“Developing a general-purpose robot, one that can competently and robustly execute a wide variety of tasks of interest in any home or office environment that humans can, has been perhaps the holy grail of robotics since the inception of the field. And given the recent progress of foundation models, it seems possible that scaling existing network architectures by training them on very large datasets might actually be the key to that grail.”
FUTURE OF FOODGene-Edited Salad Greens Are Coming to US Stores This Fall
Emily Mullin | Wired“Last year, startup Pairwise started selling the first food in the US made with CRISPR technology: a new type of mustard greens with an adjusted flavor. …The company introduced the greens to the food service industry—select restaurants, cafeterias, hotels, retirement centers, and caterers—in just a few cities. A single grocery store in New York City also stocked them. Now, biotech giant Bayer has licensed the greens from Pairwise and plans to distribute them to grocery stores across the country.”
TECHThe New ChatGPT Offers a Lesson in AI Hype
Brian X. Chen | The New York Times“When OpenAI unveiled the latest version of its immensely popular ChatGPT chatbot this month, it had a new voice possessing humanlike inflections and emotions. The online demonstration also featured the bot tutoring a child on solving a geometry problem. To my chagrin, the demo turned out to be essentially a bait and switch. The new ChatGPT was released without most of its new features, including the improved voice (which the company told me it postponed to make fixes). The ability to use a phone’s video camera to get real-time analysis of something like a math problem isn’t available yet, either.”
COMPUTINGWorld’s Thinnest Lens Is Just Three Atoms Thick
Michael Irving | New Atlas“[A Fresnel lens uses] a series of concentric circles of material to diffract light into a focal point, sacrificing some image clarity but allowing for much thinner lenses. And now, scientists have pushed that almost to the limit, creating a lens that’s just 0.6 nanometers (nm) thick, or only three measly atoms. That makes it the thinnest lens ever built, beating the previous record from 2016 which was 10 times thicker at 6.3 nm.”
ARTIFICIAL INTELLIGENCEWhy Google’s AI Overviews Gets Things Wrong
Rhiannon Williams | MIT Technology Review“On Thursday, Liz Reid, head of Google Search, announced that the company has been making technical improvements to the system to make it less likely to generate incorrect answers, including better detection mechanisms for nonsensical queries. It is also limiting the inclusion of satirical, humorous, and user-generated content in responses, since such material could result in misleading advice. But why is AI Overviews returning unreliable, potentially dangerous information? And what, if anything, can be done to fix it?”
ENERGY1-bit LLMs Could Solve AI’s Energy Demands
Matthew Hutson | IEEE Spectrum“For LLMs that are cheap, fast, and environmentally friendly, they’ll need to shrink, ideally small enough to run directly on devices like cellphones. …Researchers have long compressed networks by reducing the precision of [their] parameters—a process called quantization—so that instead of taking up 16 bits each, they might take up 8 or 4. Now researchers are pushing the envelope to a single bit.”
AUGMENTED REALITYA New Computer Uses AR Glasses to Create a 100-Inch Virtual Workspace
Kyle Barr | Gizmodo“The problem with the term ‘spatial computer’ is that most devices using the obtuse marketing term don’t actually look like computers. Yes, the Apple Vision Pro or Meta Quest 3 meet the definition of ‘computer,’ but most people still think of ‘PC’ as a desktop or a laptop. So now there’s the Spacetop G1, an AR laptop, trying to kick both the desktop and VR markets for being too stuck in their ways.”
BIOTECHWorld-First Tooth-Regrowing Drug Will Be Given to Humans in September
Bronwyn Thompson | New Atlas“The world’s first human trial of a drug that can regenerate teeth will begin in a few months, less than a year on from news of its success in animals. This paves the way for the medicine to be commercially available as early as 2030. …The intravenous treatment will be tested for its efficacy on human dentition, after it successfully grew new teeth in ferret and mouse models with no significant side effects.”
ETHICSHumanity Needs an Ethical Upgrade to Keep Up With New Technologies
Marcelo Gleiser | BigThink“Technological advancements like nuclear weapons, genetic engineering with CRISPR, and artificial intelligence present significant ethical challenges and responsibilities. While it’s rational to be concerned about technological threats and dilemmas, many people react by blaming scientists or science itself, failing to recognize the difference between those who control the application of scientific findings.”
Image Credit: NASA, ESA, Joel Kastner (RIT)
We all know the drill for reproduction—sperm meets egg.
For the past decade, scientists have been pushing the boundaries of where the two halves come from. Thanks to induced pluripotent stem cell technology, it’s now possible to scrape skin cells from mice and transform them into functional sperm or egg cells that give rise to healthy pups born from two moms or dads. The recipe may even open the door for single parent offspring—at least for mice.
But mice aren’t people. And the same recipe doesn’t work for human reproductive cells.
One reason, according to Dr. Mitinori Saitou at Kyoto University, is due to the chemical “tags,” known collectively as the epigenome, that control when certain genes are turned on or off. Like a ledger, these tags maintain a sort of memory in early reproductive cells. For cells to eventually develop into sperm or eggs, those memories must be wiped clean.
This month, Saitou’s team developed a recipe to do just that. Starting from primordial germ cells—a type of cell that eventually develops into either sperm or egg—they added a single protein ingredient that nudged their growth further along by erasing their chemical memories.
“Our study represents…a fundamental advance in our understanding of human biology and the principles behind epigenetic reprogramming in humans,” Saitou said in a press release. It’s also “a true milestone” for generating sperm and eggs in the lab, he said, which could potentially help couples struggling with infertility.
Tag, You’re ItEpigenetic tags control how our genes are expressed. Picture DNA’s double helix. Then imagine sticking small chemical “pins” into the structure. These pins readily block the transcription of our genetic blueprint into biological messages—make this protein, not that.
It may sound nefarious, but epigenetic tags are fundamental to our bodies. Most of our cells have the same DNA—it’s the expression of that DNA that allows them to form different tissues and organs and guides biological processes. Depending on where, and which, tags are added to the double helix, some shut down entire genes—sometimes for life.
But epigenetics becomes a hurdle when growing gametes—egg or sperm cells—in a dish. Called in vitro gametogenesis, the technology allows scientists to take a closer look at how gametes develop and potentially help couples struggling with infertility.
It also offers a way to fix DNA that leads to inherited diseases, especially if only one parent has the mutation. Although scientist have edited genes directly in early human embryos using CRISPR-Cas9, the approach is prone to errors and can cause potentially dangerous side effects. Rewriting DNA in sperm and egg cells is simpler—the cells can readily repair DNA, a critical step in gene editing—and lab-grown specimens are the perfect canvas to experiment on.
The problem? Their epigenetic chemical tags form a sort of “memory,” which eventually causes them to stop developing. The body naturally wipes the tags away, a process dubbed epigenetic reprogramming, so that early reproductive cells can grow into healthy egg or sperm.
While scientists can already replicate the process in mice, the same recipe doesn’t work in human cells. Why this is so is still a mystery. In the new study, Saitou’s team set out to find an epigenetic “reset” button tailored to human reproductive cells.
One Protein to Rule Them All?Saitou is hardly new to the field. Previously, his team coaxed stem cells into another type of cell that roughly resembled early reproductive cells. They recapitulated several fundamental characteristics of their natural counterpart—such as the capacity to rapidly multiply.
However, their epigenetic landscape remained intact, eventually halting the cells’ development. As a workaround, the team mixed them with mouse cells from reproductive regions in the body to mimic the microenvironment of the ovary or testis. It worked—molecular signals from the supporting cells eased off the epigenetic brake, allowing the lab-grown early reproductive cells to further develop into immature gametes that, in theory, could become egg and sperm.
While successful in concept, the process was highly inefficient, with roughly one cell out of ten able to develop further. And mixing human cells with mouse cells could have unintended consequences, which isn’t ideal or practical for studying human reproduction. But the results sparked an idea: Some molecules could switch on epigenetic reprogramming in lab-grown early reproductive cells—they just needed to find them.
The new study homed in on one. Dubbed BMP2, the protein is familiar to scientists for its role in development—for example, forming bones and cartilage. But “it was highly unexpected that it also drives…epigenetic reprogramming,” said Saitou.
When added to lab-grown early reproductive cells, they developed further than previous attempts, forming precursors to human sperm and eggs inside a dish. The cells had a similar genetic and epigenetic profile to their natural counterparts and could rapidly proliferate—in some cases, over 10 billion-fold.
It’s “near-indefinite amplification…we now also have the ability to store and re-expand these cells as needed,” said Saitou.
However, even with BMP2, the treated cells couldn’t develop into fully mature sperm and eggs. Carefully analyzing the cell’s epigenome, the team found some epigenetic marks still remained—suggesting the reprogramming wasn’t complete.
While a headache for research, these epigenetic “stragglers” could have devastating consequences if lab-grown reproductive cells were ever used in the clinic to assist reproduction. If even a single gene is wrongly imprinted by epigenetic marks, it could lead to serious disease.
Digging deeper, the team found an entire network of molecules that could explain why BMP2 triggered epigenetic reprogramming—even though it wasn’t complete. One possible reason is it altered the activity of a protein that adds epigenetic “tags” to DNA, “but further investigation will be necessary to determine the precise mechanism and whether this is direct or indirect,” said Saitou.
In vitro gametogenesis is still in its infancy, and there’s much left to learn. But with mice already born from lab-made eggs, there’s no doubt the field is rapidly advancing—along with weighty ethical and social questions. Lab-grown gametes offer a way to rapidly experiment with gene editing to cure diseases for the next generation. But like the notorious CRISPR babies, if fertilized, they could lead to permanently gene-edited humans with the capacity to pass genetic modifications along to their children.
Saitou is well aware of the risks, and he welcomes public discussion.
“Many challenges remain and the path will certainly be long, especially when considering the ethical, legal, and social implications associated with the clinical application of human IVG [in vitro gametogenesis],” he said. “Nevertheless, we have now made one significant leap forward towards the potential translation of IVG into reproductive medicine.”
Image Credit: Gerd Altmann / Pixabay
Quantum computers could solve some of the world’s most challenging problems, but only if we can make them big enough. A new modular design for quantum chips could make building large-scale quantum computers far more feasible.
While there has been significant progress in building ever larger quantum processors, the technology is still light years from the kind of scale seen in conventional computer chips.
The inherent fragility of most qubit technologies combined with the complex control systems required to manipulate them mean that leading quantum computers based on superconducting qubits have only just crossed the 1,000-qubit mark.
A new platform designed by engineers at MIT and the MITRE Corporation could present a more scalable solution though. In a recent paper in Nature, they incorporated more than 4,000 qubits made from tiny defects in diamonds onto an integrated circuit, which was used to control them. In the future, several of these so-called “quantum systems-on-a-chip” could be connected using optical networking to create large-scale quantum computers, the researchers say.
“We will need a large number of qubits, and great control over them, to really leverage the power of a quantum system and make it useful,” lead author Linsen Li from MIT said in a press release. “We are proposing a brand-new architecture and a fabrication technology that can support the scalability requirements of a hardware system for a quantum computer.”
Defects in diamonds known as color centers are promising qubit candidates because they hold their quantum states for much longer than competing technologies and can be entangled with distant qubits using light signals. What’s more, they’re solid-state systems compatible with conventional electronics manufacturing.
One of the main downsides is diamond color centers are not uniform. Information is stored in a quantum property known as “spin,” but scientists use optical signals to manipulate or read the qubits. The frequency of light each color center uses can vary significantly. In one sense, this is beneficial because they can be individually addressed, but it also makes controlling large numbers of them challenging.
The researchers got around this by integrating their qubits on top of a chip that can apply voltages to them. They can then use these voltages to tune the qubits’ frequencies. This makes it possible to tune all 4,000 to the same frequency and allows every qubit to be connected to every other one.
“The conventional assumption in the field is that the inhomogeneity of the diamond color center is a drawback,” MIT’s Dirk Englund said in the press release. “However, we turn this challenge into an advantage by embracing the diversity of the artificial atoms: Each atom has its own spectral frequency. This allows us to communicate with individual atoms by voltage tuning them into resonance with a laser, much like tuning the dial on a tiny radio.”
Key to their breakthrough was a novel fabrication technique allowing the team to create 64 “quantum microchiplets”—small slivers of diamond featuring multiple color centers—which they then slotted into sockets on the integrated circuits.
They say the approach could be applied to other solid-state quantum technologies and predict they’ll ultimately achieve qubit densities comparable to the transistor densities found in conventional electronics.
However, the team has yet to actually use the device to do any computing. They show they can efficiently prepare and measure spin states, but there’s still some way to go before they can run quantum algorithms on the device.
They’re not the only ones assembling large numbers of qubits that can’t do very much yet. Earlier this year researchers from Caltech reported they had made an array of 6,100 “neutral-atom” qubits.
Nonetheless, this highly scalable modular architecture holds considerable promise for getting us closer to the millions of qubits needed to achieve the technology’s true promise.
Image Credit: Sampson Wilcox and Linsen Li, RLE
The opaque inner workings of AI systems are a barrier to their broader deployment. Now, startup Anthropic has made a major breakthrough in our ability to peer inside artificial minds.
One of the great strengths of deep learning neural networks is they can, in a certain sense, think for themselves. Unlike previous generations of AI, which were painstakingly hand coded by humans, these algorithms come up with their own solutions to problems by training on reams of data.
This makes them much less brittle and easier to scale to large problems, but it also means we have little insight into how they reach their decisions. That makes it hard to understand or predict errors or to identify where bias may be creeping into their output.
A lack of transparency limits deployment of these systems in sensitive areas like medicine, law enforcement, or insurance. More speculatively, it also raises concerns around whether we would be able to detect dangerous behaviors, such as deception or power seeking, in more powerful future AI models.
Now though, a team from Anthropic has made a significant advance in our ability to parse what’s going on inside these models. They’ve shown they can not only link particular patterns of activity in a large language model to both concrete and abstract concepts, but they can also control the behavior of the model by dialing this activity up or down.
The research builds on years of work on “mechanistic interpretability,” where researchers reverse engineer neural networks to understand how the activity of different neurons in a model dictate its behavior.
That’s easier said than done because the latest generation of AI models encode information in patterns of activity, rather than particular neurons or groups of neurons. That means individual neurons can be involved in representing a wide range of different concepts.
The researchers had previously shown they could extract activity patterns, known as features, from a relatively small model and link them to human interpretable concepts. But this time, the team decided to analyze Anthropic’s Claude 3 Sonnet large language model to show the approach could work on commercially useful AI systems.
They trained another neural network on the activation data from one of Sonnet’s middle layers of neurons, and it was able to pull out roughly 10 million unique features related to everything from people and places to abstract ideas like gender bias or keeping secrets.
Interestingly, they found that features for similar concepts were clustered together, with considerable overlap in active neurons. The team says this suggests that the way ideas are encoded in these models corresponds to our own conceptions of similarity.
More pertinently though, the researchers also discovered that dialing up and down the activity of neurons involved in encoding these features could have significant impacts on the model’s behavior. For example, massively amplifying the feature for the Golden Gate Bridge led the model to force it into every response no matter how irrelevant, even claiming that the model itself was the iconic landmark.
The team also experimented with some more sinister manipulations. In one, they found that over-activating a feature related to spam emails could get the model to bypass restrictions and write one of its own. They could also get the model to use flattery as a means of deception by amping up a feature related to sycophancy.
The team say there’s little danger of attackers using the approach to get models to produce unwanted or dangerous output, mostly because there are already much simpler ways to achieve the same goals. But it could prove a useful way to monitor models for worrying behavior. Turning the activity of different features up or down could also be a way to steer models towards desirable outputs and away from less positive ones.
However, the researchers were keen to point out that the features they’ve discovered make up just a small fraction of all of those contained within the model. What’s more, extracting all features would take huge amounts of computing resources, even more than were used to train the model in the first place.
That means we’re still a long way from having a complete picture of how these models “think.” Nonetheless, the research shows that it is, at least in principle, possible to make these black boxes slightly less inscrutable.
Image Credit: mohammed idris djoudi / Unsplash
Imagine trying to breathe through a plastic bag. You struggle. Your heart starts to race. Your head gets dizzy. And all you did was stroll through a park full of blooming flowers.
Asthma is one of the most common breathing disorders in the world, affecting over 300 million people. Inhalers can temporarily open airways in emergencies, but there’s no cure.
Now, scientists from Tsinghua University in Beijing have found a way to keep asthma attacks in check with just a single shot. Using CAR T cell therapy, they designed a “living drug” to hunt down one of the main triggers for asthma, a group of immune cells called eosinophils. The treatment reduced lung inflammation and warded off asthma symptoms in mice for six months. In multiple asthma models, the therapy lowered lung inflammation, quieted eosinophils in the airways, and re-opened airways.
Managing asthma currently focuses on relieving symptoms and requires nearly life-long treatment. The study suggests a single CAR T dose could keep asthma at bay, “marking a potential paradigm shift in the management of common chronic disease,” the team wrote in a paper published in Nature Immunology.
Chronic DilemmaIn CAR T cell therapy, scientists extract T cells from the body, genetically engineer them to produce protein “hooks” that grab onto a target of interest—say, a cancer cell—and infuse them back into the body. The amped-up cells then hunt down and destroy the enemy.
This protein handshake is the core of CAR T. Proteins dotted along the surfaces of T cells—a type of immune cell—readily grab onto another protein target. Targets can be on the surfaces of other cells, bacteria, or even viruses, such as HIV. Thanks to CRISPR-Cas9 and other gene editing tools, adding customized protein “hooks” is easier and more efficient than ever.
Six CAR T therapies have been approved for a variety of blood cancers. Meanwhile, scientists are exploring whether CAR T can tackle glioblastoma—an especially difficult type of brain cancer—autoimmune disorders, Type 1 diabetes, or even aging.
But CAR T therapy for chronic disorders is still difficult, the authors wrote in the paper.
The first challenge is finding the right target. In blood cancer, the revamped T cells are like molecular torpedoes that mercilessly seek and destroy tumor cells. But chronic illnesses often involve a myriad of cells and proteins working together, and they can affect multiple organs. This makes finding a target more difficult, especially if the target has some normal uses too.
The second is longevity. CAR T cells divide like other cells, but over time some get worn out and no longer renew. These exhausted T cells eventually die off, and the disease rebounds.
Then there’s the potential of an immune civil war. For blood cancers, people need to undergo chemotherapeutic conditioning of the bone marrow—a source of immune cells—to make space for the new modified cells. It’s a necessary step, and though grueling, is acceptable for devasting diseases. But serious side effects make it a tough call for non-life-threatening diseases.
The new study took in all these concerns and engineered a single shot CAR T solution for asthma without the need for conditioning.
Long Live CAR TIn asthma, eosinophils infiltrate the airways.
These white blood cells usually fight off infection. But when overzealous, they trigger hyper-inflammation and damage delicate cells lining the lungs. Mucus also builds up and blocks the airways. Meanwhile, a soup of immune molecules accumulates and adds to the inflammation.
Drugs targeting immune molecules have been approved for asthma, but they only provide temporary relief, wrote the authors. They don’t cure the disorder, and because they need to be taken constantly over years or even decades, they’re not cost effective.
The new study sought to lower eosinophil levels by targeting IL-5R, a protein on the cells’ surfaces. It’s not the first time scientists have tried the strategy. A previous study that inhibited IL-5R function using antibodies was found to be safe in patients, but it required repeated dosing and risked the body becoming immune to the antibodies.
The team genetically added protein “hooks” targeting IL-5R onto T cells extracted from mice and transferred a million of the CART cells back into the mice without any conditioning. Surprisingly, the experiment was a dud. The cells died off in a week, and eosinophil levels bounced back.
What now? Earlier this year, the same team found that wiping away two genes transformed T cells into an “immortal-like” state. In a second try, they added the IL-5R-targeting protein to these T cells. It worked. When infused into mice, the upgraded cells expanded and flourished, and they lived happily alongside unmodified T cells in tissues in the lung and liver.
In multiple types of asthma models, the CAR T cells damped lung inflammation and dramatically lowered eosinophil levels. When given four weeks before an especially severe asthma attack induced by chemicals, the therapy protected against lung inflammation and symptoms.
For a more “real-life” scenario, the team next turned to dust mites—a common trigger for asthma and allergy attacks. They exposed mice to dust mites for roughly two months (a bit like living in a dusty basement) and then gave them a shot of CAR T cells or an antibody treatment previously used to reduce asthma symptoms. Four weeks later, the CAR T cell treatment was more effective at reducing mucus buildup and keeping lung inflammation at bay.
But would the therapy last? The team followed severely asthmatic mice for six months after the CAR T shot. The engineered cells lowered lung inflammation and kept the mice chipper and healthy—even when challenged with another asthma trigger.
Immune complications were a worry. But the treatment proved safe. The mice had normal bone marrow and spleen function—both sites generate blood and immune cells. During an asthma attack, CAR T cells rapidly expanded in numbers to lower eosinophil activity. With the job done, they fell back to a baseline “surveillance” number, without stressing the body.
Compared to natural T cells, the genetically altered “immortal-like” cells are perfect for long-lasting CAR T therapies, wrote the authors. In other tests, they added a mechanism into the cells to block an inflammatory protein in addition to eosinophils, with even better results.
The study is the latest example of CAR T therapy beyond cancer. But there’s still reason to be careful. In cancer treatments, the Food and Drug Administration (FDA) is investigating the risk of developing secondary tumors. This likely depends on the protein target, but long-living CAR T cells could also have unforeseen side effects, especially if used for chronic diseases.
For now, the team is planning to test a similar strategy in allergies and other breathing diseases related to eosinophils, such as chronic obstructive pulmonary disease (COPD).
Image Credit: crystal light / Shutterstock.com
By the time you finish reading this, around 30 children will have been saved thanks to vaccines.1
Over the last 50 years, that adds up to 150 million children.2 That’s more than twice the population of the United Kingdom.
That’s 150 million children who will grow up, experience life, and contribute to the world; over a hundred million sets of parents who were spared the tragedy of having to bury their children.
This figure comes from a new study from Andrew Shattock and other researchers from around the world. They estimated the number of lives saved from vaccinations against different diseases over the past 50 years.3
The two charts below show the number of lives saved, broken down by disease and region.
Vaccination against measles has had the biggest impact, saving 94 million lives over the last 50 years—more than 60 percent of the total.4
This has been a truly global effort, with more than 5 million children saved in every region, including over 50 million in Africa and 38 million in Southeast Asia. You can see the cumulative number of lives saved by WHO region in the chart below.
Vaccination Has Been a Massive Driver of Reductions in Infant MortalityChildren of all ages have benefited massively from the expansion of immunization programs. But it’s in infants that vaccines have had the most crucial impact.
Infant mortality rates have plummeted over the last 50 years.
Globally, they’ve fallen by over two-thirds, from around 10 percent in 1974 to less than 3 percent today.
The study’s researchers estimate that 40 percent of this decline is due to vaccines.
The other 60 percent of the decline has been driven by other factors, including improved nutrition, prenatal and neonatal care, access to clean water and sanitation, and other basic resources.
Coordinated vaccination programs have saved many lives50 years ago, very few children were vaccinated outside of Europe and North America. For example, fewer than 5 percent of infants received the vaccine against diphtheria, pertussis, and tetanus (DTP3).
In 1974, the World Health Assembly—the WHO’s decision-making body—formed the Essential Programme on Immunization, which aimed to vaccinate all children in the world against the main diseases for which vaccines exist, such as measles, tetanus, tuberculosis, and smallpox.
Soon after, vaccination rates increased steeply—expanding to over 60 percent of the world’s children (see the chart below). But by 2000, it was clear that progress was stalling, and many of the world’s poorest infants were still being left behind, especially in Africa and Asia.
Gavi, the Vaccine Alliance—a partnership between the Bill and Melinda Gates Foundation, the WHO, Unicef, and the World Bank—was formed to close the gaps and ensure vaccination programs were available for all.
Since then, vaccination rates have increased significantly. More than 80 percent of infants get all necessary doses of the DTP3 vaccine.
As the next chart shows, global vaccination against measles has increased from less than 20 percent in 2000 to over 70 percent today. Remember from earlier that vaccination against measles has saved the most lives.
84 percent of children are also vaccinated against tuberculosis, and 80 percent against polio and Hepatitis B.
Tens of millions of children are alive thanks to these investments in immunization programs across the world.
More Children Can Be Saved With Higher Vaccine RatesDespite this immense progress, there is still a lot more to be done.
More than a million people still die from tuberculosis every year. Hundreds of thousands from meningitis and whooping cough. Tens of thousands from measles, tetanus, and hepatitis B. Deaths caused by different vaccine-preventable diseases are shown in the chart below.
The world is also close to eradicating polio, which would make it the second human disease to be eradicated (the first was smallpox).
What’s more, scientists are now producing effective vaccines against other tragic diseases, such as malaria. There are now two recommended malaria vaccines that could potentially save hundreds of thousands of children every year.
The huge progress we’ve seen should cause us to push harder for universal vaccine coverage, not to pull back. We’re no longer helpless against the diseases our ancestors had no way to fight. We know how to stop children from dying, which makes it even more unacceptable that so many still do.
This will require increased investment, coordination from governments to provide universal immunization programs, and acceptance from the public. Most people in the world think that vaccinating children is important. However, in some countries, vaccine skepticism is much higher. Perhaps this pushback would be lower if we spent more time explaining the huge numbers of children saved by vaccines.
Tomorrow, newspapers could run the headline, “Almost 10,000 children were saved by essential vaccines yesterday.”5 They could have printed this headline daily for decades.
They won’t because this is not a groundbreaking new event. It’s progress that accumulates day after day but transforms the lives of hundreds of millions of kids and parents across the world.
This article was originally published on Our World in Data and has been republished here under a Creative Commons license. Read the original article.
Image Credit: Ian Talmacs / Unsplash
ARTIFICIAL INTELLIGENCEPocket-Sized AI Models Could Unlock a New Era of Computing
Will Knight | Wired“When ChatGPT was released in November 2023, it could only be accessed through the cloud because the model behind it was downright enormous. Today I am running a similarly capable AI program on a Macbook Air, and it isn’t even warm. The shrinkage shows how rapidly researchers are refining AI models to make them leaner and more efficient. It also shows how going to ever larger scales isn’t the only way to make machines significantly smarter.”
archive page
TECHGoogle Promised a Better Search Experience—Now It’s Telling Us to Put Glue on Our Pizza
Kylie Robison | The Verge“This is just one of many mistakes cropping up in the new feature that Google rolled out broadly this month. It also claims that former US President James Madison graduated from the University of Wisconsin not once but 21 times, that a dog has played in the NBA, NFL, and NHL, and that Batman is a cop. …Look, Google didn’t promise this would be perfect, and it even slaps a ‘Generative AI is experimental’ label at the bottom of the AI answers. But it’s clear these tools aren’t ready to accurately provide information at scale.”
BIOTECHGene Therapy Repairs Spinal Discs to Relieve Back Pain
Michael Irving | New Atlas“Assessed over 12 weeks, injured mice that received the gene therapy were found to have a host of improvements compared to injured mice given plain saline injections. The tissue in the discs was found to produce more proteins that strengthen the tissue, and help it hold water. That helped them plump back up and act more like cushions again, improving the spine’s range of motion, load bearing and flexibility. While you can’t exactly ask mice how much pain they’re feeling, behavioral tests suggested symptoms were reduced.”
AUTOMATIONOn Self-Driving, Waymo Is Playing Chess While Tesla Plays Checkers
Timothy B. Lee | Ars Technica“Many Tesla fans see [limitations like remote operators and avoiding freeways] as signs that Waymo is headed for a technological dead end. …But I predict that when Tesla begins its driverless transition, it will realize that safety requires a Waymo-style incremental rollout. So Tesla hasn’t found a different, better way to bring driverless technology to market. Waymo is just so far ahead that it’s dealing with challenges Tesla hasn’t even started thinking about. Waymo is playing chess while Tesla is still playing checkers.”
SCIENCEA Warp Drive Breakthrough Inches a Tiny Bit Closer to Star Trek
Paul Sutter | Wired“A team of physicists has discovered that it’s possible to build a real, actual, physical warp drive and not break any known rules of physics. One caveat: The vessel doing the warping can’t exceed the speed of light, so you’re not going to get anywhere interesting anytime soon. But this research still represents an important advance in our understanding of gravity.”
DIGITAL MEDIAMedia Companies Are Making a Huge Mistake With AI
Jessica Lessin | The Atlantic“For as long as I have reported on internet companies, I have watched news leaders try to bend their businesses to the will of Apple, Google, Meta, and more. Chasing tech’s distribution and cash, news firms strike deals to try to ride out the next digital wave. They make concessions to platforms that attempt to take all of the audience (and trust) that great journalism attracts, without ever having to do the complicated and expensive work of the journalism itself. And it never, ever works as planned.”
ARTIFICIAL INTELLIGENCEChatGPT, Explained
Sheena Vasani | The Verge“Some writers have declared that the debut of ChatGPT on November 30th, 2022, marked the beginning of a new chapter in history akin to the Enlightenment and the Industrial Revolution. Others have been more skeptical, wondering if this is just another overhyped tech, like blockchain or the metaverse. What history will call ChatGPT remains to be seen, but here’s one thing I do know for sure: nobody has shut up about it since. …That’s why we decided to throw together this explainer so we can cut through all the BS together. You ready? Let’s begin.”
TECHOpenAI Should Have Gone Way Beyond Scarlett Johansson
Ross Andersen | The Atlantic“Now that we can actually talk with a computer, we should be dreaming up wholly new ways to do it. Let’s hope that someone—inside or outside of OpenAI—starts giving us a sense of what those ways might be. The weirder, the better. They may not even be modeled after existing human relationships. They may take on entirely different forms.”
COMPUTINGNvidia’s Business Is Booming. Here’s What Could Slow It Down.
Asa Fitch | The Wall Street Journal“Nvidia is riding high after another quarter of blockbuster sales and earnings, even as threats are emerging that could weaken the company’s position at the center of the artificial-intelligence boom. Rivals and key customers are looking to produce chips that can close the gap with Nvidia’s products. Meanwhile, the AI market, which has proven tricky for some startups, is shifting in ways that could diminish the popularity of Nvidia’s chips.”
Image Credit: Jason Leung / Unsplash
Melanie Reid was 52 years old when she hopped onto her horse, fell, and broke her neck. The resulting injury paralyzed her body below the chest area. Fourteen years later and after extensive physical therapy, she has gradually regained some function in her right hand—“a lifeline,” she said in a press conference. But her left hand remained “useless.”
As a journalist, the injury was devasting as she couldn’t type. Even seemingly simple everyday routines—tying her hair up into a ponytail, using an ATM card, or even unwrapping candy—were a struggle.
With the help of a new device, she’s able to do all that after just two months of use. Called ARC-EX therapy, the device zaps residual neural connections around the site of spinal cord injuries. Combined with physical rehabilitation, the treatment restored some functionality in her left hand—even when the stimulation was turned off.
Reid was part of a 60-participant clinical trial that looked to use spinal cord stimulation to regain control of both hands. Similar treatments have shown promise in paraplegic patients, restoring the ability to walk in just a day. But those required surgery to place electrodes on the spinal cord.
ARC-EX therapy, by contrast, delivers two different types of electrical pulses through the skin—no surgery required. Developed by Grégoire Courtine and colleagues at the Swiss Federal Institute of Technology, the device improved hand strength, pinch, and other movements in 72 percent of participants.
Because the device is non-invasive, it’s a simple addition to physical rehabilitation programs—a sort of pilates for the fingers, explained the team. The trial only included two months of stimulation, and extending the timeline could potentially further improve results.
The stimulation only helps with finger and hand dexterity, not walking. But to Reid, that’s what matters. “Everyone thinks that [with] spinal injury all you want to do is to be able to walk again,” she said. But “what matters most is working hands…[and] the gains can be life-changing.”
Broken BridgesAll participants in the trial had a fracture in their spinal cord, roughly at the level of the neck.
With torn nerves, the brain can no longer command the body. Like a broken phone line, when you think “move your hand,” the signal gets lost at the break. Scientists have long tried to bridge this communication “gap” with electrodes to control muscle movement, essentially replacing broken biological wires with artificial ones.
Spinal cord stimulation is one solution. In 2018, a man walked across an entire football field years after he was paralyzed thanks to zaps to his spinal cord. With just a day’s stimulation, people with complete paralysis have been able to stroll around a busy downtown and go kayaking.
The key to recovery, explained the team, is to target living nerves. In a majority of cases, even patients deemed to have a “complete” spinal cord injury still have nerves left, and stimulating them can trigger regrowth and connections.
Picture the spinal cord as a tree, with branches reaching towards the skin. Now, imagine if the “trunk” was partially severed. By zapping the skin around the injury site, it’s possible to transmit electrical signals to the injured spinal cord “trunk” and allow living neurons near the injury site to rebuild neural connections.
“In preclinical models, when applying stimulation, we immediately facilitated movement,” said Courtine. But “even more importantly,” we saw new neuron growth, as if the body was repairing the broken neural system, he added. With sufficient healing, patients could potentially be able to use their hands even without stimulation.
Based on these ideas, the team built the ARC-EX device.
The TrialThe clinical trial included 60 people with neck-level spinal cord injury. All participants underwent two months of physical therapy, followed by another two months of physical therapy combined with ARC-EX stimulation. The group was roughly 46 years old on average.
For physical rehab, each person practiced movements such as pinching, grasping, or moving the whole arm for an hour each day, up to five times a week. Although the therapy improved arm and hand function, progress plateaued for most patients.
The team then added ARC-EX stimulation, with the electrodes placed above and below the site of injury. During therapy, the team controlled the frequency and strength of the zaps, so it facilitated arm and hand movements without causing any unwanted muscle “jerking.”
The trial was “open label,” meaning both participants and researchers know they’re receiving stimulation. This can be risky business because of the placebo effect—when participants recover because they think they’re getting the actual treatment, rather than, say, sham stimulation. However, with ARC-EX it’s impossible to “blind” the stimulation. Turning the device on immediately causes strange sensations in the participants, with the electrical pulses feeling like “a sort of a buzz,” said Reid. Turning the device off also immediately alerted participants. In one example, Reid said she “was holding a jar with weights with the stimulation on,” but with the stimulation off—unbeknownst to her—she immediately dropped the jar.
In just eight weeks, 72 percent of participants met or exceeded goals for hand strength and dexterity—the ability to grab a mug or pinch a tweezer—as assessed with a battery of tests. Only one person experienced unintended effects—uncontrollable muscle spasms. However, because stimulation was off when they occurred, the team says it’s likely not related to ARC-EX.
The therapy didn’t just improve hand function. The paralyzed participants also felt less pain, had fewer struggles with breathing, and slept better.
Boosted TherapyThe stimulation won’t work for the roughly 10 percent of people with spinal cord injuries that completely severed all neural connections. But for those who still have residual nerve endings, it makes restoring finger movements easier.
Sherown Campbell, another participant in the trial, said ARC-EX is a life-changer. A self-proclaimed “tech geek,” he constantly works on his computer. “Focusing on hand function was the biggest thing I was hoping to improve,” he said.
With ARC-EX therapy, his typing speed increased from 25 words per minute to 33 words per minute, “about a 30 percent increase, which is really significant for me,” he said. But more importantly, his quality of life improved. He can cook and write again—two seemingly simple everyday activities that give him joy but were robbed by his accident.
With the success of this trial, the device may hit the market soon. Because it’s non-invasive, it can easily be integrated into existing physical therapy sessions. The team is already looking for approval in the United States, with discussions with the European Union soon to follow. For now, the price is unknown, although Courtine said the aim is to make it widely accessible.
To Reid, being able to use her left hand is an enormous change to her life. It’s “extraordinary,” she said. “It makes you hold your head up and look at the world differently.”
Image Credit: ONWARD Medical N.V.
There are three ways to look for evidence of alien technological civilizations. One is to look out for deliberate attempts by them to communicate their existence, for example, through radio broadcasts. Another is to look for evidence of them visiting the solar system. And a third option is to look for signs of large-scale engineering projects in space.
A team of astronomers have taken the third approach by searching through recent astronomical survey data to identify seven candidates for alien megastructures, known as Dyson spheres, “deserving of further analysis.”
This is a detailed study looking for “oddballs” among stars—objects that might be alien megastructures. However, the authors are careful not to make any overblown claims. The seven objects, all located within 1,000 light-years of Earth, are “M-dwarfs”—a class of stars that are smaller and less bright than the sun.
Dyson spheres were first proposed by the physicist Freeman Dyson in 1960 as a way for an advanced civilization to harness a star’s power. Consisting of floating power collectors, factories, and habitats, they’d take up more and more space until they eventually surrounded almost the entire star like a sphere.
What Dyson realized is that these megastructures would have an observable signature. Dyson’s signature (which the team searched for in the recent study) is a significant excess of infrared radiation. That’s because megastructures would absorb visible light given off by the star, but they wouldn’t be able to harness it all. Instead, they’d have to “dump” excess energy as infrared light with a much longer wavelength.
Unfortunately, such light can also be a signature of a lot of other things, such as a disc of gas and dust or discs of comets and other debris. But the seven promising candidates aren’t obviously due to a disc, as they weren’t good fits to disc models.
It is worth noting there is another signature of a Dyson sphere: that visible light from the star dips as the megastructure passes in front of it. Such a signature has been found before. There was a lot of excitement about Tabby’s Star, or KIC 8462852, which showed many really unusual dips in its light that could be due to an alien megastructure.
Tabby’s Star in infrared (left) and ultraviolet (right). Image Credit: Infrared: IPAC/NASA / Ultraviolet: STScI /NASA via Wikimedia CommonsIt almost certainly isn’t an alien megastructure. A variety of natural explanations have been proposed, such as clouds of comets passing through a dust cloud. But it is an odd observation. An obvious follow up on the seven candidates would be to look for this signature as well.
The Case Against Dyson SpheresDyson spheres may well not even exist, however. I think they are unlikely to be there. That’s not to say they couldn’t exist, rather that any civilization capable of building them would probably not need to (unless it was some mega art project).
Dyson’s reasoning for considering such megastructures assumed that advanced civilizations would have vast power requirements. Around the same time, astronomer Nikolai Kardashev proposed a scale on which to rate the advancement of civilizations, which was based almost entirely on their power consumption.
In the 1960s, this sort of made sense. Looking back over history, humanity had just kept exponentially increasing its power use as technology advanced and the number of people increased, so they just extrapolated this ever-expanding need into the future.
However, our global energy use has started to grow much more slowly over the past 50 years, and especially over the last decade. What’s more, Dyson and Kardashev never specified what these vast levels of power would be used for, they just (fairly reasonably) assumed they’d be needed to do whatever it is that advanced alien civilizations do.
But as we now look ahead to future technologies, we see efficiency, miniaturization, and nanotechnologies promise vastly lower power use (the performance per watt of pretty much all technologies is constantly improving).
A quick calculation reveals that, if we wanted to collect 10 percent of the sun’s energy at the distance the Earth is from the sun, we’d need a surface area equal to 1 billion Earths. And if we had a super-advanced technology that could make the megastructure only 10 kilometers thick, that’d mean we’d need about a million Earths worth of material to build them from.
A significant problem is that our solar system only contains about 100 Earths worth of solid material, so our advanced alien civilization would need to dismantle all the planets in 10,000 planetary systems and transport it to the star to build their Dyson sphere. To do it with the material available in a single system, each part of the megastructure could only be one meter thick.
This is assuming they use all the elements available in a planetary system. If they needed, say, lots of carbon to make their structures, then we’re looking at dismantling millions of planetary systems to get hold of it. Now, I’m not saying a super-advanced alien civilization couldn’t do this, but it is one hell of a job.
I’d also strongly suspect that by the time a civilization got to the point of having the ability to build a Dyson sphere, they’d have a better way of getting the power than using a star, if they really needed it (I have no idea how, but they are a super-advanced civilization).
Maybe I’m wrong, but it can’t hurt to look.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Kevin Gill / Flickr
If you’ve ever vented to ChatGPT about troubles in life, the responses can sound empathetic. The chatbot delivers affirming support, and—when prompted—even gives advice like a best friend.
Unlike older chatbots, the seemingly “empathic” nature of the latest AI models has already galvanized the psychotherapy community, with many wondering if they can assist therapy.
The ability to infer other people’s mental states is a core aspect of everyday interaction. Called “theory of mind,” it lets us guess what’s going on in someone else’s mind, often by interpreting speech. Are they being sarcastic? Are they lying? Are they implying something that’s not overtly said?
“People care about what other people think and expend a lot of effort thinking about what is going on in other minds,” wrote Dr. Cristina Becchio and colleagues at the University Medical Center Hanburg-Eppendorf in a new study in Nature Human Behavior.”
In the study, the scientists asked if ChatGPT and other similar chatbots—which are based on machine learning algorithms called large language models—can also guess other people’s mindsets. Using a series of psychology tests tailored for certain aspects of theory of mind, they pitted two families of large language models, including OpenAI’s GPT series and Meta’s LLaMA 2, against over 1,900 human participants.
GPT-4, the algorithm behind ChatGPT, performed at, or even above, human levels in some tasks, such as identifying irony. Meanwhile, LLaMA 2 beat both humans and GPT at detecting faux pas—when someone says something they’re not meant to say but don’t realize it.
To be clear, the results don’t confirm LLMs have theory of mind. Rather, they show these algorithms can mimic certain aspects of this core concept that “defines us as humans,” wrote the authors.
What’s Not SaidBy roughly four years old, children already know that people don’t always think alike. We have different beliefs, intentions, and needs. By placing themselves into other people’s shoes, kids can begin to understand other perspectives and gain empathy.
First introduced in 1978, theory of mind is a lubricant for social interactions. For example, if you’re standing near a closed window in a stuffy room, and someone nearby says, “It’s a bit hot in here,” you have to think about their perspective to intuit they’re politely asking you to open the window.
When the ability breaks down—for example, in autism—it becomes difficult to grasp other people’s emotions, desires, intentions, and to pick up deception. And we’ve all experienced when texts or emails lead to misunderstandings when a recipient misinterprets the sender’s meaning.
So, what about the AI models behind chatbots?
Man Versus MachineBack in 2018, Dr. Alan Winfield, a professor in the ethics of robotics at the University of West England, championed the idea that theory of mind could let AI “understand” people and other robots’ intentions. At the time, he proposed giving an algorithm a programmed internal model of itself, with common sense about social interactions built in rather than learned.
Large language models take a completely different approach, ingesting massive datasets to generate human-like responses that feel empathetic. But do they exhibit signs of theory of mind?
Over the years, psychologists have developed a battery of tests to study how we gain the ability to model another’s mindset. The new study pitted two versions of OpenAI’s GPT models (GPT-4 and GPT-3.5) and Meta’s LLaMA-2-Chat against 1,907 healthy human participants. Based solely on text descriptions of social scenarios and using comprehensive tests spanning different theories of theory of mind abilities, they had to gauge the fictional person’s “mindset.”
Each test was already well-established for measuring theory of mind in humans in psychology.
The first, called “false belief,” is often used to test toddlers as they gain a sense of self and recognition of others. As an example, you listen to a story: Lucy and Mia are in the kitchen with a carton of orange juice in the cupboard. When Lucy leaves, Mia puts the juice in the fridge. Where will Lucy look for the juice when she comes back?
Both humans and AI guessed nearly perfectly that the person who’d left the room when the juice was moved would look for it where they last remembered seeing it. But slight changes tripped the AI up. When changing the scenario—for example, the juice was transported between two transparent containers—GPT models struggled to guess the answer. (Though, for the record, humans weren’t perfect on this either in the study.)
A more advanced test is “strange stories,” which relies on multiple levels of reasoning to test for advanced mental capabilities, such as misdirection, manipulation, and lying. For example, both human volunteers and AI models were told the story of Simon, who often lies. His brother Jim knows this and one day found his Ping-Pong paddle missing. He confronts Simon and asks if it’s under the cupboard or his bed. Simon says it’s under the bed. The test asks: Why would Jim look in the cupboard instead?
Out of all AI models, GPT-4 had the most success, reasoning that “the big liar” must be lying, and so it’s better to choose the cupboard. Its performance even trumped human volunteers.
Then came the “faux pas” study. In prior research, GPT models struggled to decipher these social situations. During testing, one example depicted a person shopping for new curtains, and while putting them up, a friend casually said, “Oh, those curtains are horrible, I hope you’re going to get some new ones.” Both humans and AI models were presented with multiple similar cringe-worthy scenarios and asked if the witnessed response was appropriate. “The correct answer is always no,” wrote the team.
GPT-4 correctly identified that the comment could be hurtful, but when asked whether the friend knew about the context—that the curtains were new—it struggled with a correct answer. This could be because the AI couldn’t infer the mental state of the person, and that recognizing a faux pas in this test relies on context and social norms not directly explained in the prompt, explained the authors. In contrast, LLaMA-2-Chat outperformed humans, achieving nearly 100 percent accuracy except for one run. It’s unclear why it has such as an advantage.
Under the BridgeMuch of communication isn’t what’s said, but what’s implied.
Irony is maybe one of the hardest concepts to translate between languages. When tested with an adapted psychological test for autism, GPT-4 surprisingly outperformed human participants in recognizing ironic statements—of course, through text only, without the usual accompanying eye-roll.
The AI also outperformed humans on a hinting task—basically, understanding an implied message. Derived from a test for assessing schizophrenia, it measures reasoning that relies on both memory and cognitive ability to weave and assess a coherent narrative. Both participants and AI models were given 10 written short skits, each depicting an everyday social interaction. The stories ended with a hint of how best to respond with open-ended answers. Over 10 stories, GPT-4 won against humans.
For the authors, the results don’t mean LLMs already have theory of mind. Each AI struggled with some aspects. Rather, they think the work highlights the importance of using multiple psychology and neuroscience tests—rather than relying on any one—to probe the opaque inner workings of machine minds. Psychology tools could help us better understand how LLMs “think”—and in turn, help us build safer, more accurate, and more trustworthy AI.
There’s some promise that “artificial theory of mind may not be too distant an idea,” wrote the authors.
Image Credit: Abishek / Unsplash
The origin of consciousness has teased the minds of philosophers and scientists for centuries. In the last decade, neuroscientists have begun to piece together its neural underpinnings—that is, how the brain, through its intricate connections, transforms electrical signaling between neurons into consciousness.
Yet the field is fragmented, an international team of neuroscientists recently wrote in a new paper in Neuron. Many theories of consciousness contradict each other, with different ideas about where and how consciousness emerges in the brain.
Some theories are even duking it out in a mano-a-mano test by imaging the brains of volunteers as they perform different tasks in clinical test centers across the globe.
But unlocking the neural basis of consciousness doesn’t have to be confrontational. Rather, theories can be integrated, wrote the authors, who were part of the Human Brain Project—a massive European endeavor to map and understand the brain—and specialize in decoding brain signals related to consciousness.
Not all authors agree on the specific brain mechanisms that allow us to perceive the outer world and construct an inner world of “self.” But by collaborating, they merged their ideas, showing that different theories aren’t necessarily mutually incompatible—in fact, they could be consolidated into a general framework of consciousness and even inspire new ideas that help unravel one of the brain’s greatest mysteries.
If successful, the joint mission could extend beyond our own noggins. Brain organoids, or “mini-brains,” that roughly mimic early human development are becoming increasingly sophisticated, spurring ethical concerns about their potential for developing self-awareness (to be clear, there aren’t any signs). Meanwhile, similar questions have been raised about AI. A general theory of consciousness, based on the human mind, could potentially help us evaluate these artificial constructs.
“Is it realistic to reconcile theories, or even aspire to a unified theory of consciousness?” the authors asked. “We take the standpoint that the existence of multiple theories is a sign of healthiness in this nascent field…such that multiple theories can simultaneously contribute to our understanding.”
Lost in TranslationI’m conscious. You are too. We see, smell, hear, and feel. We have an internal world that tells us what we’re experiencing. But the lines get blurry for people in different stages of coma or for those locked-in—they can still perceive their surroundings but can’t physically respond. We lose consciousness in sleep every night and during anesthesia. Yet, somehow, we regain consciousness. How?
With extensive imaging of the brain, neuroscientists today agree that consciousness emerges from the brain’s wiring and activity. But multiple theories argue about how electrical signals in the brain produce rich and intimate experiences of our lives.
Part of the problem, wrote the authors, is that there isn’t a clear definition of “consciousness.” In this paper, they separated the term into two experiences: one outer, one inner. The outer experience, called phenomenal consciousness, is when we immediately realize what we’re experiencing—for example, seeing a total solar eclipse or the northern lights.
The inner experience is a bit like a “gut feeling” in that it helps to form expectations and types of memory, so that tapping into it lets us plan behaviors and actions.
Both are aspects of consciousnesses, but the difference is hardly delineated in previous work. It makes comparing theories difficult, wrote the authors, but that’s what they set out to do.
Meet the ContendersUsing their “two experience” framework, they examined five prominent consciousness theories.
The first, the global neuronal workspace theory, pictures the brain as a city of sorts. Each local brain region “hub” dynamically interacts with a “global workspace,” which integrates and broadcasts information to other hubs for further processing—allowing information to reach the consciousness level. In other words, we only perceive something when all pieces of sensory information—sight, hearing, touch, taste—are woven into a temporary neural sketchpad. According to this theory, the seat of consciousness is in the frontal parts of the brain.
The second, integrated information theory, takes a more globalist view. The idea is that consciousness stems from a series of cause-effect reactions from the brain’s networks. With the right neural architecture, connections, and network complexity, consciousness naturally emerges. The theory suggests the back of the brain sparks consciousness.
Then there’s dendritic integration theory, the coolest new kid in town. Unlike previous ideas, this theory waved the front or back of the brain goodbye and instead zoomed in on single neurons in the cortex, the outermost part of the brain and a hub for higher cognitive functions such as reasoning and planning.
The cortex has extensive connections to other parts of the brain—for example, those that encode memories and emotions. One type of neuron, deep inside the cortex, especially stands out. Physically, these neurons resemble trees with extensive “roots” and “branches.” The roots connect to other parts of the brain, whereas the upper branches help calculate errors in the neuron’s computing. In turn, these upper branches generate an error signal that corrects mistakes through multiple rounds of learning.
The two compartments, while physically connected, go about their own business—turning a single neuron into multiple computers. Here’s the crux: There’s a theoretical “gate” between the upper and lower neural “offices” for each neuron. During consciousness, the gate opens, allowing information to flow between the cortex and other brain regions. In dreamless sleep and other unconscious states, the gate closes.
Like a light switch, this theory suggests that consciousness is supported by flicking individual neuron gates on or off on a grand scale.
The last two theories propose that recurrent processing in the brain—that is, it learns from previous experiences—is essential for consciousness. Instead of “experiencing” the world, the brain builds an internal simulation that constantly predicts the “here and now” to control what we perceive.
A Unified Theory?All the theories have extensive experiments to back up their claims. So, who’s right? To the authors, the key is to consider consciousness not as a singular concept, but as a “ladder” of sorts. The brain functions at multiple levels: cells, local networks, brain regions, and finally, the whole brain.
When examining theories of consciousness, it also makes sense to delineate between different levels. For example, the dendritic integration theory—which considers neurons and their connections—is on the level of single cells and how they contribute to consciousness. It makes the theory “neutral,” in that it can easily fit into ideas at a larger scale—those that mostly rely on neural network connections or across larger brain regions.
Although it’s seemingly difficult to reconcile various ideas about consciousness, two principles tie them together, wrote the team. One is that consciousness requires feedback, within local neural circuits and throughout the brain. The other is integration, in that any feedback signals need to be readily incorporated back into neural circuits, so they can change their outputs. Finally, all authors agree that local, short connections are vital but not enough. Long distance connections from the cortex to deeper brain areas are required for consciousness.
So, is an integrated theory of consciousness possible? The authors are optimistic. By defining multiple aspects of consciousness—immediate responses versus internal thoughts—it’ll be clearer how to explore and compare results from different experiments. For now, the global neuronal workspace theory mostly focuses on the “inner experience” that leads to consciousness, whereas others try to tackle the “outer experience”—what we immediately experience.
For the theories to merge, the latter groups will have to explain how consciousness is used for attention and planning, which are hallmarks for immediate responses. But fundamentally, wrote the authors, they are all based on different aspects of neuronal connections near and far. With more empirical experiments, and as increasingly more sophisticated brain atlases come online, they’ll move the field forward.
Hopefully, the authors write, “an integrated theory of consciousness…may come within reach within the next years or decades.”
Image Credit: SIMON LEE / Unsplash
ARTIFICIAL INTELLIGENCE*It’s Time to Believe the AI Hype
Steven Levy | Wired*“There’s universal agreement in the tech world that AI is the biggest thing since the internet, and maybe bigger. …Skeptics might try to claim that this is an industry-wide delusion, fueled by the prospect of massive profits. But the demos aren’t lying. We will eventually become acclimated to the AI marvels unveiled this week. The smartphone once seemed exotic; now it’s an appendage no less critical to our daily life than an arm or a leg. At a certain point AI’s feats, too, may not seem magical any more.”
archive page
COMPUTINGHow to Put a Datacenter in a Shoebox
Anna Herr and Quentin Herr | IEEE Spectrum“At Imec, we have spent the past two years developing superconducting processing units that can be manufactured using standard CMOS tools. A processor based on this work would be one hundred times as energy efficient as the most efficient chips today, and it would lead to a computer that fits a data-center’s worth of computing resources into a system the size of a shoebox.”
BIOTECHIndieBio’s SF Incubator Lineup Is Making Some Wild Biotech Promises
Devin Coldewey | TechCrunch“We took special note of a few, which were making some major, bordering on ludicrous, claims that could pay off in a big way. Biotech has been creeping out in recent years to touch adjacent industries, as companies find how much they rely on outdated processes or even organisms to get things done. So it may not surprise you that there’s a microbiome company in the latest batch—but you might be surprised when you hear it’s the microbiome of copper ore.”
TECHIt’s the End of Google Search as We Know It
Lauren Goode | Wired“It’s as though Google took the index cards for the screenplay it’s been writing for the past 25 years and tossed them into the air to see where the cards might fall. Also: The screenplay was written by AI. These changes to Google Search have been long in the making. Last year the company carved out a section of its Search Labs, which lets users try experimental new features, for something called Search Generative Experience. The big question since has been whether, or when, those features would become a permanent part of Google Search. The answer is, well, now.”
AUTOMATIONWaymo Says Its Robotaxis Are Now Making 50,000 Paid Trips Every Week
Mariella Moon | Engadget“If you’ve been seeing more Waymo robotaxis recently in Phoenix, San Francisco, and Los Angeles, that’s because more and more people are hailing one for a ride. The Alphabet-owned company has announced on Twitter/X that it’s now serving more than 50,000 paid trips every week across three cities. Waymo One operates 24/7 in parts of those cities. If the company is getting 50,000 rides a week, that means it receives an average of 300 bookings every hour or five bookings every minute.”
CULTURETechnology Is Probably Changing Us for the Worse—or So We Always Think
Timothy Maher | MIT Technology Review“We’ve always greeted new technologies with a mixture of fascination and fear, says Margaret O’Mara, a historian at the University of Washington who focuses on the intersection of technology and American politics. ‘People think: “Wow, this is going to change everything affirmatively, positively,”‘ she says. ‘And at the same time: ‘It’s scary—this is going to corrupt us or change us in some negative way.”‘ And then something interesting happens: ‘We get used to it,’ she says. ‘The novelty wears off and the new thing becomes a habit.'”
TECHThis Is the Next Smartphone Evolution
Matteo Wong | The Atlantic“Earlier [this week], OpenAI announced its newest product: GPT-4o, a faster, cheaper, more powerful version of its most advanced large language model, and one that the company has deliberately positioned as the next step in ‘natural human-computer interaction.’ …Watching the presentation, I felt that I was witnessing the murder of Siri, along with that entire generation of smartphone voice assistants, at the hands of a company most people had not heard of just two years ago.”
SPACEIn the Race for Space Metals, Companies Hope to Cash In
Sarah Scoles | Undark“Previous companies have rocketed toward similar goals before but went bust about a half decade ago. In the years since that first cohort left the stage, though, ‘the field has exploded in interest,’ said Angel Abbud-Madrid, director of the Center for Space Resources at the Colorado School of Mines. …The economic picture has improved with the cost of rocket launches decreasing, as has the regulatory environment, with countries creating laws specifically allowing space mining. But only time will tell if this decade’s prospectors will cash in where others have drilled into the red or be buried by their business plans.”
FUTUREWhat I Got Wrong in a Decade of Predicting the Future of Tech
Christopher Mims | The Wall Street Journal“Anniversaries are typically a time for people to get misty-eyed and recount their successes. But after almost 500 articles in The Wall Street Journal, one thing I’ve learned from covering the tech industry is that failures are far more instructive. Especially when they’re the kind of errors made by many people. Here’s what I’ve learned from a decade of embarrassing myself in public—and having the privilege of getting an earful about it from readers.”
FUTURE OF FOODLab-Grown Meat Is on Shelves Now. But There’s a Catch
Matt Reynolds | Wired“Now cultivated meat is available in one store in Singapore. There is a catch, however: The chicken on sale at Huber’s Butchery contains just 3 percent animal cells. The rest will be made of plant protein—the same kind of ingredients you’d find in plant-based meats that are already on supermarket shelves worldwide. This might feel like a bit of a bait and switch. Didn’t cultivated meat firms promise us real chicken? And now we’re getting plant-based products with a sprinkling of animal cells? That criticism wouldn’t be entirely fair, though.”
Image Credit: Pawel Czerwinski / Unsplash
Some of the hardest sectors to decarbonize are industries that require high temperatures like steel smelting and cement production. A new approach uses a synthetic quartz solar trap to generate temperatures of over 1,000 degrees Celsius (1,832 degrees Fahrenheit)—hot enough for a host of carbon-intensive industries.
While most of the focus on the climate fight has been on cleaning up the electric grid and transportation, a surprisingly large amount of fossil fuel usage goes into industrial heat. As much as 25 percent of global energy consumption goes towards manufacturing glass, steel, and cement.
Electrifying these processes is challenging because it’s difficult to reach the high temperatures required. Solar receivers, which use thousands of sun-tracking mirrors to concentrate energy from the sun, have shown promise as they can hit temperatures of 3,000 C. But they’re very inefficient when processes require temperatures over 1,000 C because much of the energy is radiated back out.
To get around this, researchers from ETH Zurich in Switzerland showed that adding semi-transparent quartz to a solar receiver could trap solar energy at temperatures as high as 1,050 C. That’s hot enough to replace fossil fuels in a range of highly polluting industries, the researchers say.
“Previous research has only managed to demonstrate the thermal-trap effect up to 170 C,” lead researcher Emiliano Casati said in a press release. “Our research showed that solar thermal trapping works not just at low temperatures, but well above 1,000 C. This is crucial to show its potential for real-world industrial applications.”
The researchers used a silicon carbide disk to absorb solar energy but attached a roughly one-foot-long quartz rod to it. Because quartz is semi-transparent, light is able pass through it, but it also readily absorbs heat and prevents it from being radiated back out.
That meant that when the researchers subjected the quartz rod to simulated sunlight equivalent to 136 suns, the solar energy readily passed through to the silicon plate and was then trapped there. This allowed the plate to heat up to 1,050 C, compared to just 600 C at the other end of the rod.
Simulations of the device found that the quartz’s thermal trapping capabilities could significantly boost the efficiency of solar receivers. Adding a quartz rod to a state-of-the-art receiver could boost efficiency from 40 percent to 70 percent when attempting to hit temperatures of 1,200 C. That kind of efficiency gain could drastically reduce the size, and therefore cost, of solar heat installations.
While still just a proof of concept, the simplicity of the approach means it would probably not be too difficult to apply to existing receiver technology. Companies like Heliogen, which is backed by Bill Gates, has already developed solar furnace technology designed to generate the high temperatures required in a wide range of industries.
Casati says the promise is clear, but work remains to be done to prove its commercial feasibility.
“Solar energy is readily available, and the technology is already here,” he says. “To really motivate industry adoption, we need to demonstrate the economic viability and advantages of this technology at scale.”
But the prospect of replacing such a big chunk of our fossil fuel usage with solar power should be motivation enough to bring this technology to fruition.
Image Credit: A new solar trap built by a team of ETH Zurich scientists reaches 1050 C (Device/Casati et al.)
A quantum internet would essentially be unhackable. In the future, sensitive information—financial or national security data, for instance, as opposed to memes and cat pictures—would travel through such a network in parallel to a more traditional internet.
Of course, building and scaling systems for quantum communications is no easy task. Scientists have been steadily chipping away at the problem for years. A Harvard team recently took another noteworthy step in the right direction. In a paper published this week in Nature, the team says they’ve sent entangled photons between two quantum memory nodes 22 miles (35 kilometers) apart on existing fiber optic infrastructure under the busy streets of Boston.
“Showing that quantum network nodes can be entangled in the real-world environment of a very busy urban area is an important step toward practical networking between quantum computers,” Mikhail Lukin, who led the project and is a physics professor at Harvard, said in a press release.
The team leased optical fiber under the Boston streets, connecting the two memory nodes located at Harvard by way of a 22-mile (35-kilometer) loop of cable. Image Credit: Can Knaut via OpenStreetMapOne way a quantum network can transmit information is by using entanglement, a quantum property where two particles, likely photons in this case, are linked so a change in the state of one tells us about the state of the other. If the sender and receiver of information each have one of a pair of entangled photons, they can securely transmit data using them. This means quantum communications will rely on generating enormous numbers of entangled photons and reliably sending them to far-off destinations.
Scientists have sent entangled particles long distances over fiber optic cables before, but to make a quantum internet work, particles will need to travel hundreds or thousands of miles. Because cables tend to absorb photons over such distances, the information will be lost—unless it can be periodically refreshed.
Enter quantum repeaters.
You can think of a repeater as a kind of internet gas station. Information passing through long stretches of fiber optic cables naturally degrades. A repeater refreshes that information at regular intervals, strengthening the signal and maintaining its fidelity. A quantum repeater is the same thing, only it also preserves entanglement.
That scientists have yet to build a quantum repeater is one reason we’re still a ways off from a working quantum internet at scale. Which is where the Harvard study comes in.
The team of researchers from Harvard and Amazon Web Services (AWS) have been working on quantum memory nodes. Each node houses a piece of diamond with an atom-sized hole, or silicon-vacancy center, containing two qubits: one for storage, one for communication. The nodes are basically small quantum computers, operating at near absolute zero, that can receive, record, and transmit quantum information. The Boston experiment, according to the team, is the longest distance anyone has sent information between such devices and a big step towards a quantum repeater.
“Our experiment really put us in a position where we’re really close to working on a quantum repeater demonstration,” Can Knaut, a Harvard graduate student in Lukin’s lab, told New Scientist.
Next steps include expanding the system to include multiple nodes.
Along those lines, a separate group in China, using a different technique for quantum memory involving clouds of rubidium atoms, recently said they’d linked three nodes 6 miles (10 kilometers) apart. The same group, led by Xiao-Hui Bao at the University of Science and Technology of China, had previously entangled memory nodes 13.6 miles (22 kilometers) apart.
It’ll take a lot more work to make the technology practical. Researchers need to increase the rate at which their machines entangle photons, for example. But as each new piece falls into place, the prospect of unhackable communications gets a bit closer.
Image Credit: Visax / Unsplash
Many people understand the concept of bias at some intuitive level. In society, and in artificial intelligence systems, racial and gender biases are well documented.
If society could somehow remove bias, would all problems go away? The late Nobel laureate Daniel Kahneman, who was a key figure in the field of behavioral economics, argued in his last book that bias is just one side of the coin. Errors in judgments can be attributed to two sources: bias and noise.
Bias and noise both play important roles in fields such as law, medicine, and financial forecasting, where human judgments are central. In our work as computer and information scientists, my colleagues and I have found that noise also plays a role in AI.
Statistical NoiseNoise in this context means variation in how people make judgments of the same problem or situation. The problem of noise is more pervasive than initially meets the eye. A seminal work, dating back all the way to the Great Depression, has found that different judges gave different sentences for similar cases.
Worryingly, sentencing in court cases can depend on things such as the temperature and whether the local football team won. Such factors, at least in part, contribute to the perception that the justice system is not just biased but also arbitrary at times.
Other examples: Insurance adjusters might give different estimates for similar claims, reflecting noise in their judgments. Noise is likely present in all manner of contests, ranging from wine tastings to local beauty pageants to college admissions.
Noise in the DataOn the surface, it doesn’t seem likely that noise could affect the performance of AI systems. After all, machines aren’t affected by weather or football teams, so why would they make judgments that vary with circumstance? On the other hand, researchers know that bias affects AI, because it is reflected in the data that the AI is trained on.
For the new spate of AI models like ChatGPT, the gold standard is human performance on general intelligence problems such as common sense. ChatGPT and its peers are measured against human-labeled commonsense datasets.
Put simply, researchers and developers can ask the machine a commonsense question and compare it with human answers: “If I place a heavy rock on a paper table, will it collapse? Yes or No.” If there is high agreement between the two—in the best case, perfect agreement—the machine is approaching human-level common sense, according to the test.
So where would noise come in? The commonsense question above seems simple, and most humans would likely agree on its answer, but there are many questions where there is more disagreement or uncertainty: “Is the following sentence plausible or implausible? My dog plays volleyball.” In other words, there is potential for noise. It is not surprising that interesting commonsense questions would have some noise.
But the issue is that most AI tests don’t account for this noise in experiments. Intuitively, questions generating human answers that tend to agree with one another should be weighted higher than if the answers diverge—in other words, where there is noise. Researchers still don’t know whether or how to weigh AI’s answers in that situation, but a first step is acknowledging that the problem exists.
Tracking Down Noise in the MachineTheory aside, the question still remains whether all of the above is hypothetical or if in real tests of common sense there is noise. The best way to prove or disprove the presence of noise is to take an existing test, remove the answers and get multiple people to independently label them, meaning provide answers. By measuring disagreement among humans, researchers can know just how much noise is in the test.
The details behind measuring this disagreement are complex, involving significant statistics and math. Besides, who is to say how common sense should be defined? How do you know the human judges are motivated enough to think through the question? These issues lie at the intersection of good experimental design and statistics. Robustness is key: One result, test, or set of human labelers is unlikely to convince anyone. As a pragmatic matter, human labor is expensive. Perhaps for this reason, there haven’t been any studies of possible noise in AI tests.
To address this gap, my colleagues and I designed such a study and published our findings in Nature Scientific Reports, showing that even in the domain of common sense, noise is inevitable. Because the setting in which judgments are elicited can matter, we did two kinds of studies. One type of study involved paid workers from Amazon Mechanical Turk, while the other study involved a smaller-scale labeling exercise in two labs at the University of Southern California and the Rensselaer Polytechnic Institute.
You can think of the former as a more realistic online setting, mirroring how many AI tests are actually labeled before being released for training and evaluation. The latter is more of an extreme, guaranteeing high quality but at much smaller scales. The question we set out to answer was how inevitable is noise, and is it just a matter of quality control?
The results were sobering. In both settings, even on commonsense questions that might have been expected to elicit high—even universal—agreement, we found a nontrivial degree of noise. The noise was high enough that we inferred that between 4 percent and 10 percent of a system’s performance could be attributed to noise.
To emphasize what this means, suppose I built an AI system that achieved 85 percent on a test, and you built an AI system that achieved 91 percent. Your system would seem to be a lot better than mine. But if there is noise in the human labels that were used to score the answers, then we’re not sure anymore that the 6 percent improvement means much. For all we know, there may be no real improvement.
On AI leaderboards, where large language models like the one that powers ChatGPT are compared, performance differences between rival systems are far narrower, typically less than 1 percent. As we show in the paper, ordinary statistics do not really come to the rescue for disentangling the effects of noise from those of true performance improvements.
Noise AuditsWhat is the way forward? Returning to Kahneman’s book, he proposed the concept of a “noise audit” for quantifying and ultimately mitigating noise as much as possible. At the very least, AI researchers need to estimate what influence noise might be having.
Auditing AI systems for bias is somewhat commonplace, so we believe that the concept of a noise audit should naturally follow. We hope that this study, as well as others like it, leads to their adoption.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Michael Dziedzic / Unsplash
Scientists just published the most detailed map of a cubic millimeter of the human brain. Smaller than a grain of rice, the mapped section of brain includes over 57,000 cells, 230 millimeters of blood vessels, and 150 million synapses.
The project, a collaboration between Harvard and Google, is looking to accelerate connectomics—the study of how neurons are wired together—over a much larger scale.
Our brains are like a jungle.
Neuron branches crisscross regions, forming networks that process perception, memories, and even consciousness. Blood vessels tightly wrap around these branches to provide nutrients and energy. Other brain cell types form intricate connections with neurons, support the brain’s immune function, and fine-tune neural network connections.
In biology, structure determines function. Like tracing wires of a computer, mapping components of the brain and their connections can improve our understanding of how the brain works—and when and why it goes wrong. A brain map that charts the jungle inside our heads could help us tackle some of the most perplexing neurological disorders, such as Alzheimer’s disease, and decipher the origins of emotions, thoughts, and behaviors.
Aided by machine learning tools from Google Research, the Harvard team traced neurons, blood vessels, and other brain cells at nanoscale levels. The images revealed previously unknown quirks in the human brain—including mysterious tangles in neuron wiring and neurons that connect through multiple “contacts” to other cells. Overall, the dataset incorporates a massive 1.4 petabytes of information—roughly the storage amount of a thousand high-end laptops—and is free to explore.
“It’s a little bit humbling,” Dr. Viren Jain, a neuroscientist at Google and study author, told Nature. “How are we ever going to really come to terms with all this complexity?” The database, first released as a preprint paper in 2021, has already garnered much enthusiasm in the scientific field.
“It’s probably the most computer-intensive work in all of neuroscience,” Dr. Michael Hawrylycz, a computational neuroscientist at the Allen Institute for Brain Science, who was not involved in the project, told MIT Technology Review.
Why So Complicated?Many types of brain maps exist. Some chart gene expression in brain cells; others map different cell types across the brain. But the goal is the same. They aim to help scientists understand how the brain works in health and disease.
The connectome details highways between brain regions that “talk” to each other. These connections, called synapses, number in the hundreds of trillions in human brains—on the scale of the number of stars in the universe.
Decades ago, the first whole-brain wiring map detailed all 302 neurons in the roundworm Caenorhabditis elegans. Because its genetics are largely known, the lowly worm delivered insights, such as how the brain and body communicate to increase healthy longevity. Next, scientists charted the fruit fly connectome and found the underpinnings of spatial navigation.
More recently, the MouseLight Project and MICrONS have been deciphering a small chunk of a mouse’s brain—the outermost area called the cortex. It’s hoped such work can help inform neuro-inspired AI algorithms with lower power requirements and higher efficacy.
But mice are not people. In the new study, scientists mapped a cubic millimeter of human brain tissue from the temporal cortex—a nexus that’s important for memory, emotions, and sensations. Although just one-millionth of a human brain, the effort reconstructed connections in 3D at nanoscale resolution.
Slice It UpSourcing is a challenge when mapping the human brain. Brain tissues rapidly deteriorate after trauma or death, which changes their wiring and chemistry. Brain organoids—”mini-brains” grown in test tubes—somewhat resemble the brain’s architecture, but they can’t replicate the real thing.
Here, the team took a tiny bit of brain tissue from a 45-year-old woman with epilepsy during surgery—the last resort for those who suffer severe seizures and don’t respond to medication.
Using a machine like a deli-meat slicer armed with a diamond knife, the Harvard team, led by connectome expert Dr. Jeff Lichtman, meticulously sliced the sample into 5,019 cross sections. Each was roughly 30 nanometers thick—a fraction of the width of a human hair. They imaged the slices with an electron microscope, capturing nanoscale cellular details, including the “factories” inside cells that produce energy, eliminate waste, or transport molecules.
Piecing these 2D images into a 3D reconstruction is a total headache. A decade ago, scientists had to do it by hand. Jain’s team at Google developed an AI to automate the job. The AI was able to track fragments of whole components—say, a part of a neuron (its body or branches)—and stick them back together throughout the images.
In total, the team pieced together thousands of neurons and over a hundred million synaptic connections. Other brain components included blood vessels and myelin—a protective molecular “sheath” covering neurons. Like electrical insulation, when myelin deteriorates, it causes multiple brain disorders.
“I remember this moment, going into the map and looking at one individual synapse from this woman’s brain, and then zooming out into these other millions of pixels,” Jain told Nature. “It felt sort of spiritual.”
A Whole New WorldEven a cursory look at the data led to surprising insights into the brain’s intricate neural wiring.
Cortical neurons have a forest-like structure for input and a single “cable” that delivers output signals. Called axons, these are dotted with thousands of synapses connecting to other cells.
Usually, a synapse grabs onto just one spot of a neighboring neuron. But the new map found a rare, strange group that connects with up to 50 points. “We’ve always had a theory that there would be super connections, if you will, amongst certain cells…But it’s something we’ve never had the resolution to prove,” Dr. Tim Mosca, who was not involved in the work, told Popular Science. These could be extra-potent connections that allow neural communications to go into “autopilot mode,” like when riding a bike or navigating familiar neighborhoods.
More strange structures included “axon whorls” that wrapped around themselves like tangled headphones. An axon’s main purpose is to reach out and connect with other neurons—so why do some fold into themselves? Do they serve a purpose, or are they just a hiccup in brain wiring? It’s a mystery. Another strange observation found pairs of neurons that perfectly mirrored each other. What this symmetry does for the brain is also unknown.
The bottom line: Our understanding of the brain’s connections and inner workings is still only scratching the surface. The new database is a breakthrough, but it’s not perfect. The results are from a single person with epilepsy, which can’t represent everyone. Some wiring changes, for example, may be due to the disorder. The team is planning a follow-up to separate epilepsy-related circuits from those that are more universal in people.
Meanwhile, they’ve opened the entire database for anyone to explore. And the team is also working with scientists to manually examine the results and eliminate potential AI-induced errors during reconstruction. So far, hundreds of cells have been “proofread” and validated by humans, but it’s just a fraction of the 50,000 neurons in the database.
The technology can also be used for other species, such as the zebrafish—another animal model often used in neuroscience research—and eventually the entire mouse brain.
Although this study only traced a tiny nugget of the human brain, the atlas is a stunning way to peek inside its seemingly chaotic wiring and make sense of things. “Further studies using this resource may bring valuable insights into the mysteries of the human brain,” wrote the team.
Image Credit: Google Research and Lichtman Lab
ARTIFICIAL INTELLIGENCE*OpenAI Could Unveil Its Google Search Competitor on Monday
Jess Weatherbed | The Verge*“OpenAI is reportedly gearing up to announce a search product powered by artificial intelligence on Monday that could threaten Google’s dominance. That target date, provided to Reuters by ‘two sources familiar with the matter,’ would time the announcement a day before Google kicks off its annual I/O conference, which is expected to focus on the search giant’s own AI model offerings like Gemini and Gemma.”
archive page
ROBOTICSDeepMind Is Experimenting With a Nearly Indestructible Robot Hand
Jeremy Hsu | New Scientist“This latest robotic hand developed by the UK-based Shadow Robot Company can go from fully open to closed within 500 milliseconds and perform a fingertip pinch with up to 10 newtons of force. It can also withstand repeated punishment such as pistons punching the fingers from multiple angles or a person smashing the device with a hammer.”
BIOTECHFirst Patient Begins Newly Approved Sickle Cell Gene Therapy
Gina Kolata | The New York Times“On Wednesday, Kendric Cromer, a 12-year-old boy from a suburb of Washington, became the first person in the world with sickle cell disease to begin a commercially approved gene therapy that may cure the condition. For the estimated 20,000 people with sickle cell in the United States who qualify for the treatment, the start of Kendric’s monthslong medical journey may offer hope. But it also signals the difficulties patients face as they seek a pair of new sickle cell treatments.”
SPACECommercial Space Stations Approach Launch Phase
Andrew Jones | IEEE Spectrum“A changing of the guard in space stations is on the horizon as private companies work towards providing new opportunities for science, commerce, and tourism in outer space. …The challenge [new space stations like Blue Origin’s] Orbital Reef faces is considerable: reimagining successful earthbound technologies—such as regenerative life support systems, expandable habitats and 3D printing—but now in orbit, on a commercially viable platform.”
FUTUREThis Gigantic 3D Printer Could Reinvent Manufacturing
Nate Berg | Fast Company“This machine isn’t just spitting out basic building materials like some massive glue gun. It’s also able to do subtractive manufacturing, like milling, as well as utilize a robotic arm for more complicated tasks. A built-in system allows it to lay down fibers in a printed object that give it greater structural integrity, allowing printed spans to stretch farther, and enabling factory-based 3D printed buildings to become even larger.”
AUTOMATIONWayve Raises $1B to Take Its Tesla-Like Technology for Self-Driving to Many Carmakers
Mike Butcher | TechCrunch“Wayve calls its hardware-agnostic mapless product an ‘Embodied AI,’ and it plans to distribute its platform not just to car makers but also to robotics companies serving manufacturers of all descriptions, allowing the platform to learn from human behavior in a wide variety of real-world environments.”
BIOTECHThe US Is Cracking Down on Synthetic DNA
Emily Mullin | Wired“Synthesizing DNA has been possible for decades, but it’s become increasingly easier, cheaper, and faster to do so in recent years thanks to new technology that can ‘print’ custom gene sequences. Now, dozens of companies around the world make and ship synthetic nucleic acids en masse. And with AI, it’s becoming possible to create entirely new sequences that don’t exist in nature—including those that could pose a threat to humans or other living things.”
SPACEFall Into a Black Hole in Mind-Bending NASA Animation
Robert Lea | Space.com“If you’ve ever wondered what would happen if you were unlucky enough to fall into a black hole, NASA has your answer. A visualization created on a NASA supercomputer to celebrate the beginning of black hole week on Monday (May 6) takes the viewer on a one-way plunge beyond the event horizon of a black hole.”
ENERGYA Company Is Building a Giant Compressed-Air Battery in the Australian Outback
Dan Gearino | Wired“Toronto-based Hydrostor is one of the businesses developing long-duration energy storage that has moved beyond lab scale and is now focusing on building big things. The company makes systems that store energy underground in the form of compressed air, which can be released to produce electricity for eight hours or longer.”
SCIENCEThe Way Whales Communicate Is Closer to Human Language Than We Realized
Rhiannon Williams | MIT Technology Review“A team of researchers led by Pratyusha Sharma at MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) working with Project CETI, a nonprofit focused on using AI to understand whales, used statistical models to analyze whale codas and managed to identify a structure to their language that’s similar to features of the complex vocalizations humans use. Their findings represent a tool future research could use to decipher not just the structure but the actual meaning of whale sounds.”
Image Credit: Benjamin Cheng / Unsplash
Pulling carbon dioxide out of the atmosphere is likely to be a crucial weapon in the battle against climate change. And now global carbon capture capacity has quadrupled with the opening of the world’s largest direct air capture plant in Iceland.
Scientists and policymakers initially resisted proposals to remove CO2 from the atmosphere, due to concerns it could lead to a reduced sense of urgency around emissions reductions. But with progress on that front falling behind schedule, there’s been growing acceptance that carbon capture will be crucial if we want to avoid the worst consequences of climate change.
A variety of approaches, including reforestation, regenerative agriculture, and efforts to lock carbon up in minerals, could play a role. But the approach garnering most of the attention is direct air capture, which relies on large facilities powered by renewable energy to suck CO2 out of the air.
One of the leaders in this space is Swiss company Climeworks, whose Orca plant in Iceland previously held the title for world’s largest. But this week, the company started operations at a new plant called Mammoth that has nearly ten times the capacity. The facility, also in Iceland, will be able to extract 36,000 tons of CO2 a year, which is nearly four times the 10,000 tons a year currently being captured globally.
“Starting operations of our Mammoth plant is another proof point in Climeworks’ scale-up journey to megaton capacity by 2030 and gigaton by 2050,” co-CEO of Climeworks Jan Wurzbacher said in a statement. “Constructing multiple real-world plants in rapid sequences makes Climeworks the most deployed carbon removal company with direct air capture at the core.”
Climeworks plants use fans to suck air into large collector units filled with a material called a sorbent, which absorbs CO2. Once the sorbent is saturated, the collector shuts and is heated to roughly 212 degrees Fahrenheit to release the CO2.
The Mammoth plant will eventually feature 72 of these collector units, though only 12 are currently operational. That’s still more than Orca’s eight units, which allows it to capture roughly 4,000 tons of CO2 a year. Adding an extra level to the stacks of collectors has also reduced land use per ton of CO2 captured, while a new V-shaped configuration improves airflow, boosting performance.
To permanently store the captured carbon, Climeworks has partnered with Icelandic company Carbfix, which has developed a process to inject CO2 dissolved in water deep into porous rock formations made of basalt. Over the course of a couple years, the dissolved CO2 reacts with the rocks to form solid carbonate minerals that are stable for thousands of years.
With the Orca plant, CO2 had to be transported through hundreds of meters of pipeline to Carbfix’s storage site. But Mammoth features two injection wells on-site reducing transportation costs. It also has a new CO2 absorption tower that dissolves the gas in water at lower pressures, reducing energy costs compared to the previous approach.
Climeworks has much bigger ambitions than Mammoth though. The US government has earmarked $3.5 billion to build four direct air capture hubs, each capable of capturing one million tons of CO2 a year, and Climeworks will provide the technology for one of the proposed facilities in Louisiana.
The company says it’s aiming to reach megaton-scale—removing one million tons a year—by 2030 and gigaton-scale—a billion tons a year by 2050. Hopefully, they won’t be the only ones, because climate forecasts suggest we’ll need to be removing 3.5 gigatons of CO2 a year by 2050 to keep warming below 1.5 degrees Celsius.
There’s also little clarity on the economics of the approach. According to Reuters, Climeworks did not reveal how much it costs Mammoth to remove each ton of CO2, though it said it’s targeting $400-600 per ton by 2030 and $200-350 per ton by 2040. And while plants in Iceland can take advantage of abundant, green geothermal energy, it’s less clear what they will rely on elsewhere.
Either way, there’s growing agreement that carbon capture will be an important part of our efforts to tackle climate change. While Mammoth might not make much of a dent in emissions, it’s a promising sign that direct air capture technology is maturing.
Image Credit: Climeworks
Proteins are biological workhorses.
They build our bodies and orchestrate the molecular processes in cells that keep them healthy. They also present a wealth of targets for new medications. From everyday pain relievers to sophisticated cancer immunotherapies, most current drugs interact with a protein. Deciphering protein architectures could lead to new treatments.
That was the promise of AlphaFold 2, an AI model from Google DeepMind that predicted how proteins gain their distinctive shapes based on the sequences of their constituent molecules alone. Released in 2020, the tool was a breakthrough half a decade in the making.
But proteins don’t work alone. They inhabit an entire cellular universe and often collaborate with other molecular inhabitants like, for example, DNA, the body’s genetic blueprint.
This week, DeepMind and Isomorphic Labs released a big new update that allows the algorithm to predict how proteins work inside cells. Instead of only modeling their structures, the new version—dubbed AlphaFold 3—can also map a protein’s interactions with other molecules.
For example, could a protein bind to a disease-causing gene and shut it down? Can adding new genes to crops make them resilient to viruses? Can the algorithm help us rapidly engineer new vaccines to tackle existing diseases—or whatever new ones nature throws at us?
“Biology is a dynamic system…you have to understand how properties of biology emerge due to the interactions between different molecules in the cell,” said Demis Hassabis, the CEO of DeepMind, in a press conference.
AlphaFold 3 helps explain “not only how proteins talk to themselves, but also how they talk to other parts of the body,” said lead author Dr. John Jumper.
The team is releasing the new AI online for academic researchers by way of an interface called the AlphaFold Server. With a few clicks, a biologist can run a simulation of an idea in minutes, compared to the weeks or months usually needed for experiments in a lab.
Dr. Julien Bergeron at King’s College London, who builds nano-protein machines but was not involved in the work, said the AI is “transformative science” for speeding up research, which could ultimately lead to nanotech devices powered by the body’s mechanisms alone.
For Dr. Frank Uhlmann at the Francis Crick Laboratory, who gained early access to AlphaFold 3 and used it to study how DNA divides when cells divide, the AI is “democratizing discovery research.”
Molecular UniverseProteins are finicky creatures. They’re made of strings of molecules called amino acids that fold into intricate three-dimensional shapes that determine what the protein can do.
Sometimes the folding processes goes wrong. In Alzheimer’s disease, misfolded proteins clump into dysfunctional blobs that clog up around and inside brain cells.
Scientists have long tried to engineer drugs to break up disease-causing proteins. One strategy is to map protein structure—know thy enemy (and friends). Before AlphaFold, this was done with electron microscopy, which captures a protein’s structure at the atomic level. But it’s expensive, labor intensive, and not all proteins can tolerate the scan.
Which is why AlphaFold 2 was revolutionary. Using amino acid sequences alone—the constituent molecules that make up proteins—the algorithm could predict a protein’s final structure with startling accuracy. DeepMind used AlphaFold to map the structure of nearly all proteins known to science and how they interact. According to the AI lab, in just three years, researchers have mapped roughly six million protein structures using AlphaFold 2.
But to Jumper, modeling proteins isn’t enough. To design new drugs, you have to think holistically about the cell’s whole ecosystem.
It’s an idea championed by Dr. David Baker at the University of Washington, another pioneer in the protein-prediction space. In 2021, Baker’s team released AI-based software called RoseTTAFold All-Atom to tackle interactions between proteins and other biomolecules.
Picturing these interactions can help solve tough medical challenges, allowing scientists to design better cancer treatments or more precise gene therapies, for example.
“Properties of biology emerge through the interactions between different molecules in the cell,” said Hassabis in the press conference. “You can think about AlphaFold 3 as our first big sort of step towards that.”
A RevampAlphaFold 3 builds on its predecessor, but with significant renovations.
One way to gauge how a protein interacts with other molecules is to examine evolution. Another is to map a protein’s 3D structure and—with a dose of physics—predict how it can grab onto other molecules. While AlphaFold 2 mostly used an evolutionary approach—training the AI on what we already know about protein evolution in nature—the new version heavily embraces physical and chemical modeling.
Some of this includes chemical changes. Proteins are often tagged with different chemicals. These tags sometimes change protein structure but are essential to their behavior—they can literally determine a cell’s fate, for example, life, senescence, or death.
The algorithm’s overall setup makes some use of its predecessor’s machinery to map proteins, DNA, and other molecules and their interactions. But the team also looked to diffusion models—the algorithms behind OpenAI’s DALL-E 2 image generator—to capture structures at the atomic level. Diffusion models are trained to reverse noisy images in steps until they arrive at a prediction for what the image (or in this case a 3D model of a biomolecule) should look like without the noise. This addition made a “substantial change” to performance, said Jumper.
Like AlphaFold 2, the new version has a built-in “sanity check” that indicates how confident it is in a generated model so scientists can proofread its outputs. This has been a core component of all their work, said the DeepMind team. They trained the AI using the Protein Data Bank, an open-source compilation of 3D protein structures that’s constantly updated, including new experimentally validated structures of proteins binding to DNA and other biomolecules
Pitted against existing software, AlphaFold 3 broke records. One test for molecular interactions between proteins and small molecules—ones that could become medications—succeeded 76 percent of the time. Previous attempts were successful in roughly 42 percent of cases.
When it comes to deciphering protein functions, AlphaFold 3 “seeks to solve the exact same problem [as RoseTTAFold All-Atom]…but is clearly more accurate,” Baker told Singularity Hub.
But the tool’s accuracy depends on which interaction is being modeled. The algorithm isn’t yet great at protein-RNA interactions, for example, Columbia University’s Mohammed AlQuraishi told MIT Technology Review. Overall, accuracy ranged from 40 to more than 80 percent.
AI to Real LifeUnlike previous iterations, DeepMind isn’t open-sourcing AlphaFold 3’s code. Instead, they’re releasing the tool as a free online platform, called AlphaFold Server, that allows scientists to test their ideas for protein interactions with just a few clicks.
AlphaFold 2 required technical expertise to install and run the software. The server, in contrast, can help people unfamiliar with code to use the tool. It’s for non-commercial use only and can’t be reused to train other machine learning models for protein prediction. But it is freely available for scientists to try. The team envisions the software helping develop new antibodies and other treatments at a faster rate. Isomorphic Labs, a spin-off of DeepMind, is already using AlphaFold 3 to develop medications for a variety of diseases.
For Bergeron, the upgrade is “transformative.” Instead of spending years in the lab, it’s now possible to mimic protein interactions in silico—a computer simulation—before beginning the labor- and time-intensive work of investigating promising solutions using cells.
“I’m pretty certain that every structural biology and protein biochemistry research group in the world will immediately adopt this system,” he said.
Image Credit: Google DeepMind
There are well over a million asteroids in the solar system. Most don’t cross paths with Earth, but some do and there’s a risk one of these will collide with our planet. Taking a census of nearby space rocks, then, is prudent. As conventional wisdom would have it, we’ll need lots of telescopes, time, and teams of astronomers to find them.
But maybe not, according to the B612 Foundation’s Asteroid Institute.
In tandem with Google Cloud, the Asteroid Institute recently announced they’ve spotted 27,500 new asteroids—more than all discoveries worldwide last year—without requiring a single new observation. Instead, over a period of just a few weeks, the team used new software to scour 1.7 billion points of light in some 400,000 images taken over seven years and archived by the National Optical-Infrared Astronomy Research Laboratory (NOIRLab).
To discover new asteroids, astronomers usually need multiple images over several nights (or more) to find moving objects and calculate their orbits. This means they have to make new observations with asteroid discovery in mind. There is also, however, a trove of existing one-time observations made for other purposes, and these are likely packed with photobombing asteroids. But identifying them is difficult and computationally intensive.
Working with the University of Washington, the Asteroid Institute team developed an algorithm, Tracklet-less Heliocentric Orbit Recovery, or THOR, to scan archived images recorded at different times or even by different telescopes. The tool can tell if moving points of light recorded in separate images are the same object. Many of these will be asteroids.
Running THOR on Google Cloud, the team scoured the NOIRLab data and found plenty. Most of the new asteroids are in the main asteroid belt, but more than 100 are near-Earth asteroids. Though the team classified their findings as “high-confidence,” these near-Earth asteroids have not yet been confirmed. They’ll submit their findings to the Minor Planet Center, and ESA and NASA will then verify orbits and assess risk. (The team says they have no reason to believe any pose a risk to Earth.)
While the new software could speed up the pace of discovery, the process still requires volunteers and scientists to manually review the algorithm’s finds. The team plans to use the raw data from the recent run including human review to train an AI model. The hope is that some or all of the manual review process can be automated, making the process even faster.
In the future, the algorithm will go to work on data from the Vera C. Rubin Observatory, a telescope in Chile’s Atacama desert. The telescope, set to begin operations next year, will make twice nightly observations of the sky with asteroid detection in mind. THOR may be able to make discoveries with only one nightly run, freeing the telescope up for other work.
All this is in service of the plan to discover as many Earth-crossing asteroids as possible.
According to NASA, we’ve found over 1.3 million asteroids, 35,000 of which are near-Earth asteroids. Of these, over 90 percent of the biggest and most dangerous—in the same class as the impact that ended the dinosaurs—have been discovered. Scientists are now filling out the list of smaller but still dangerous asteroids. The vast majority of all known asteroids were catalogued this century. Before that we were flying blind.
While no dangerous asteroids are known to be headed our way soon, space agencies are working on a plan of action—sans nukes and Bruce Willis—should we discover one.
In 2022, NASA rammed the DART spacecraft into an asteroid, Dymorphos, to see if it would deflect the space rock’s orbit. This is a planetary defense strategy known as a “kinetic impactor.” Scientists thought DART might change the asteroid’s orbit by 7 minutes. Instead, DART changed Dymorphos’ orbit by a whopping 33 minutes, much of which was due to recoil produced by a giant plume of material ejected by the impact.
The conclusion of scientists studying the aftermath? “Kinetic impactor technology is a viable technique to potentially defend Earth if necessary.” With the caveat: If we have enough time. Such impacts amount to a nudge, so we need years of advance notice.
Algorithms like THOR could help give us that crucial heads up.
Image Credit: B612 Foundation
In March, we saw the launch of a “ChatGPT for music” called Suno, which uses generative AI to produce realistic songs on demand from short text prompts. A few weeks later, a similar competitor—Udio—arrived on the scene.
I’ve been working with various creative computational tools for the past 15 years, both as a researcher and a producer, and the recent pace of change has floored me. As I’ve argued elsewhere, the view that AI systems will never make “real” music like humans do should be understood more as a claim about social context than technical capability.
The argument “sure, it can make expressive, complex-structured, natural-sounding, virtuosic, original music which can stir human emotions, but AI can’t make proper music” can easily begin to sound like something from a Monty Python sketch.
After playing with Suno and Udio, I’ve been thinking about what it is exactly they change—and what they might mean not only for the way professionals and amateur artists create music, but the way all of us consume it.
Expressing Emotion Without Feeling ItGenerating audio from text prompts in itself is nothing new. However, Suno and Udio have made an obvious development: from a simple text prompt, they generate song lyrics (using a ChatGPT-like text generator), feed them into a generative voice model, and integrate the “vocals” with generated music to produce a coherent song segment.
This integration is a small but remarkable feat. The systems are very good at making up coherent songs that sound expressively “sung” (there I go anthropomorphizing).
The effect can be uncanny. I know it’s AI, but the voice can still cut through with emotional impact. When the music performs a perfectly executed end-of-bar pirouette into a new section, my brain gets some of those little sparks of pattern-processing joy that I might get listening to a great band.
To me this highlights something sometimes missed about musical expression: AI doesn’t need to experience emotions and life events to successfully express them in music that resonates with people.
Music as an Everyday LanguageLike other generative AI products, Suno and Udio were trained on vast amounts of existing work by real humans—and there is much debate about those humans’ intellectual property rights.
Nevertheless, these tools may mark the dawn of mainstream AI music culture. They offer new forms of musical engagement that people will just want to use, to explore, to play with, and actually listen to for their own enjoyment.
AI capable of “end-to-end” music creation is arguably not technology for makers of music, but for consumers of music. For now it remains unclear whether users of Udio and Suno are creators or consumers—or whether the distinction is even useful.
A long-observed phenomenon in creative technologies is that as something becomes easier and cheaper to produce, it is used for more casual expression. As a result, the medium goes from an exclusive high art form to more of an everyday language—think what smartphones have done to photography.
So imagine you could send your father a professionally produced song all about him for his birthday, with minimal cost and effort, in a style of his preference—a modern-day birthday card. Researchers have long considered this eventuality, and now we can do it. Happy birthday, Dad!
Mr Bown’s Blues. Generated by Oliver Bown using Udio [3.75 MB (download)]Can You Create Without Control?Whatever these systems have achieved and may achieve in the near future, they face a glaring limitation: the lack of control.
Text prompts are often not much good as precise instructions, especially in music. So these tools are fit for blind search—a kind of wandering through the space of possibilities—but not for accurate control. (That’s not to diminish their value. Blind search can be a powerful creative force.)
Viewing these tools as a practicing music producer, things look very different. Although Udio’s about page says “anyone with a tune, some lyrics, or a funny idea can now express themselves in music,” I don’t feel I have enough control to express myself with these tools.
I can see them being useful to seed raw materials for manipulation, much like samples and field recordings. But when I’m seeking to express myself, I need control.
Using Suno, I had some fun finding the most gnarly dark techno grooves I could get out of it. The result was something I would absolutely use in a track.
Cheese Lovers’ Anthem. Generated by Oliver Bown using Suno [2.75 MB (download)]But I found I could also just gladly listen. I felt no compulsion to add anything or manipulate the result to add my mark.
And many jurisdictions have declared that you won’t be awarded copyright for something just because you prompted it into existence with AI.
For a start, the output depends just as much on everything that went into the AI—including the creative work of millions of other artists. Arguably, you didn’t do the work of creation. You simply requested it.
New Musical Experiences in the No-Man’s Land Between Production and ConsumptionSo Udio’s declaration that anyone can express themselves in music is an interesting provocation. The people who use tools like Suno and Udio may be considered more consumers of music AI experiences than creators of music AI works, or as with many technological impacts, we may need to come up with new concepts for what they’re doing.
A shift to generative music may draw attention away from current forms of musical culture, just as the era of recorded music saw the diminishing (but not death) of orchestral music, which was once the only way to hear complex, timbrally rich and loud music. If engagement in these new types of music culture and exchange explodes, we may see reduced engagement in the traditional music consumption of artists, bands, radio and playlists.
While it is too early to tell what the impact will be, we should be attentive. The effort to defend existing creators’ intellectual property protections, a significant moral rights issue, is part of this equation.
But even if it succeeds I believe it won’t fundamentally address this potentially explosive shift in culture, and claims that such music might be inferior also have had little effect in halting cultural change historically, as with techno or even jazz, long ago. Government AI policies may need to look beyond these issues to understand how music works socially and to ensure that our musical cultures are vibrant, sustainable, enriching, and meaningful for both individuals and communities.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Pawel Czerwinski / Unsplash
We all know that exercise is good for us.
A brisk walk of roughly an hour a day can stave off chronic diseases, including heart or blood vessel issues and Type 2 diabetes. Regular exercise delays memory loss due to aging, boosts the immune system, slashes stress, and may even increase lifespan.
For decades, scientists have tried to understand why. Throughout the body, our organs and tissues release a wide variety of molecules during—and even after—exercise to reap its benefits. But no single molecule works alone. The hard part is understanding how they collaborate in networks after exercise.
Enter the Molecular Transducers of Physical Activity Consortium (MoTrPAC) project. Established nearly a decade ago and funded by the National Institutes of Health (NIH), the project aims to create comprehensive molecular maps of how genes and proteins change after exercise in both rodents and people. Rather than focusing on single proteins or genes, the project takes a Google Earth approach—let’s see the overall picture.
It’s not simply for scientific curiosity. If we can find important molecular processes that trigger exercise benefits, we could potentially mimic those reactions using medications and help people who physically can’t work out—a sort of “exercise in a pill.”
This month, the project announced multiple results.
In one study, scientists built an atlas of bodily changes before, during, and after exercise in rats. Altogether, the team collected nearly 9,500 samples across multiple tissues to examine how exercise changes gene expression across the body. Another study detailed differences between sexes after exercise. A third team mapped exercise-related genes to those associated with diseases.
According to the project’s NIH webpage: “When the MoTrPAC study is completed, it will be the largest research study examining the link between exercise and its improvement of human health.”
Work ItOur tissues are chatterboxes. The gut “talks” to the brain through a vast maze of molecules. Muscles pump out proteins to fine-tune immune system defenses. Plasma—the liquid part of blood—can transfer the learning and memory benefits of running when injected into “couch potato” mice and delay cognitive decline.
Over the years, scientists have identified individual molecules and processes that could mediate these effects, but the health benefits are likely due to networks of molecules working together.
“MoTrPAC was launched to fill an important gap in exercise research,” said former NIH director Dr. Francis Collins in a 2020 press release. “It shifts focus from a specific organ or disease to a fundamental understanding of exercise at the molecular level—an understanding that may lead to personalized, prescribed exercise regimens based on an individual’s needs and traits.”
The project has two arms. One observes rodents before, during, and after wheel running to build comprehensive maps of molecular changes due to exercise. These maps aim to capture gene expression alongside metabolic and epigenetic changes in multiple organs.
Another arm will recruit roughly 2,600 healthy volunteers aged 10 to over 60 years old. With a large pool of participants, the team hopes to account for variation between people and even identify differences in the body’s response to exercise based on age, gender, or race. The volunteers will undergo 12 weeks of exercise, either endurance training—such as long-distance running—or weightlifting.
Altogether, the goal is to detect how exercise affects cells at a molecular level in multiple tissue types—blood, fat, and muscle.
Exercise EncyclopediaLast week, MoTrPAC released an initial wave of findings.
In one study, the group collected blood and 18 different tissue samples from adult rats, both male and female, as they happily ran for a week to two months. The team then screened how the body changes with exercise by comparing rats that work out with “couch potato” rats as a baseline. Physical training increased the rats’ aerobic capacity—the amount of oxygen the body can use—by roughly 17 percent.
Next, the team analyzed the molecular fingerprints of exercise in whole blood, plasma, and 18 solid tissues, including heart, liver, lung, kidney, fat tissue, and the hippocampus, a brain region associated with memory. They used an impressive array of tools that, for example, captured changes in overall gene expression and the epigenetic landscape. Others mapped differences in the body’s proteins, fat, immune system, and metabolism.
“Altogether, datasets were generated from 9,466 assays across 211 combinations of tissues and molecular platforms,” wrote the team.
Using an AI-based method, they integrated the results across time into a comprehensive molecular map. The map pinpointed multiple molecular changes that could dampen liver diseases, inflammatory bowel disease, and protect against heart health and tissue injuries.
All this represents “the first whole-organism molecular map” capturing how exercise changes the body, wrote the team. (All of the data is free to explore.)
Venus and MarsMost previous studies on exercise in rodents focused on males. What about the ladies?
After analyzing the MoTrPAC database, another study found that exercise changes the body’s molecular signaling differently depending on biological sex.
After running, female rats triggered genes in white fat—the type under the skin—related to insulin signaling and the body’s ability to form fat. Meanwhile, males showed molecular signatures of a ramped up metabolism.
With consistent exercise, male rats rapidly lost fat and weight, whereas females maintained their curves but with improved insulin signaling, which might protect them against heart diseases.
A third study integrated gene expression data collected from exercised rats with disease-relevant gene databases previously found in humans. The goal is to link workout-related genes in a particular organ or tissue with a disease or other health outcome—what the authors call “trait-tissue-gene triplets.” Overall, they found 5,523 triplets “to serve as a valuable starting point for future investigations,” they wrote.
We’re only scratching the surface of the complex puzzle that is exercise. Through extensive mapping efforts, the project aims to eventually tailor workout regimens for people with chronic diseases or identify key “druggable” components that could confer some health benefits of exercise with a pill.
“This is an unprecedented large-scale effort to begin to explore—in extreme detail—the biochemical, physiological, and clinical impact of exercise,” Dr. Russell Tracy at the University of Vermont, a MoTrPAC member, said in a press release.
Image Credit: Fitsum Admasu / Unsplash
ARTIFICIAL INTELLIGENCE*Sam Altman Says Helpful Agents Are Poised to Become AI’s Killer Function
James O’Donnell | MIT Technology Review*“Altman, who was visiting Cambridge for a series of events hosted by Harvard and the venture capital firm Xfund, described the killer app for AI as a ‘super-competent colleague that knows absolutely everything about my whole life, every email, every conversation I’ve ever had, but doesn’t feel like an extension.’ It could tackle some tasks instantly, he said, and for more complex ones it could go off and make an attempt, but come back with questions for you if it needs to.”archive page
COMPUTINGExpect a Wave of Wafer-Scale Computers
Samuel K. Moore | IEEE Spectrum“At TSMC’s North American Technology Symposium on Wednesday, the company detailed both its semiconductor technology and chip-packaging technology road maps. While the former is key to keeping the traditional part of Moore’s Law going, the latter could accelerate a trend toward processors made from more and more silicon, leading quickly to systems the size of a full silicon wafer. …In 2027, you will get a full-wafer integration that delivers 40 times as much compute power, more than 40 reticles’ worth of silicon, and room for more than 60 high-bandwidth memory chips, TSMC predicts.”
FUTURENick Bostrom Made the World Fear AI. Now He Asks: What if It Fixes Everything?
Will Knight | Wired“With the publication of his last book, Superintelligence: Paths, Dangers, Strategies, in 2014, Bostrom drew public attention to what was then a fringe idea—that AI would advance to a point where it might turn against and delete humanity. …Bostrom’s new book takes a very different tack. Rather than play the doomy hits, Deep Utopia: Life and Meaning in a Solved World, considers a future in which humanity has successfully developed superintelligent machines but averted disaster.”
TECHAI Start-Ups Face a Rough Financial Reality Check
Cade Metz, Karen Weise, and Tripp Mickle | The New York Times“The AI revolution, it is becoming clear in Silicon Valley, is going to come with a very big price tag. And the tech companies that have bet their futures on it are scrambling to figure out how to close the gap between those expenses and the profits they hope to make somewhere down the line.”
ROBOTICSEvery Tech Company Wants to Be Like Boston Dynamics
Jacob Stern | The Atlantic“Clips of robots running faster than Usain Bolt and dancing in sync, among many others, have helped [Boston Dynamics] reach true influencer status. Its videos have now been viewed more than 800 million times, far more than those of much bigger tech companies, such as Tesla and OpenAI. The creator of Black Mirror even admitted that an episode in which killer robot dogs chase a band of survivors across an apocalyptic wasteland was directly inspired by Boston Dynamics’ videos.”
ETHICSChatGPT Shows Better Moral Judgment Than a College Undergrad
Kyle Orland | Ars Technica“In ‘Attributions toward artificial agents in a modified Moral Turing Test’…[Georgia State University] researchers found that morality judgments given by ChatGPT4 were ‘perceived as superior in quality to humans’ along a variety of dimensions like virtuosity and intelligence. But before you start to worry that philosophy professors will soon be replaced by hyper-moral AIs, there are some important caveats to consider.”
SPACENew Space Company Seeks to Solve Orbital Mobility With High Delta-V Spacecraft
Eric Berger | Ars Technica“[Portal Space Systems founder, Jeff Thornburg] envisions a fleet of refuelable Supernova vehicles at medium-Earth and geostationary orbit capable of swooping down to various orbits and providing services such as propellant delivery, mobility, and observation for commercial and military satellites. His vision is to provide real-time, responsive capability for existing satellites. If one needs to make an emergency maneuver, a Supernova vehicle could be there within a couple of hours. ‘If we’re going to have a true space economy, that means logistics and supply services,’ he said.”
AUTOMATIONGoogle’s Waymo Is Expanding Its Self-Driving ‘Robotaxi’ Testing
William Gavin | Quartz“Waymo plans to soon start testing fully autonomous rides across California’s San Francisco Peninsula, despite criticism and concerns from residents and city officials. In the coming weeks, Waymo employees will begin testing rides without a human driver on city streets north of San Mateo, the company said Friday.”
VIRTUAL REALITYUkraine Unveils AI-Generated Foreign Ministry Spokesperson
Agence France-Presse | The Guardian“Dressed in a dark suit, the spokesperson introduced herself as Victoria Shi, a ‘digital person,’ in a presentation posted on social media. The figure gesticulates with her hands and moves her head as she speaks. The foreign ministry’s press service said that the statements given by Shi would not be generated by AI but ‘written and verified by real people.'”
Image Credit: Drew Walker / Unsplash
Getting microbes to eat plastic is a frequently touted solution to our growing waste problem, but making the approach practical is tricky. A new technique that impregnates plastic with the spores of plastic-eating bacteria could make the idea a reality.
The impact of plastic waste on the environment and our health has gained increasing attention in recent years. The latest round of UN talks aiming for a global treaty to end plastic pollution just concluded in Ottawa, Canada earlier this week, though considerable disagreements remain.
Recycling will inevitably be a crucial ingredient in any plan to deal with the problem. But a 2022 report from the Organization for Economic Cooperation and Development found only 9 percent of plastic waste ever gets recycled. That’s partly due to the fact that existing recycling approaches are energy intensive and time consuming.
This has spurred a search for new approaches, and one of the most promising is the use of bacteria to break down plastics, either by rendering them harmless or using them to produce building blocks that can be repurposed into other valuable materials and chemicals. The main problem with the approach is making sure plastic waste ends up in the same place as these plastic-loving bacteria.
Now, researchers have come up with an ingenious solution: embed microbes in plastic during the manufacturing process. Not only did the approach result in 93 percent of the plastic biodegrading within five months, but it even increased the strength and stretchability of the material.
“What’s remarkable is that our material breaks down even without the presence of additional microbes,” project co-leader Jon Pokorski from the University of California San Diego said in a press release.
“Chances are, most of these plastics will likely not end up in microbially rich composting facilities. So this ability to self-degrade in a microbe-free environment makes our technology more versatile.”
The main challenge when it came to incorporating bacteria into plastics was making sure they survived the high temperatures involved in manufacturing the material. The researchers worked with a soft plastic called thermoplastic polyurethane (TPU), which is used in footwear, cushions, and memory foam. TPU is manufactured by melting pellets of the material at around 275 degrees Fahrenheit and then extruding it into the desired shape.
Given the need to survive these high temperatures, the researchers selected a plastic-eating bacteria called Bacillus subtilis, which can form spores allowing it to survive harsh conditions. Even then, they discovered more than 90 percent of the bacteria were killed in under a minute at those temperatures.
So, the team used a technique called adaptive laboratory evolution to create a more heat-tolerant strain of the bacteria. They dunked the spores in boiling water for increasing lengths of time, collecting the survivors, growing the population back up, and then repeating the process. Over time, this selected for mutations that conferred greater heat tolerance, until the researchers were left with a strain that was able to withstand the manufacturing process.
When they incorporated the spores into the plastic, they were surprised to find the bacteria actually improved the mechanical properties of the material. In essence, the spores acted like steel rebar in concrete, making it harder to break and increasing its stretchability.
To test whether the impregnated spores could help the plastic biodegrade, the researchers took small strips of the plastic and put them in sterilized compost. After five months, they found the strips had lost 93 percent of their mass compared to 44 percent for TPU without spores, which suggests the spores were reactivated by nutrients in the compost and helped degrade the plastic substantially faster.
It’s unclear if the approach would work with other plastics, though the researchers say they plan to find out. There is also a danger the spores could reactivate before the plastic is disposed of, which could shorten the life of any products made with it. Perhaps most crucially, plastics researcher Steve Fletcher from the University of Portsmouth in the UK told the BBC that this kind of technology could distract from efforts to limit plastic waste.
“Care must be taken with potential solutions of this sort, which could give the impression that we should worry less about plastic pollution because any plastic leaking into the environment will quickly, and ideally safely, degrade,” he said. “For the vast majority of plastics, this is not the case.”
Given the scale of the plastic pollution problem today though, any attempt to mitigate the harm should be welcomed. While it’s early days, the prospect of making plastic that can biodegrade itself could go a long way towards tackling the problem.
Image Credit: David Baillot/UC San Diego Jacobs School of Engineering
To anyone living in a city where autonomous vehicles operate, it would seem they need a lot of practice. Robotaxis travel millions of miles a year on public roads in an effort to gather data from sensors—including cameras, radar, and lidar—to train the neural networks that operate them.
In recent years, due to a striking improvement in the fidelity and realism of computer graphics technology, simulation is increasingly being used to accelerate the development of these algorithms. Waymo, for example, says its autonomous vehicles have already driven some 20 billion miles in simulation. In fact, all kinds of machines, from industrial robots to drones, are gathering a growing amount of their training data and practice hours inside virtual worlds.
According to Gautham Sholingar, a senior manager at Nvidia focused on autonomous vehicle simulation, one key benefit is accounting for obscure scenarios for which it would be nearly impossible to gather training data in the real world.
“Without simulation, there are some scenarios that are just hard to account for. There will always be edge cases which are difficult to collect data for, either because they are dangerous and involve pedestrians or things that are challenging to measure accurately like the velocity of faraway objects. That’s where simulation really shines,” he told me in an interview for Singularity Hub.
While it isn’t ethical to have someone run unexpectedly into a street to train AI to handle such a situation, it’s significantly less problematic for an animated character inside a virtual world.
Industrial use of simulation has been around for decades, something Sholingar pointed out, but a convergence of improvements in computing power, the ability to model complex physics, and the development of the GPUs powering today’s graphics indicate we may be witnessing a turning point in the use of simulated worlds for AI training.
Graphics quality matters because of the way AI “sees” the world.
When a neural network processes image data, it’s converting each pixel’s color into a corresponding number. For black and white images, the number ranges from 0, which indicates a fully black pixel, up to 255, which is fully white, with numbers in between representing some variation of grey. For color images, the widely used RGB (red, green, blue) model can correspond to over 16 million possible colors. So as graphics rendering technology becomes ever more photorealistic, the distinction between pixels captured by real-world cameras and ones rendered in a game engine is falling away.
Simulation is also a powerful tool because it’s increasingly able to generate synthetic data for sensors beyond just cameras. While high-quality graphics are both appealing and familiar to human eyes, which is useful in training camera sensors, rendering engines are also able to generate radar and lidar data as well. Combining these synthetic datasets inside a simulation allows the algorithm to train using all the various types of sensors commonly used by AVs.
Due to their expertise in producing the GPUs needed to generate high-quality graphics, Nvidia have positioned themselves as leaders in the space. In 2021, the company launched Omniverse, a simulation platform capable of rendering high-quality synthetic sensor data and modeling real-world physics relevant to a variety of industries. Now, developers are using Omniverse to generate sensor data to train autonomous vehicles and other robotic systems.
In our discussion, Sholingar described some specific ways these types of simulations may be useful in accelerating development. The first involves the fact that with a bit of retraining, perception algorithms developed for one type of vehicle can be re-used for other types as well. However, because the new vehicle has a different sensor configuration, the algorithm will be seeing the world from a new point of view, which can reduce its performance.
“Let’s say you developed your AV on a sedan, and you need to go to an SUV. Well, to train it then someone must change all the sensors and remount them on an SUV. That process takes time, and it can be expensive. Synthetic data can help accelerate that kind of development,” Sholingar said.
Another area involves training algorithms to accurately detect faraway objects, especially in highway scenarios at high speeds. Since objects over 200 meters away often appear as just a few pixels and can be difficult for humans to label, there isn’t typically enough training data for them.
“For the far ranges, where it’s hard to annotate the data accurately, our goal was to augment those parts of the dataset,” Sholingar said. “In our experiment, using our simulation tools, we added more synthetic data and bounding boxes for cars at 300 meters and ran experiments to evaluate whether this improves our algorithm’s performance.”
According to Sholingar, these efforts allowed their algorithm to detect objects more accurately beyond 200 meters, something only made possible by their use of synthetic data.
While many of these developments are due to better visual fidelity and photorealism, Sholingar also stressed this is only one aspect of what makes capable real-world simulations.
“There is a tendency to get caught up in how beautiful the simulation looks since we see these visuals, and it’s very pleasing. What really matters is how the AI algorithms perceive these pixels. But beyond the appearance, there are at least two other major aspects which are crucial to mimicking reality in a simulation.”
First, engineers need to ensure there is enough representative content in the simulation. This is important because an AI must be able to detect a diversity of objects in the real world, including pedestrians with different colored clothes or cars with unusual shapes, like roof racks with bicycles or surfboards.
Second, simulations have to depict a wide range of pedestrian and vehicle behavior. Machine learning algorithms need to know how to handle scenarios where a pedestrian stops to look at their phone or pauses unexpectedly when crossing a street. Other vehicles can behave in unexpected ways too, like cutting in close or pausing to wave an oncoming vehicle forward.
“When we say realism in the context of simulation, it often ends up being associated only with the visual appearance part of it, but I usually try to look at all three of these aspects. If you can accurately represent the content, behavior, and appearance, then you can start moving in the direction of being realistic,” he said.
It also became clear in our conversation that while simulation will be an increasingly valuable tool for generating synthetic data, it isn’t going to replace real-world data collection and testing.
“We should think of simulation as an accelerator to what we do in the real world. It can save time and money and help us with a diversity of edge-case scenarios, but ultimately it is a tool to augment datasets collected from real-world data collection,” he said.
Beyond Omniverse, the wider industry of helping “things that move” develop autonomy is undergoing a shift toward simulation. Tesla announced they’re using similar technology to develop automation in Unreal Engine, while Canadian startup, Waabi, is taking a simulation-first approach to training their self-driving software. Microsoft, meanwhile, has experimented with a similar tool to train autonomous drones, although the project was recently discontinued.
While training and testing in the real world will remain a crucial part of developing autonomous systems, the continued improvement of physics and graphics engine technology means that virtual worlds may offer a low-stakes sandbox for machine learning algorithms to mature into functional tools that can power our autonomous future.
Image Credit: Nvidia
Imagine the tap of a card that bought you a cup of coffee this morning also let a hacker halfway across the world access your bank account and buy themselves whatever they liked. Now imagine it wasn’t a one-off glitch, but it happened all the time: Imagine the locks that secure our electronic data suddenly stopped working.
This is not a science fiction scenario. It may well become a reality when sufficiently powerful quantum computers come online. These devices will use the strange properties of the quantum world to untangle secrets that would take ordinary computers more than a lifetime to decipher.
We don’t know when this will happen. However, many people and organizations are already concerned about so-called “harvest now, decrypt later” attacks, in which cybercriminals or other adversaries steal encrypted data now and store it away for the day when they can decrypt it with a quantum computer.
As the advent of quantum computers grows closer, cryptographers are trying to devise new mathematical schemes to secure data against their hypothetical attacks. The mathematics involved is highly complex—but the survival of our digital world may depend on it.
‘Quantum-Proof’ EncryptionThe task of cracking much current online security boils down to the mathematical problem of finding two numbers that, when multiplied together, produce a third number. You can think of this third number as a key that unlocks the secret information. As this number gets bigger, the amount of time it takes an ordinary computer to solve the problem becomes longer than our lifetimes.
Future quantum computers, however, should be able to crack these codes much more quickly. So the race is on to find new encryption algorithms that can stand up to a quantum attack.
The US National Institute of Standards and Technology has been calling for proposed “quantum-proof” encryption algorithms for years, but so far few have withstood scrutiny. (One proposed algorithm, called Supersingular Isogeny Key Encapsulation, was dramatically broken in 2022 with the aid of Australian mathematical software called Magma, developed at the University of Sydney.)
The race has been heating up this year. In February, Apple updated the security system for the iMessage platform to protect data that may be harvested for a post-quantum future.
Two weeks ago, scientists in China announced they had installed a new “encryption shield” to protect the Origin Wukong quantum computer from quantum attacks.
Around the same time, cryptographer Yilei Chen announced he had found a way quantum computers could attack an important class of algorithms based on the mathematics of lattices, which were considered some of the hardest to break. Lattice-based methods are part of Apple’s new iMessage security, as well as two of the three frontrunners for a standard post-quantum encryption algorithm.
What Is a Lattice-Based Algorithm?A lattice is an arrangement of points in a repeating structure, like the corners of tiles in a bathroom or the atoms in a diamond crystal. The tiles are two dimensional and the atoms in diamond are three dimensional, but mathematically we can make lattices with many more dimensions.
Most lattice-based cryptography is based on a seemingly simple question: If you hide a secret point in such a lattice, how long will it take someone else to find the secret location starting from some other point? This game of hide and seek can underpin many ways to make data more secure.
A variant of the lattice problem called “learning with errors” is considered to be too hard to break even on a quantum computer. As the size of the lattice grows, the amount of time it takes to solve is believed to increase exponentially, even for a quantum computer.
The lattice problem—like the problem of finding the factors of a large number on which so much current encryption depends—is closely related to a deep open problem in mathematics called the “hidden subgroup problem.”
Yilei Chen’s approach suggested quantum computers may be able to solve lattice-based problems more quickly under certain conditions. Experts scrambled to check his results—and rapidly found an error. After the error was discovered, Chen published an updated version of his paper describing the flaw.
Despite this discovery, Chen’s paper has made many cryptographers less confident in the security of lattice-based methods. Some are still assessing whether Chen’s ideas can be extended to new pathways for attacking these methods.
More Mathematics RequiredChen’s paper set off a storm in the small community of cryptographers who are equipped to understand it. However, it received almost no attention in the wider world—perhaps because so few people understand this kind of work or its implications.
Last year, when the Australian government published a national quantum strategy to make the country “a leader of the global quantum industry” where “quantum technologies are integral to a prosperous, fair and inclusive Australia,” there was an important omission: It didn’t mention mathematics at all.
Australia does have many leading experts in quantum computing and quantum information science. However, making the most of quantum computers—and defending against them—will require deep mathematical training to produce new knowledge and research.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: ZENG YILI / Unsplash
Blood transfusions save lives. In the US alone, people receive around 10 million units each year. But blood banks are always short in supply—especially when it comes to the “universal donor” type O.
Surprisingly, the gut microbiome may hold a solution for boosting universal blood supplies by chemically converting other blood types into the universal O.
Infusing the wrong blood type—say, type A to type B—triggers deadly immune reactions. Type O blood, however, is compatible with nearly everyone. It’s in especially high demand following hurricanes, earthquakes, wildfires, and other crises because doctors have to rapidly treat as many people as possible.
Sometimes, blood banks have an imbalance of different blood types—for example, too much type A, not enough universal O. This week, a team from Denmark and Sweden discovered a cocktail of enzymes that readily converts type A and type B blood into the universal donor. Found in gut bacteria, the enzymes chew up an immune-stimulating sugar molecule dotted on the surfaces of type A and B blood cells, removing their tendency to spark an immune response.
Compared to previous attempts, the blend of enzymes converted A and B blood types to type O blood with “remarkably high efficiencies,” the authors wrote.
Wardrobe ChangeBlood types can be characterized in multiple ways, but roughly speaking, the types come in four main forms: A, B, AB, and O.
These types are distinguished by what kinds of sugar molecules—called antigens—cover the surfaces of red blood cells. Antigens can trigger immune rejection if mismatched. Type A blood has A antigens; type B has B antigens; type AB has both. Type O has neither.
This is why type O blood can be used for most people. It doesn’t normally trigger an immune response and is highly coveted during emergencies when it’s difficult to determine a person’s blood type. One obvious way to boost type O stock is to recruit more donors, but that’s not always possible. As a workaround, scientists have tried to artificially produce type O blood using stem cell technology. While successful in the lab, it’s expensive and hard to scale up for real-world demands.
An alternative is removing the A and B antigens from donated blood. First proposed in the 1980s, this approach uses enzymes to break down the immune-stimulating sugar molecules. Like licking an ice cream cone, as the antigens gradually melt away, the blood cells are stripped of their A or B identity, eventually transforming into the universal O blood type.
The technology sounds high-tech, but breaking down sugars is something our bodies naturally do every day, thanks to microbes in the gut that happily digest our food. This got scientists wondering: Can we hunt down enzymes in the digestive track to convert blood types?
Over a half decade ago, a team from the University of British Columbia made headlines by using bacterial enzymes found in the gut microbiome to transform type A blood to type O. Some gut bugs eat away at mucus—a slimy substance made of sugary molecules covering the gut. These mucus linings are molecularly similar to the antigens on red blood cells.
So, digestive enzymes from gut microbes could potentially chomp away A and B antigens.
In one test, the team took samples of human poop (yup), which carry enzymes from the gut microbiome and looked for DNA that could break down red blood cell sugar chains.
They eventually discovered two enzymes from a single bacterial strain. Tested in human blood, the duo readily stripped away type A antigens, converting it into universal type O.
The study was a proof of concept for transforming one blood type into another, with potentially real-world implications. Type A blood—common in Europe and the US—makes up roughly one-third of the supply of donations. A technology that converts it to universal O could boost blood transplant resources in this part of the world.
“This is a first, and if these data can be replicated, it is certainly a major advance,” Dr. Harvey Klein at the National Institutes of Health’s Clinical Center, who was not involved in the work, told Science at the time.
There’s one problem though. Converted blood doesn’t always work.
Let’s Talk ABO+When tested in clinical trials, converted blood has raised safety concerns. Even when removing A or B antigens completely from donated blood, small hints from earlier studies found an immune mismatch between the transformed donor blood and the recipient. In other words, the engineered O blood sometimes still triggered an immune response.
Why?
There’s more to blood types than classic ABO. Type A is composed of two different subtypes—one with higher A antigen levels than the other. Type B, common in people of Asian and African descent, also comes in “extended” forms. These recently discovered sugar chains are longer and harder to break down than in the classic versions. Called “extended antigens,” they could be why some converted blood still stimulates the immune system after transfusion.
The new study tackled these extended forms by again peeking into gut bacteria DNA. One bacterial strain, A. muciniphila, stood out. These bugs contain enzymes that work like a previously discovered version that chops up type A and B antigens, but surprisingly, they also strip away extended versions of both antigens.
These enzymes weren’t previously known to science, with just 30 percent similarity when compared to a previous benchmark enzyme that cuts up B and extended B antigens.
Using cells from different donors, the scientists engineered an enzyme soup that rapidly wiped out blood antigens. The strategy is “unprecedented,” wrote the team.
Although the screen found multiple enzymes capable of blood type conversion, each individually had limited effects. But when mixed and matched, the recipe transformed donated B type cells into type O, with limited immune responses when mixed with other blood types.
A similar strategy yielded three different enzymes to cut out the problematic A antigen and, in turn, transform the blood to type O. Some people secrete the antigen into other bodily fluids—for example, saliva, sweat, or tears. Others, dubbed non-secreters, have less of these antigens floating around their bodies. Using blood donated from both secreters and non-secreters, the team treated red blood cells to remove the A antigen and its extended versions.
When mixed with other blood types, the enzyme cocktail lowered their immune response, although with lower efficacy than cells transformed from type B to O.
By mapping the structures of these enzymes, the team found some parts increased their ability to chop up sugar chains. Focusing on these hot-spot structures, scientists are set to hunt down other naturally-derived enzymes—or use AI to engineer ones with better efficacy and precision.
The system still needs to be tested in humans. And the team didn’t address other blood antigens, such as the Rh system, which is what makes blood types positive or negative. Still, bacterial enzymes appear to be an unexpected but promising way to engineer universal blood.
Image Credit: Zeiss Microscopy / Flickr
ARTIFICIAL INTELLIGENCEMeta’s Open Source Llama 3 Is Already Nipping at OpenAI’s Heels
Will Knight | Wired“OpenAI changed the world with ChatGPT, setting off a wave of AI investment and drawing more than 2 million developers to its cloud APIs. But if open source models prove competitive, developers and entrepreneurs may decide to stop paying to access the latest model from OpenAI or Google and use Llama 3 or one of the other increasingly powerful open source models that are popping up.”
BIOTECH‘Real Hope’ for Cancer Cure as Personal mRNA Vaccine for Melanoma Trialed
Andrew Gregory | The Guardian“Experts are testing new jabs that are custom-built for each patient and tell their body to hunt down cancer cells to prevent the disease ever coming back. A phase 2 trial found the vaccines dramatically reduced the risk of the cancer returning in melanoma patients. Now a final, phase 3, trial has been launched and is being led by University College London Hospitals NHS Foundation Trust (UCLH). Dr Heather Shaw, the national coordinating investigator for the trial, said the jabs had the potential to cure people with melanoma and are being tested in other cancers, including lung, bladder and kidney.”
DIGITAL MEDIAAn AI Startup Made a Hyperrealistic Deepfake of Me That’s So Good It’s Scary
Melissa Heikkilä | MIT Technology Review“Until now, all AI-generated videos of people have tended to have some stiffness, glitchiness, or other unnatural elements that make them pretty easy to differentiate from reality. Because they’re so close to the real thing but not quite it, these videos can make people feel annoyed or uneasy or icky—a phenomenon commonly known as the uncanny valley. Synthesia claims its new technology will finally lead us out of the valley.”
ENERGYNuclear Fusion Experiment Overcomes Two Key Operating Hurdles
Matthew Sparkes | New Scientist“A nuclear fusion reaction has overcome two key barriers to operating in a ‘sweet spot’ needed for optimal power production: boosting the plasma density and keeping that denser plasma contained. The milestone is yet another stepping stone towards fusion power, although a commercial reactor is still probably years away.”
FUTUREDaniel Dennett: ‘ Why Civilization Is More Fragile Than We Realized’
Tom Chatfield | BBC“[Dennett’s] warning was not of a takeover by some superintelligence, but of a threat he believed that nonetheless could be existential for civilization, rooted in the vulnerabilities of human nature. ‘If we turn this wonderful technology we have for knowledge into a weapon for disinformation,’ he told me, ‘we are in deep trouble.’ Why? ‘Because we won’t know what we know, and we won’t know who to trust, and we won’t know whether we’re informed or misinformed. We may become either paranoid and hyper-skeptical, or just apathetic and unmoved. Both of those are very dangerous avenues. And they’re upon us.'”
ENVIRONMENTCalifornia Just Went 9.25 Hours Using Only Renewable Energy
Adele Peters | Fast Company“Last Saturday, as 39 million Californians went about their daily lives—taking showers, doing laundry, or charging their electric cars—the whole state ran on 100% clean electricity for more than nine hours. The same thing happened on Sunday, as the state was powered without fossil fuels for more than eight hours. It was the ninth straight day that solar, wind, hydropower, geothermal, and battery storage fully powered the electric grid for at least some portion of the time. Over the last six and a half weeks, that’s happened nearly every day. In some cases, it’s just for 15 minutes. But often it’s for hours at a time.”
archive pa
TECHAI Hype Is Deflating. Can AI Companies Find a Way to Turn a Profit?
Gerrit De Vynck | The Washington Post“Some once-promising start-ups have cratered, and the suite of flashy products launched by the biggest players in the AI race—OpenAI, Microsoft, Google and Meta—have yet to upend the way people work and communicate with one another. While money keeps pouring into AI, very few companies are turning a profit on the tech, which remains hugely expensive to build and run. The road to widespread adoption and business success is still looking long, twisty and full of roadblocks, say tech executives, technologists and financial analysts.”
ARTIFICIAL INTELLIGENCEApple Releases Eight Small AI Language Models Aimed at On-Device Use
Benj Edwards | Ars Technica“In the world of AI, what might be called ‘small language models’ have been growing in popularity recently because they can be run on a local device instead of requiring data center-grade computers in the cloud. On Wednesday, Apple introduced a set of tiny source-available AI language models called OpenELM that are small enough to run directly on a smartphone. They’re mostly proof-of-concept research models for now, but they could form the basis of future on-device AI offerings from Apple.”
SPACE*If Starship Is Real, We’re Going to Need Big Cargo Movers on the Moon and Mars
Eric Berger | Ars Technica*“Unloading tons of cargo on the Moon may seem like a preposterous notion. During Apollo, mass restrictions were so draconian that the Lunar Module could carry two astronauts, their spacesuits, some food, and just 300 pounds (136 kg) of scientific payload down to the lunar surface. By contrast, Starship is designed to carry 100 tons, or more, to the lunar surface in a single mission. This is an insane amount of cargo relative to anything in spaceflight history, but that’s the future that [Jaret] Matthews is aiming toward.”
Image Credit: CARTIST / Unsplash
Genomics is revolutionizing medicine and science, but current approaches still struggle to capture the breadth of human genetic diversity. Pangenomes that incorporate many people’s DNA could be the answer, and a new project thinks quantum computers will be a key enabler.
When the Human Genome Project published its first reference genome in 2001, it was based on DNA from just a handful of humans. While less than one percent of our DNA varies from person to person, this can still leave important gaps and limit what we can learn from genomic analyses.
That’s why the concept of a pangenome has become increasingly popular. This refers to a collection of genomic sequences from many different people that have been merged to cover a much greater range of human genetic possibilities.
Assembling these pangenomes is tricky though, and their size and complexity make carrying out computational analyses on them daunting. That’s why the University of Cambridge, the Wellcome Sanger Institute, and the European Molecular Biology Laboratory’s European Bioinformatics Institute have teamed up to see if quantum computers can help.
“We’ve only just scratched the surface of both quantum computing and pangenomics,” David Holland of the Wellcome Sanger Institute said in a press release. “So to bring these two worlds together is incredibly exciting. We don’t know exactly what’s coming, but we see great opportunities for major new advances.”
Pangenomes could be crucial for discovering how different genetic variants impact human biology, or that of other species. The current reference genome is used as a guide to assemble genetic sequences, but due to the variability of human genomes there are often significant chunks of DNA that don’t match up. A pangenome would capture a lot more of that diversity, making it easier to connect the dots and giving us a more complete view of possible human genomes.
Despite their power, pangenomes are difficult to work with. While the genome of a single person is just a linear sequence of genetic data, a pangenome is a complex network that tries to capture all the ways in which its constituent genomes do and don’t overlap.
These so-called “sequence graphs” are challenging to construct and even more challenging to analyze. And it will require high levels of computational power and novel techniques to make use of the rich representation of human diversity contained within.
That’s where this new project sees quantum computers lending a hand. Relying on the quirks of quantum mechanics, they can tackle certain computational problems that are near impossible for classical computers.
While there’s still considerable uncertainty about what kinds of calculations quantum computers will actually be able to run, many hope they will dramatically improve our ability to solve problems relating to complex systems with large numbers of variables. This new project is aimed at developing quantum algorithms that speed up both the production and analysis of pangenomes, though the researchers admit it’s early days.
“We’re starting from scratch because we don’t even know yet how to represent a pangenome in a quantum computing environment,” David Yuan from the European Bioinformatics Institute said in the press release. “If you compare it to the first moon landings, this project is the equivalent of designing a rocket and training the astronauts.”
The project has been awarded $3.5 million, which will be used to develop new algorithms and then test them on simulated quantum hardware using supercomputers. The researchers think the tools they develop could lead to significant breakthroughs in personalized medicine. They could also be applied to pangenomes of viruses and bacteria, improving our ability to track and manage disease outbreaks.
Given its exploratory nature and the difficulty of getting quantum computers to do anything practical, it could be some time before the project bears fruit. But if they succeed, the researchers could significantly expand our ability to make sense of the genes that shape our lives.
Image Credit: Gerd Altmann / Pixabay
CRISPR has revolutionized science. AI is now taking the gene editor to the next level.
Thanks to its ability to accurately edit the genome, CRISPR tools are now widely used in biotechnology and across medicine to tackle inherited diseases. In late 2023, a therapy using the Nobel Prize-winning tool gained approval from the FDA to treat sickle cell disease. CRISPR has also enabled CAR T cell therapy to battle cancers and been used to lower dangerously high cholesterol levels in clinical trials.
Outside medicine, CRISPR tools are changing the agricultural landscape, with projects ongoing to engineer hornless bulls, nutrient-rich tomatoes, and livestock and fish with more muscle mass.
Despite its real-world impact, CRISPR isn’t perfect. The tool snips both strands of DNA, which can cause dangerous mutations. It also can inadvertently nip unintended areas of the genome and trigger unpredictable side effects.
CRISPR was first discovered in bacteria as a defense mechanism, suggesting that nature hides a bounty of CRISPR components. For the past decade, scientists have screened different natural environments—for example, pond scum—to find other versions of the tool that could potentially increase its efficacy and precision. While successful, this strategy depends on what nature has to offer. Some benefits, such as a smaller size or greater longevity in the body, often come with trade-offs like lower activity or precision.
Rather than relying on evolution, can we fast-track better CRISPR tools with AI?
This week, Profluent, a startup based in California, outlined a strategy that uses AI to dream up a new universe of CRISPR gene editors. Based on large language models—the technology behind the popular ChatGPT—the AI designed several new gene-editing components.
In human cells, the components meshed to reliably edit targeted genes. The efficiency matched classic CRISPR, but with far more precision. The most promising editor, dubbed OpenCRISPR-1, could also precisely swap out single DNA letters—a technology called base editing—with an accuracy that rivals current tools.
“We demonstrate the world’s first successful editing of the human genome using a gene editing system where every component is fully designed by AI,” wrote the authors in a blog post.
Match Made in HeavenCRISPR and AI have had a long romance.
The CRISPR recipe has two main parts: A “scissor” Cas protein that cuts or nicks the genome and a “bloodhound” RNA guide that tethers the scissor protein to the target gene.
By varying these components, the system becomes a toolbox, with each setup tailored to perform a specific type of gene editing. Some Cas proteins cut both strands of DNA; others give just one strand a quick snip. Alternative versions can also cut RNA, a type of genetic material found in viruses, and can be used as diagnostic tools or antiviral treatments.
Different versions of Cas proteins are often found by searching natural environments or through a process called direct evolution. Here, scientist rationally swap out some parts of the Cas protein to potentially boost efficacy.
It’s a highly time-consuming process. Which is where AI comes in.
Machine learning has already helped predict off-target effects in CRISPR tools. It’s also homed in on smaller Cas proteins to make downsized editors easier to deliver into cells.
Profluent used AI in a novel way: Rather than boosting current systems, they designed CRISPR components from scratch using large language models.
The basis of ChatGPT and DALL-E, these models launched AI into the mainstream. They learn from massive amounts of text, images, music, and other data to distill patterns and concepts. It’s how the algorithms generate images from a single text prompt—say, “unicorn with sunglasses dancing over a rainbow”—or mimic the music style of a given artist.
The same technology has also transformed the protein design world. Like words in a book, proteins are strung from individual molecular “letters” into chains, which then fold in specific ways to make the proteins work. By feeding protein sequences into AI, scientists have already fashioned antibodies and other functional proteins unknown to nature.
“Large generative protein language models capture the underlying blueprint of what makes a natural protein functional,” wrote the team in the blog post. “They promise a shortcut to bypass the random process of evolution and move us towards intentionally designing proteins for a specific purpose.”
Do AIs Dream of CRISPR Sheep?All large language models need training data. The same is true for an algorithm that generates gene editors. Unlike text, images, or videos that can be easily scraped online, a CRISPR database is harder to find.
The team first screened over 26 terabytes of data about current CRISPR systems and built a CRISPR-Cas atlas—the most extensive to date, according to the researchers.
The search revealed millions of CRISPR-Cas components. The team then trained their ProGen2 language model—which was fine-tuned for protein discovery—using the CRISPR atlas.
The AI eventually generated four million protein sequences with potential Cas activity. After filtering out obvious deadbeats with another computer program, the team zeroed in on a new universe of Cas “protein scissors.”
The algorithm didn’t just dream up proteins like Cas9. Cas proteins come in families, each with its own quirks in gene-editing ability. The AI also designed proteins resembling Cas13, which targets RNA, and Cas12a, which is more compact than Cas9.
Overall, the results expanded the universe of potential Cas proteins nearly five-fold. But do any of them work?
Hello, CRISPR WorldFor the next test, the team focused on Cas9, because it’s already widely used in biomedical and other fields. They trained the AI on roughly 240,000 different Cas9 protein structures from multiple types of animals, with the goal of generating similar proteins to replace natural ones—but with higher efficacy or precision.
The initial results were surprising: The generated sequences, roughly a million of them, were totally different than natural Cas9 proteins. But using DeepMind’s AlphaFold2, a protein structure prediction AI, the team found the generated protein sequences could adopt similar shapes.
Cas proteins can’t function without a bloodhound RNA guide. With the CRISPR-Cas atlas, the team also trained AI to generate an RNA guide when given a protein sequence.
The result is a CRISPR gene editor with both components—Cas protein and RNA guide— designed by AI. Dubbed OpenCRISPR-1, its gene editing activity was similar to classic CRISPR-Cas9 systems when tested in cultured human kidney cells. Surprisingly, the AI-generated version slashed off-target editing by roughly 95 percent.
With a few tweaks, OpenCRISPR-1 could also perform base editing, which can change single DNA letters. Compared to classic CRISPR, base editing is likely more precise as it limits damage to the genome. In human kidney cells, OpenCRISPR-1 reliably converted one DNA letter to another in three sites across the genome, with an editing rate similar to current base editors.
To be clear, the AI-generated CRISPR tools have only been tested in cells in a dish. For treatments to reach the clinic, they’d need to undergo careful testing for safety and efficacy in living creatures, which can take a long time.
Profluent is openly sharing OpenCRISPR-1 with researchers and commercial groups but keeping the AI that created the tool in-house. “We release OpenCRISPR-1 publicly to facilitate broad, ethical usage across research and commercial applications,” they wrote.
As a preprint, the paper describing their work has yet to be analyzed by expert peer reviewers. Scientists will also have to show OpenCRISPR-1 or variants work in multiple organisms, including plants, mice, and humans. But tantalizingly, the results open a new avenue for generative AI—one that could fundamentally change our genetic blueprint.
Image Credit: Profluent
The origin of life on Earth is still enigmatic, but we are slowly unraveling the steps involved and the necessary ingredients. Scientists believe life arose in a primordial soup of organic chemicals and biomolecules on the early Earth, eventually leading to actual organisms.
It’s long been suspected that some of these ingredients may have been delivered from space. Now a new study, published in Science Advances, shows that a special group of molecules, known as peptides, can form more easily under the conditions of space than those found on Earth. That means they could have been delivered to the early Earth by meteorites or comets—and that life may be able to form elsewhere, too.
The functions of life are upheld in our cells (and those of all living beings) by large, complex carbon-based (organic) molecules called proteins. How to make the large variety of proteins we need to stay alive is encoded in our DNA, which is itself a large and complex organic molecule.
However, these complex molecules are assembled from a variety of small and simple molecules such as amino acids—the so-called building blocks of life.
To explain the origin of life, we need to understand how and where these building blocks form and under what conditions they spontaneously assemble themselves into more complex structures. Finally, we need to understand the step that enables them to become a confined, self-replicating system—a living organism.
This latest study sheds light on how some of these building blocks might have formed and assembled and how they ended up on Earth.
Steps to LifeDNA is made up of about 20 different amino acids. Like letters of the alphabet, these are arranged in DNA’s double helix structure in different combinations to encrypt our genetic code.
Peptides are also an assemblage of amino acids in a chain-like structure. Peptides can be made up of as little as two amino acids, but also range to hundreds of amino acids.
The assemblage of amino acids into peptides is an important step because peptides provide functions such as catalyzing, or enhancing, reactions that are important to maintaining life. They are also candidate molecules that could have been further assembled into early versions of membranes, confining functional molecules in cell-like structures.
However, despite their potentially important role in the origin of life, it was not so straightforward for peptides to form spontaneously under the environmental conditions on the early Earth. In fact, the scientists behind the current study had previously shown that the cold conditions of space are actually more favorable to the formation of peptides.
The interstellar medium. Image Credit: Charles Carter/Keck Institute for Space StudiesIn the very low density clouds of molecules and dust particles in a part of space called the interstellar medium (see above), single atoms of carbon can stick to the surfaces of dust grains together with carbon monoxide and ammonia molecules. They then react to form amino acid-like molecules. When such a cloud becomes denser and dust particles also start to stick together, these molecules can assemble into peptides.
In their new study, the scientists look at the dense environment of dusty disks, from which a new solar system with a star and planets emerges eventually. Such disks form when clouds suddenly collapse under the force of gravity. In this environment, water molecules are much more prevalent—forming ice on the surfaces of any growing agglomerates of particles that could inhibit the reactions that form peptides.
By emulating the reactions likely to occur in the interstellar medium in the laboratory, the study shows that, although the formation of peptides is slightly diminished, it is not prevented. Instead, as rocks and dust combine to form larger bodies such as asteroids and comets, these bodies heat up and allow for liquids to form. This boosts peptide formation in these liquids, and there’s a natural selection of further reactions resulting in even more complex organic molecules. These processes would have occurred during the formation of our own solar system.
Many of the building blocks of life such as amino acids, lipids, and sugars can form in the space environment. Many have been detected in meteorites.
Because peptide formation is more efficient in space than on Earth, and because they can accumulate in comets, their impacts on the early Earth might have delivered loads that boosted the steps towards the origin of life on Earth.
So, what does all this mean for our chances of finding alien life? Well, the building blocks for life are available throughout the universe. How specific the conditions need to be to enable them to self-assemble into living organisms is still an open question. Once we know that, we’ll have a good idea of how widespread, or not, life might be.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Aldebaran S / Unsplash
From Covid boosters to annual flu shots, most of us are left wondering: Why so many, so often?
There’s a reason to update vaccines. Viruses rapidly mutate, which can help them escape the body’s immune system, putting previously vaccinated people at risk of infection. Using AI modeling, scientists have increasingly been able to predict how viruses will evolve. But they mutate fast, and we’re still playing catch up.
An alternative strategy is to break the cycle with a universal vaccine that can train the body to recognize a virus despite mutation. Such a vaccine could eradicate new flu strains, even if the virus has transformed into nearly unrecognizable forms. The strategy could also finally bring a vaccine for the likes of HIV, which has so far notoriously evaded decades of efforts.
This month, a team from UC California Riverside, led by Dr. Shou-Wei Ding, designed a vaccine that unleashed a surprising component of the body’s immune system against invading viruses.
In baby mice without functional immune cells to ward off infections, the vaccine defended against lethal doses of a deadly virus. The protection lasted at least 90 days after the initial shot.
The strategy relies on a controversial theory. Most plants and fungi have an innate defense against viruses that chops up their genetic material. Called RNA interference (RNAi), scientists have long debated whether the same mechanism exists in mammals—including humans.
“It’s an incredible system because it can be adapted to any virus,” Dr. Olivier Voinnet at the Swiss Federal Institute of Technology, who championed the theory with Ding, told Nature in late 2013.
A Hidden RNA UniverseRNA molecules are usually associated with the translation of genes into proteins.
But they’re not just biological messengers. A wide array of small RNA molecules roam our cells. Some shuttle protein components through the cell during the translation of DNA. Others change how DNA is expressed and may even act as a method of inheritance.
But fundamental to immunity are small interfering RNA molecules, or siRNAs. In plants and invertebrates, these molecules are vicious defenders against viral attacks. To replicate, viruses need to hijack the host cell’s machinery to copy their genetic material—often, it’s RNA. The invaded cells recognize the foreign genetic material and automatically launch an attack.
During this attack, called RNA interference, the cell chops the invading viruses’ RNA genome into tiny chunks–siRNA. The cell then spews these viral siRNA molecules into the body to alert the immune system. The molecules also directly grab onto the invading viruses’ genome, blocking it from replicating.
Here’s the kicker: Vaccines based on antibodies usually target one or two locations on a virus, making them vulnerable to mutation should those locations change their makeup. RNA interference generates thousands of siRNA molecules that cover the entire genome—even if one part of a virus mutates, the rest is still vulnerable to the attack.
This powerful defense system could launch a new generation of vaccines. There’s just one problem. While it’s been observed in plants and flies, whether it exists in mammals has been highly controversial.
“We believe that RNAi has been antiviral for hundreds of millions of years,” Ding told Nature in 2013. “Why would we mammals dump such an effective defense?”
Natural Born Viral KillersIn the 2013 study in Science, Ding and colleagues suggested mammals also have an antiviral siRNA mechanism—it’s just being repressed by a gene carried by most viruses. Dubbed B2, the gene acts like a “brake,” smothering any RNA interference response from host cells by destroying their ability to make siRNA snippets.
Getting rid of B2 should kick RNA interference back into gear. To prove the theory, the team genetically engineered a virus without a functioning B2 gene and tried to infect hamster cells and immunocompromised baby mice. Called Nodamura virus, it’s transmitted by mosquitoes in the wild and is often deadly.
But without B2, even a lethal dose of the virus lost its infectious power. The baby mice rapidly generated a hefty dose of siRNA molecules to clear out the invaders. As a result, the infection never took hold, and the critters—even when already immunocompromised—survived.
“I truly believe that the RNAi response is relevant to at least some viruses that infect mammals,” said Ding at the time.
New-Age VaccinesMany vaccines contain either a dead or a living but modified version of a virus to train the immune system. When faced with the virus again, the body produces T cells to kill off the target, B cells that pump out antibodies, and other immune “memory” cells to alert against future attacks. But their effects don’t always last, especially if a virus mutates.
Rather than rallying T and B cells, triggering the body’s siRNA response offers another type of immune defense. This can be done by deleting the B2 gene in live viruses. These viruses can be formulated into a new type of vaccine, which the team has been working to develop, relying on RNA interference to ward off invaders. The resulting flood of siRNA molecules triggered by the vaccine would, in theory, also provide some protection against future infection.
“If we make a mutant virus that cannot produce the protein to suppress our RNAi [RNA interference], we can weaken the virus. It can replicate to some level, but then loses the battle to the host RNAi response,” Ding said in a press release about the most recent study. “A virus weakened in this way can be used as a vaccine for boosting our RNAi immune system.”
In the study, his team tried the strategy against Nodamura virus by removing its B2 gene.
The team vaccinated baby and adult mice, both of which were genetically immunocompromised in that they couldn’t mount T cell or B cell defenses. In just two days, the single shot fully protected the mice against a deadly dose of virus, and the effect lasted over three months.
Viruses are most harmful to vulnerable populations—infants, the elderly, and immunocompromised individuals. Because of their weakened immune systems, current vaccines aren’t always as effective. Triggering siRNA could be a life-saving alternative strategy.
Although it works in mice, whether humans respond similarly remains to be seen. But there’s much to look forward to. The B2 “brake” protein has also been found in lots of other common viruses, including dengue, flu, and a family of viruses that causes fever, rash, and blisters.
The team is already working on a new flu vaccine, using live viruses without the B2 protein. If successful, the vaccine could potentially be made as a nasal spray—forget the needle jab. And if their siRNA theory holds up, such a vaccine might fend off the virus even as it mutates into new strains. The playbook could also be adapted to tackle new Covid variants, RSV, or whatever nature next throws at us.
This vaccine strategy is “broadly applicable to any number of viruses, broadly effective against any variant of a virus, and safe for a broad spectrum of people,” study author Dr. Rong Hai said in the press release. “This could be the universal vaccine that we have been looking for.”
Image Credit: Diana Polekhina / Unsplash
ARTIFICIAL INTELLIGENCE15 Graphs That Explain the State of AI in 2024
Eliza Strickland | IEEE Spectrum“Each year, the AI Index lands on virtual desks with a louder virtual thud—this year, its 393 pages are a testament to the fact that AI is coming off a really big year in 2023. For the past three years, IEEE Spectrum has read the whole damn thing and pulled out a selection of charts that sum up the current state of AI.”
NEUROSCIENCEThe Next Frontier for Brain Implants Is Artificial Vision
Emily Mullin | Wired“Elon Musk’s Neuralink and others are developing devices that could provide blind people with a crude sense of sight. …’This is not about getting biological vision back,’ says Philip Troyk, a professor of biomedical engineering at Illinois Tech, who’s leading the study Bussard is in. ‘This is about exploring what artificial vision could be.'”
DIGITAL MEDIAMicrosoft’s VASA-1 Can Deepfake a Person With One Photo and One Audio Track
Benj Edwards | Ars Technica“On Tuesday, Microsoft Research Asia unveiled VASA-1, an AI model that can create a synchronized animated video of a person talking or singing from a single photo and an existing audio track. In the future, it could power virtual avatars that render locally and don’t require video feeds—or allow anyone with similar tools to take a photo of a person found online and make them appear to say whatever they want.”
TECHMeta Is Already Training a More Powerful Successor to Llama 3
Will Knight | Wired“On Thursday morning, Meta released its latest artificial intelligence model, Llama 3, touting it as the most powerful to be made open source so that anyone can use it. The same afternoon, Yann LeCun, Meta’s chief AI scientist, said an even more powerful successor to Llama is in the works. He suggested it could potentially outshine the world’s best closed AI models, including OpenAI’s GPT-4 and Google’s Gemini.”
COMPUTINGIntel Reveals World’s Biggest ‘Brain-Inspired’ Neuromorphic Computer
Matthew Sparkes | New Scientist“Hala Point contains 1.15 billion artificial neurons across 1152 Loihi 2 achips, and is capable of 380 trillion synaptic operations per second. Mike Davies at Intel says that despite this power it occupies just six racks in a standard server case—a space similar to that of a microwave oven. Larger machines will be possible, says Davies. ‘We built this scale of system because, honestly, a billion neurons was a nice round number,’ he says. ‘I mean, there wasn’t any particular technical engineering challenge that made us stop at this level.'”
AUTOMATIONUS Air Force Confirms First Successful AI Dogfight
Emma Roth | The Verge“Human pilots were on board the X-62A with controls to disable the AI system, but DARPA says the pilots didn’t need to use the safety switch ‘at any point.’ The X-62A went against an F-16 controlled solely by a human pilot, where both aircraft demonstrated ‘high-aspect nose-to-nose engagements’ and got as close as 2,000 feet at 1,200 miles per hour. DARPA doesn’t say which aircraft won the dogfight, however.”
CULTUREWhat If Your AI Girlfriend Hated You?
Kate Knibbs | Wired“It seems as though we’ve arrived at the moment in the AI hype cycle where no idea is too bonkers to launch. This week’s eyebrow-raising AI project is a new twist on the romantic chatbot—a mobile app called AngryGF, which offers its users the uniquely unpleasant experience of getting yelled at via messages from a fake person.”
NEUROSCIENCEInsects and Other Animals Have Consciousness, Experts Declare
Dan Falk | Quanta“For decades, there’s been a broad agreement among scientists that animals similar to us—the great apes, for example—have conscious experience, even if their consciousness differs from our own. In recent years, however, researchers have begun to acknowledge that consciousness may also be widespread among animals that are very different from us, including invertebrates with completely different and far simpler nervous systems.”
SCIENCETwo Lifeforms Merge in Once-in-a-Billion-Years Evolutionary Event
Michael Irving | New Atlas“Scientists have caught a once-in-a-billion-years evolutionary event in progress, as two lifeforms have merged into one organism that boasts abilities its peers would envy. Last time this happened, Earth got plants. …A species of algae called Braarudosphaera bigelowii was found to have engulfed a cyanobacterium that lets them do something that algae, and plants in general, can’t normally do—’fixing’ nitrogen straight from the air, and combining it with other elements to create more useful compounds.”
Image Credit: Shubham Dhage / Unsplash
Dubbed “living drugs,” CAR T cells are bioengineered from a patient’s own immune cells to make them better able to hunt and destroy cancer.
The treatment is successfully tackling previously untreatable blood cancers. Six therapies are already approved by the FDA. Over a thousand clinical trials are underway. These aren’t limited to cancer—they cover a range of difficult medical problems such as autoimmune diseases, heart conditions, and viral infections including HIV. They may even slow down the biological processes that contribute to aging.
But CAR T has an Achilles heel.
Once injected into the body, the cells often slowly dwindle. Called “exhaustion,” this process erodes therapeutic effect over time and has dire medical consequences. According to Dr. Evan Weber at the University of Pennsylvania, more than 50 percent of people who respond to CAR T therapies eventually relapse. This may also be why CAR T cells have struggled to fight off solid tumors in breast, pancreatic, or deadly brain cancers.
This month, two teams found a potential solution—make CAR T cells more like stem cells. Known for their regenerative abilities, stem cells easily repopulate the body. Both teams identified the same protein “master switch” to make engineered cells resemble stem cells.
One study, led by Weber, found that adding the protein, called FOXO1, revved up metabolism and health in CAR T cells in mice. Another study from a team at the Peter MacCallum Cancer Center in Australia found FOXO1-boosted cells appeared genetically similar to immune stem cells and were better able to fend off solid tumors.
While still early, “these findings may help improve the design of CAR T cell therapies and potentially benefit a wider range of patients,” said Weber in a press release.
I RememberHere’s how CAR T cell therapy usually works.
The approach focuses on T cells, a particular type of immune cell that naturally hunts downs and eliminates infections and cancers inside the body. Enemy cells are dotted with a specific set of proteins, a kind of cellular fingerprint, that T cells recognize and latch onto.
Tumors also have a unique signature. But they can be sneaky, with some eventually developing ways to evade immune surveillance. In solid cancers, for example, they can pump out chemicals that fight off immune cell defenders, allowing the cancer to grow and spread.
CAR T cells are designed to override these barriers.
To make them, medical practitioners remove T cells from the body and genetically engineer them to produce tailormade protein hooks targeting a particular protein on tumor cells. The supercharged T cells are then grown in petri dishes and transfused back into the body.
In the beginning, CAR T was a last-resort blood cancer treatment, but now it’s a first-line therapy. Keeping the engineered cells around inside the body, however, has been a struggle. With time, the cells stop dividing and become dysfunctional, potentially allowing the cancer to relapse.
The TranslatorTo tackle cell exhaustion, Weber’s team found inspiration in the body itself.
Our immune system has a cellular ledger tracking previous infections. The cells making up this ledger are called memory T cells. They’re a formidable military reserve, a portion of which resemble stem cells. When the immune system detects an invader it’s seen before—a virus, bacteria, or cancer cell—these reserve cells rapidly proliferate to fend off the attack.
CAR T cells don’t usually have this ability. Inside multiple cancers, they eventually die off—allowing cancers to return. Why?
In 2012, Dr. Crystal Mackall at Stanford University found several changes in gene expression that lead to CAR T cell exhaustion. In the new study, together with Weber, the team discovered a protein, FOXO1, that could lengthen CAR T’s effects.
In one test, a drug that inhibited FOXO1 caused CAR T cells to rapidly fail and eventually die in petri dishes. Erasing genes encoding FOXO1 also hindered the cells and increased signs of CAR T exhaustion. When infused into mice with leukemia, CAR T cells without FOXO1 couldn’t treat the cancer. By contrast, increasing levels of FOXO1 helped the cells readily fight it off.
Analyzing genes related to FOXO1, the team found they were mostly connected to immune cell memory. It’s likely that adding the gene encoding FOXO1 to CAR T cells promotes a stable memory for the cells, so they can easily recognize potential harm—be it cancer or pathogen—long after the initial infection.
When treating mice with leukemia, a single dose of the FOXO1-enhanced cells decreased cancer growth and increased survival up to five-fold compared to standard CAR T therapy. The enhanced treatment also tackled a type of bone cancer in mice, which is often hard to treat without surgery and chemotherapy.
An Immune LinkMeanwhile, the Australian team also zeroed in on FOXO1. Led by Drs. Junyun Lai, Paul Beavis, and Phillip Darcy, the team was looking for protein candidates to enhance CAR T longevity.
The idea was, like their natural counterparts, engineered CAR T cells also need a healthy metabolism to thrive and divide.
They started by analyzing a protein previously shown to enhance CAR T metabolism, potentially lowering the chances of exhaustion. Mapping the epigenome and transcriptome in CAR T cells—both of which tell us how genes are expressed—they also discovered FOXO1 regulating CAR T cell longevity.
As a proof of concept, the team induced exhaustion in the engineered cells by increasingly restricting their ability to divide.
In mice with cancer, cells supercharged with FOXO1 lasted longer by months than those that hadn’t been boosted. The critters’ liver and kidney functions remained normal, and they didn’t lose weight during the treatment, a marker of overall health. The FOXO1 boost also changed how genes were expressed in the cells—they looked younger, as if in a stem cell-like state.
The new recipe also worked in T cells donated by six people with cancer who had undergone standard CAR T therapy. Adding a dose of FOXO1 to these cells increased their metabolism.
Multiple CAR T clinical trials are ongoing. But “the effects of such cells are transient and do not provide long-term protection against exhaustion,” wrote Darcy and team. In other words, durability is key for CAR T cells to live up to their full potential.
A FOXO1 boost offers a way—although it may not be the only way.
“By studying factors that drive memory in T cells, like FOXO1, we can enhance our understanding of why CAR T cells persist and work more effectively in some patients compared to others,” said Weber.
Image Credit: Gerardo Sotillo, Stanford Medicine
Graphene has been hailed as a wonder material, but it also set off a rush to find other promising atomically thin materials. Now researchers have managed to create a 2D version of gold they call “goldene,” which could have a host of applications in chemistry.
Scientists had speculated about the possibility of creating layers of carbon just a single atom thick for many decades. But it wasn’t until 2004 that a team from the University of Manchester in the UK first produced graphene sheets using the remarkably simple technique of peeling them off a lump of graphite with common sticky tape.
The resulting material’s high strength, high conductivity, and unusual optical properties set off a stampede to find applications for it. But it also spurred researchers to investigate what kinds of exotic capabilities other ultra-thin materials could have.
Gold is one material scientists have long been eager to make as thin as graphene, but so far, efforts have been in vain. Now though, researchers from Linköping University in Sweden have borrowed from an old Japanese forging technique to create ultra-thin flakes of what they’re calling “goldene.”
“If you make a material extremely thin, something extraordinary happens,” Shun Kashiwaya, who led the research, said in a press release. “The same thing happens with gold.”
Making goldene has proven tough in the past because its atoms tend to clump together. So, even if you can create a 2D sheet of gold atoms they quickly roll up to create nanoparticles instead.
The researchers got around this by taking a ceramic called titanium silicon carbide, which features ultra-thin layers of silicon between layers of titanium carbide, and coating it with gold. They then heated it in a furnace, which caused the gold to diffuse into the material and replace the silicon layers in a process known as intercalation.
This created atomically thin layers of gold embedded in the ceramic. To get it out, they had to borrow a century-old technique developed by Japanese knife makers. They used a chemical formulation known as Murakami’s reagent, which etches away carbon residue, to slowly reveal the gold sheets.
The researchers had to experiment with different concentrations of the reagent and various etching times. They also had to add a detergent-like chemical called a surfactant that protected the gold sheets from the etching liquid and prevented them from curling up. The gold flakes could then be sieved out of the solution to be examined more closely.
In a paper in Nature Synthesis, the researchers describe how they used an electron microscope to confirm that the gold layers were indeed just one atom thick. They also showed that the goldene flakes were semiconductors.
It’s not the first time someone has claimed to have created goldene, notes Nature. But previous attempts have involved creating the ultra-thin sheets sandwiched between other materials, and the Linköping team say their effort is the first to create a “free-standing 2D metal.”
The material could have a range of use cases, the researchers say. Gold nanoparticles already show promise as catalysts that can turn plastic waste and biomass into valuable materials, they note in their paper, and they have properties that could prove useful for energy harvesting, creating photonic devices, or even splitting water to create hydrogen fuel.
It will take work to tweak the synthesis method so it can produce commercially useful amounts of the material, a challenge that has delayed the full arrival of graphene as a widely used product too. But the team is also investigating whether similar approaches can be applied to other useful catalytic metals. Graphene might not be the only wonder material in town for long.
Image Credit: Nature Synthesis (CC BY 4.0)
Yesterday, Boston Dynamics announced it was retiring its hydraulic Atlas robot. Atlas has long been the standard bearer of advanced humanoid robots. Over the years, the company was known as much for its research robots as it was for slick viral videos of them working out in military fatigues, forming dance mobs, and doing parkour. Fittingly, the company put together a send-off video of Atlas’s greatest hits and blunders.
But there were clues this wasn’t really the end, not least of which was the specific inclusion of the word “hydraulic” and the last line of the video, “‘Til we meet again, Atlas.” It wasn’t a long hiatus. Today, the company released hydraulic Atlas’s successor—electric Atlas.
The new Atlas is notable for several reasons. First, and most obviously, Boston Dynamics has finally done away with hydraulic actuators in favor of electric motors. To be clear, Atlas has long had an onboard battery pack—but now it’s fully electric. The advantages of going electric include less cost, noise, weight, and complexity. It also allows for a more polished design. From the company’s own Spot robot to a host of other humanoid robots, fully electric models are the norm these days. So, it’s about time Atlas made the switch.
Without a mess of hydraulic hoses to contend with, the new Atlas can now also contort itself in new ways. As you’ll note in the release video, the robot rises to its feet—a crucial skill for a walking robot—in a very, let’s say, special way. It folds its legs up along its torso and impossibly, for a human at least, pivots up through its waist (no hands). Once standing Atlas swivels its head 180 degrees, then does the same thing at each hip joint and the waist. It takes a few watches to really appreciate all the weirdness there.
The takeaway is that while Atlas looks like us, it’s capable of movements we aren’t and therefore has more flexibility in how it completes future tasks.
This theme of same-but-different is evident in its head too. Instead of opting for a human-like head that risks slipping into the uncanny valley, the team chose a featureless (for now) lighted circle. In an interview with IEEE Spectrum, Boston Dynamics CEO, Robert Playter, said the human-like designs they tried seemed “a little bit threatening or dystopian.”
“We’re trying to project something else: a friendly place to look to gain some understanding about the intent of the robot,” he said. “The design borrows from some friendly shapes that we’d seen in the past. For example, there’s the old Pixar lamp that everybody fell in love with decades ago, and that informed some of the design for us.”
While most of these upgrades are improvements, there is one area where it’s not totally clear how well the new form will fair: strength and power.
Hydraulics are known to provide both, and Atlas pushed its hydraulics to their limits carrying heavy objects, executing backflips, and doing 180-degree, in-air twists. According to the press release and Playter’s interviews, little has been lost in this category. In fact, they say, electric Atlas is stronger than hydraulic Atlas. Still, as with all things robotics, the ultimate proof of how capable it is will likely be in video form, which we’ll eagerly await.
Despite big design updates, the company’s messaging is perhaps more notable. Atlas used to be a research robot. Now, the company intends to sell them commercially.
This isn’t terribly surprising. There are now a number of companies competing in the humanoid robots space, including Agility, 1X, Tesla, Apptronik, and Figure—which just raised $675 million at a $2.6 billion valuation. Several are making rapid progress, with a heavy focus on AI, and have kicked off real-world pilots.
Where does Boston Dynamics fit in? With Atlas, the company has been the clear leader for years. So, it’s not starting from the ground floor. Also, thanks to its Spot and Stretch robots, the company already has experience commercializing and selling advanced robots, from identifying product-market fit to dealing with logistics and servicing. But AI was, until recently, less of a focus. Now, they’re folding reinforcement learning into Spot, have begun experimenting with generative AI too, and promise more is coming.
Hyundai acquired Boston Dynamics for $1.1 billion in 2021. This may prove advantageous, as they have access to a world-class manufacturing company along with its resources and expertise producing and selling machines at scale. It’s also an opportunity to pilot Atlas in real-world situations and perfect it for future customers. Plans are already in motion to put Atlas to work at Hyundai next year.
Still, it’s worth noting that, although humanoid robots are attracting attention, getting big time investment, and being tried out in commercial contexts, there’s likely a ways to go before they reach the kind of generality some companies are touting. Playter says Boston Dynamics is going for multi-purpose, but still niche, robots in the near term.
“It definitely needs to be a multi-use case robot. I believe that because I don’t think there’s very many examples where a single repetitive task is going to warrant these complex robots,” he said. “I also think, though, that the practical matter is that you’re going to have to focus on a class of use cases, and really making them useful for the end customer.”
Humanoid robots that tidy your house and do the dishes may not be imminent, but the field is hot, and AI is bringing a degree of generality not possible a year ago. Now that Boston Dynamics has thrown its name in the hat, things will only get more interesting from here. We’ll be keeping a close eye on YouTube to see what new tricks Atlas has up its sleeve.
Image Credit: Boston Dynamics
Stars like the sun are remarkably constant. They vary in brightness by only 0.1 percent over years and decades, thanks to the fusion of hydrogen into helium that powers them. This process will keep the sun shining steadily for about 5 billion more years, but when stars exhaust their nuclear fuel, their deaths can lead to pyrotechnics.
The sun will eventually die by growing large and then condensing into a type of star called a white dwarf. But stars over eight times more massive than the sun die violently in an explosion called a supernova.
Supernovae happen across the Milky Way only a few times a century, and these violent explosions are usually remote enough that people here on Earth don’t notice. For a dying star to have any effect on life on our planet, it would have to go supernova within 100 light years from Earth.
I’m an astronomer who studies cosmology and black holes.
In my writing about cosmic endings, I’ve described the threat posed by stellar cataclysms such as supernovae and related phenomena such as gamma-ray bursts. Most of these cataclysms are remote, but when they occur closer to home they can pose a threat to life on Earth.
The Death of a Massive StarVery few stars are massive enough to die in a supernova. But when one does, it briefly rivals the brightness of billions of stars. At one supernova per 50 years, and with 100 billion galaxies in the universe, somewhere in the universe a supernova explodes every hundredth of a second.
The dying star emits high-energy radiation as gamma rays. Gamma rays are a form of electromagnetic radiation with wavelengths much shorter than light waves, meaning they’re invisible to the human eye. The dying star also releases a torrent of high-energy particles in the form of cosmic rays: subatomic particles moving at close to the speed of light.
Supernovae in the Milky Way are rare, but a few have been close enough to Earth that historical records discuss them. In 185 AD, a star appeared in a place where no star had previously been seen. It was probably a supernova.
Observers around the world saw a bright star suddenly appear in 1006 AD. Astronomers later matched it to a supernova 7,200 light years away. Then, in 1054 AD, Chinese astronomers recorded a star visible in the daytime sky that astronomers subsequently identified as a supernova 6,500 light years away.
Johannes Kepler, the astronomer who observed what was likely a supernova in 1604. Image Credit: Kepler-Museum in Weil der StadtJohannes Kepler observed the last supernova in the Milky Way in 1604, so in a statistical sense, the next one is overdue.
At 600 light years away, the red supergiant Betelgeuse in the constellation of Orion is the nearest massive star getting close to the end of its life. When it goes supernova, it will shine as bright as the full moon for those watching from Earth, without causing any damage to life on our planet.
Radiation DamageIf a star goes supernova close enough to Earth, the gamma-ray radiation could damage some of the planetary protection that allows life to thrive on Earth. There’s a time delay due to the finite speed of light. If a supernova goes off 100 light years away, it takes 100 years for us to see it.
Astronomers have found evidence of a supernova 300 light years away that exploded 2.5 million years ago. Radioactive atoms trapped in seafloor sediments are the telltale signs of this event. Radiation from gamma rays eroded the ozone layer, which protects life on Earth from the sun’s harmful radiation. This event would have cooled the climate, leading to the extinction of some ancient species.
Safety from a supernova comes with greater distance. Gamma rays and cosmic rays spread out in all directions once emitted from a supernova, so the fraction that reach the Earth decreases with greater distance. For example, imagine two identical supernovae, with one 10 times closer to Earth than the other. Earth would receive radiation that’s about a hundred times stronger from the closer event.
A supernova within 30 light years would be catastrophic, severely depleting the ozone layer, disrupting the marine food chain and likely causing mass extinction. Some astronomers guess that nearby supernovae triggered a series of mass extinctions 360 to 375 million years ago. Luckily, these events happen within 30 light years only every few hundred million years.
When Neutron Stars CollideBut supernovae aren’t the only events that emit gamma rays. Neutron star collisions cause high-energy phenomena ranging from gamma rays to gravitational waves.
Left behind after a supernova explosion, neutron stars are city-size balls of matter with the density of an atomic nucleus, so 300 trillion times denser than the sun. These collisions created many of the gold and precious metals on Earth. The intense pressure caused by two ultradense objects colliding forces neutrons into atomic nuclei, which creates heavier elements such as gold and platinum.
A neutron star collision generates an intense burst of gamma rays. These gamma rays are concentrated into a narrow jet of radiation that packs a big punch.
If the Earth were in the line of fire of a gamma-ray burst within 10,000 light years, or 10 percent of the diameter of the galaxy, the burst would severely damage the ozone layer. It would also damage the DNA inside organisms’ cells, at a level that would kill many simple life forms like bacteria.
That sounds ominous, but neutron stars do not typically form in pairs, so there is only one collision in the Milky Way about every 10,000 years. They are 100 times rarer than supernova explosions. Across the entire universe, there is a neutron star collision every few minutes.
Gamma-ray bursts may not hold an imminent threat to life on Earth, but over very long time scales, bursts will inevitably hit the Earth. The odds of a gamma-ray burst triggering a mass extinction are 50 percent in the past 500 million years and 90 percent in the 4 billion years since there has been life on Earth.
By that math, it’s quite likely that a gamma-ray burst caused one of the five mass extinctions in the past 500 million years. Astronomers have argued that a gamma-ray burst caused the first mass extinction 440 million years ago, when 60 percent of all marine creatures disappeared.
A Recent ReminderThe most extreme astrophysical events have a long reach. Astronomers were reminded of this in October 2022, when a pulse of radiation swept through the solar system and overloaded all of the gamma-ray telescopes in space.
It was the brightest gamma-ray burst to occur since human civilization began. The radiation caused a sudden disturbance to the Earth’s ionosphere, even though the source was an explosion nearly two billion light years away. Life on Earth was unaffected, but the fact that it altered the ionosphere is sobering—a similar burst in the Milky Way would be a million times brighter.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA, ESA, Joel Kastner (RIT)
Are we alone? This question is nearly as old as humanity itself. Today, the question in astronomy focuses on finding life beyond our planet. Are we, as a species, and as a planet, alone? Or is there life somewhere else? Usually the question inspires visions of weird, green versions of humans. However, life is more […]
Forget sperm meets egg. Using human stem cells, scientists have created human embryo-like structures inside petri dishes. These lab-grown blobs develop multiple structures that mimic a human embryo after implantation into the uterus—a major milestone for fertility—and last at least 14 days. A decade ago, manufacturing embryo-like structures, or embryoids, without reproductive cells would have […]
Of all the forms of human intellect that one might expect artificial intelligence to emulate, few people would likely place creativity at the top of their list. Creativity is wonderfully mysterious—and frustratingly fleeting. It defines us as human beings—and seemingly defies the cold logic that lies behind the silicon curtain of machines. Yet, the use […]
ARTIFICIAL INTELLIGENCEWhat OpenAI Really Wants
Steven Levy | Wired“For Altman and his company, ChatGPT and GPT-4 are merely stepping stones along the way to achieving a simple and seismic mission, one these technologists may as well have branded on their flesh. That mission is to build artificial general intelligence—a concept that’s so far been grounded more in science fiction than science—and to make it safe for humanity. The people who work at OpenAI are fanatical in their pursuit of that goal.”
COMPUTINGThe Secret to Nvidia’s AI Success
Samuel K. Moore | IEEE Spectrum“[Nvidia] has managed to increase the performance of its chips on AI tasks a thousandfold over the past 10 years, it’s raking in money, and it’s reportedly very hard to get your hands on its newest AI-accelerating GPU, the H100. How did Nvidia get here? …Moore’s Law was a surprisingly small part of Nvidia’s magic and new number formats a very large part. Put it all together and you get what Dally called Huang’s Law (for Nvidia CEO Jensen Huang).”
VIRTUAL REALITYRoblox’s New AI Chatbot Will Help You Build Virtual Worlds
Jay Peters | The Verge“The new tool, the Roblox Assistant, builds on previously announced features that let creators build virtual assets and write code with the help of generative AI. …Down the line, Roblox has bigger visions for Roblox Assistant, and Sturman teased that it could generate sophisticated gameplay and even make 3D models from scratch. If that all works, it could bring Roblox in line with CEO David Baszucki’s vision of Westworld-like ease of design.”
TECHApple Is Reportedly Spending ‘Millions of Dollars a Day’ Training AI
Monica Chin | The Verge“The company is reportedly working on multiple AI models across several teams. Apple’s unit that works on conversational AI is called ‘Foundational Models,’ per The Information’s reporting. It has ‘around 16’ members, including several former Google engineers. Additional teams at Apple are also working on artificial intelligence, per The Information. A Visual Intelligence unit is developing an image generation model, and another group is researching ‘multimodal AI, which can recognize and produce images or video as well as text.'”
SPACESpaceX Broke Its Record for Number of Launches in a Year
Stephen Clark | Ars Technica“SpaceX is leading the world not just in the number of launches, but also in the total payload mass the company has launched into orbit this year. In the first half of 2023, SpaceX delivered about 447 metric tons of cargo into orbit, roughly 80 percent of all the material launched into orbit worldwide, according to data from the space analytics firm BryceTech. Musk said SpaceX will launch about 90 percent of the world’s total payload mass into orbit next year, based on the company’s launch manifest for 2024.”
ARTRefik Anadol Just Turned the Las Vegas Sphere Into the World’s Largest AI Artwork
Jesus Diaz | Fast Company“With Sphere, the building is the canvas—a bland engineering marvel that transforms into something visually arresting once Anadol gets his hands on it. ‘I think this is one of the most Blade Runner moments ever,’ he says. ‘A science fiction moment that, finally, merges media arts and architecture, embedding technology into a physical environment that exists in the real world.'”
BIOTECHRedwire Space Prints Human Knee Cartilage in Space for the First Time
Aria Alamalhodaei | TechCrunch“Redwire Space has successfully ‘bioprinted’ a human knee meniscus aboard the International Space Station, a landmark development that could help people recovering from meniscus injuries here on Earth. The meniscus cartilage was printed on Redwire’s BioFabrication Facility (BFF) on the ISS. …After the BFF printed the meniscus with living human cells, it was transferred to Redwire’s Advanced Space Experiment Processor for a 14-day enculturation process. After the culture process was complete, the meniscus was packaged up and sent back to Earth aboard SpaceX’s Crew-6 mission.”
AUTOMATIONFAA Clears UPS Delivery Drones for Longer-Range Flights
Sheena Vasani | The Verge“UPS Flight Forward, a UPS subsidiary focused on drone delivery, can now deliver small packages beyond the visual line of sight (BVLOS) without spotters on the ground monitoring the route and skies for other aircraft, using Matternet M2 drones. The FAA also announced authorizations for two other companies to fly beyond sight for commercial purposes.”
Image Credit: Marek Piwnicki / Unsplash
Renewable power was already rapidly replacing fossil fuels as the cheapest source of electricity. Thanks to rocketing fuel prices last year, it is now the clear winner when it comes to cost-effectiveness.
For decades, solar and wind power was substantially more expensive than fossil fuels and most projects were heavily reliant on government subsidies to survive. But rapidly falling costs mean renewables now match or even outperform traditional power sources in a wide range of markets.
That transition has now accelerated significantly, according to a new report from the International Renewable Energy Agency (IRENA). Thanks in large part to a major spike in fossil fuel prices, 86 percent of newly commissioned, grid-scale renewable electricity capacity in 2022 had lower costs than fossil-fuel-derived electricity. That’s despite all kinds of costs having gone up across the world due to rising inflation and disruption to supply chains caused by the Covid pandemic and war in Ukraine.
“IRENA sees 2022 as a veritable turning point in the deployment for renewables as its cost-competitiveness has never been greater despite the lingering commodity and equipment cost inflation around the world,” IRENA’s director-general Francesco La Camera said in a press release.
The findings are just the latest data point showing the dramatic fall in prices renewables have experienced in recent years. According to the report, in 2010 solar power was 710 percent more expensive than the cheapest fossil fuel option, while onshore wind was 95 percent more expensive.
Last year, the average cost of electricity from solar fell by 3 percent to almost one-third less than the cheapest fossil fuel globally, while onshore wind costs fell by 5 percent to slightly less than half that of the cheapest fossil fuel option.
Cost declines weren’t evenly distributed though, the report notes. The significant improvements in both solar and onshore wind were both driven by deployments in China. If the Asian giant had been excluded from the calculations, the average cost of onshore wind would have remained level. And countries like France, Germany, and Greece experienced significant increases in the cost of solar.
The costs of offshore wind projects and hydropower projects also both increased in 2022. The former saw a 2 percent rise due to a drop in China’s rate of deployment, while the latter saw costs jump 18 percent due to overruns in a number of large projects.
Nonetheless, the report found the combined renewable power capacity deployed around the world since the year 2000 saved roughly $521 billion in fuel costs in 2022. The authors suggest the rapid build-out of green energy in recent years probably prevented the spike in fossil fuel prices from developing into an all-out energy crisis last year, highlighting the energy security benefits of renewables.
“The most affected regions by the historic price shock were remarkably resilient, in large part thanks to the massive increase of solar and wind in the last decade,” said La Camera.
Even in places where renewable installation costs increased, the report says that fossil fuel prices typically rose by far more. With those prices expected to remain high for, the authors conclude that this will cement a structural change in the energy market with renewables becoming the cheapest source of power globally.
Whether this shift in cost dynamics will be enough to avert the climate crisis remains to be seen. La Camera notes that annual deployments of renewable power need to hit 1,000 gigawatts every year until 2030 if we want to keep alive the goal of limiting global warming to 1.5 degrees Celsius. That’s an ambitious goal that will need all the help it can get from market forces.
Image Credit: Chelsea / Unsplash
In late 2020, AI pioneer DeepMind achieved a breakthrough 50 years in the making. By predicting the shape of proteins with atomic accuracy, its deep learning algorithm, AlphaFold, all but solved one of biology’s grand challenges. From metabolism to brain function, proteins are the molecules that make our bodies go. When they go wrong, things […]
This summer, the AI division of Mark Zuckerberg’s Meta unveiled its Llama 2 chatbot. Microsoft has been appointed as Meta’s preferred partner on Llama 2, which will be available through the Windows operating system.
Meta’s approach with Llama 2 contrasts with that of the company OpenAI, which created the AI chatbot ChatGPT. That’s because Meta has made its product open source—meaning that the original code is freely available, allowing it to be researched and modified.
This strategy has sparked a vast wave of discussions. Will it foster greater public scrutiny and regulation of large language models (LLMs)—the technology that underlies AI chatbots such as Llama 2 and ChatGPT? Could it inadvertently empower criminals to use the technology to help them carry out phishing attacks or develop malware? And could the move help Meta gain an advantage over OpenAI and Google in this fast-moving field?
Whatever happens, this strategic move looks set to reshape the current landscape of generative AI. In February 2023, Meta released its first version of the LLM, called Llama, but made it available for academic use only. Its updated version, Llama 2, features improved performance and is more suitable for business use.
Like other AI chatbots, Llama 2 had to be trained using online data. Exposure to this vast resource of information helps it improve what it does—providing users with useful responses to their questions.
An initial version of Llama 2 was created through “supervised fine-tuning,” a technique that uses high-quality question-and-answer data to calibrate it for use by the public. It was further refined with human feedback reinforcement learning which, as the name suggests, incorporates people’s assessments of the AI’s performance to align it with human preferences.
Guaranteed BenefitsMeta’s embrace of the open-source ethos with Llama 2 allows it to capitalize on what appears to be an approach that has worked for the company in the past. Meta’s engineers are known for their development of products to aid developers such as React and PyTorch. Both are open source and have become the industry standard. Through them, Meta has set a precedent of innovation through collaboration.
The release of Llama 2 holds the promise of safer generative AI. Through shared wisdom and collective exploration, users can identify erroneous information and any vulnerabilities that could be exploited by criminals. Unexpected applications have already emerged, such as a version of Llama 2 that can be installed on iPhones and was created by users, underscoring the potential for creativity within this community.
But there are limits to how far Meta will allow Llama 2 users to commercialize its AI system. If any party achieves more than 700 million active users in the preceding calendar month for a product based on Llama 2, it will have to request a license from Meta. For Meta, this opens up the potential for profit-sharing on successful products based on Llama 2.
Meta’s strategy contrasts starkly with the more guarded approach of its primary competitor, OpenAI. Even as some question Meta’s ability to compete in this area and commercialize products as OpenAI has done with ChatGPT, Meta’s decision to invite worldwide developers into the fold suggests a broader vision. It’s a move that positions Zuckerberg’s company not merely as a player but a facilitator, harnessing global talent to contribute to the growing ecosystem of Llama 2.
This strategy could also be a shrewd hedge against potential competition from fellow tech giants such as Google. With a large population of users exploring the potential of Llama 2, any successful advance can be promptly integrated into Meta’s other products. Only time will reveal the full impact of this decision, but the immediate effects on the industry are already resonating far and wide.
Advantages and Pitfalls for UsersThe public experimentation aspect of open source technology allows for greater scrutiny, providing an opportunity for a community of users to assess Llama 2’s strengths and weaknesses, including its vulnerability to attacks. The public’s watchful eye may reveal flaws in LLMs, prompting the development of defenses against them.
On the downside, concerns have emerged that this is akin to “handing a knife to criminals,” as it could also allow malicious users to exploit the technology. For example, its power could help fraudsters build a dialogue system that generates plausible automated conversations for telephone scams. This potential for misuse has led some to call for regulation of the technology.
But exactly what rules are devised, who gets the power to supervise this process, and exactly what needs more or less scrutiny, all require careful planning to make sure that regulation does not simply prop up monopolies for the big tech companies.
In the evolving saga of AI development, the debate over open sourcing serves as a reminder that technological advancements are rarely simple or one-dimensional. The implications of Meta’s decision are likely to ripple across the tech world for years to come. While Llama 2 may not yet rival the capabilities of ChatGPT, it opens the door to the development of a host of innovative products.
Google will also be under scrutiny, as speculation grows about how it may respond. In an era where open source culture thrives, it would not be surprising to see Google follow suit with its own releases.
The term “tech for good” has become a common mantra to describe technology companies using some of their resources to make a positive impact on all our lives. Ultimately, though, this objective remains a shared responsibility, not just something that a handful of companies should be engaged in.
It’s also an aim that demands collaboration and a concerted effort across academia, industry, and beyond. As LLM technologies continue to evolve, the stakes are high, and the path forward is laden with both opportunities and challenges.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Shawn Suttle / Pixabay
Switching from combustion-engine cars to electric vehicles is going to be an important part of the renewable energy transition. But for the switch to truly make a difference, the electricity powering the next generation of cars will also have to be all-green, and the grid is quite a ways from that reality. One small loophole is solar cars—the loophole is small because the technology isn’t advanced enough to put any sort of significant dent in the vehicles’ electricity need, but it’s a start.
There are a handful of solar cars in production, from the $6,800 golf-cart-like Squad car to the sleek $250,000 Lightyear 0 (and let’s not forget the OG Aptera and more recent arrival Sion). Now a Swedish manufacturer is taking the solar concept and going bigger with it—big-rig big, that is. Scania’s hybrid solar truck was tested on public roads for the first time last week.
One of the first vehicles of its kind, the truck is a research project involving both academia and industry, and its creators hope it will be a step toward reducing the trucking industry’s environmental footprint. It could also cut fuel costs for drivers and ultimately reduce the total cost of moving goods from one place to another.
Hybrid solar vehicles have a battery that can be plugged in to charge, but they’re also decked out with solar arrays that provide an alternate energy source. Scania’s truck has solar panels covering an area of 100 square meters (1,076 square feet), all on the sides and top of the 59-foot-long trailer. The panels were specially made for this project, and the team says they’re more lightweight and efficient than the current industry standard.
The truck has a 560 horsepower engine, and its solar array can provide up to 8,000 kilowatt hours (kWh) of energy per year in Sweden or similar climates—that equates to around 5,000 kilometers (3,107 miles) of driving range. Being as far north as it is, Sweden’s not the sunniest place; the researchers say that in sunnier climates (the example they give is Spain) the range could double, reaching about 6,200 miles a year.
The team is also working on developing tandem solar cells with an even higher efficiency, which they say could double the solar energy generation a second time over. The truck’s batteries have a total capacity of 300 kWh, 100 kWh placed on the truck and the remaining 200 kWh on the trailer.
Besides gauging how much solar energy the truck’s panels can produce under various conditions, the researchers are monitoring how much an average truck’s carbon emissions would decrease if outfitted with a comparable solar setup. They’re also looking at various ways solar trucks could interact with the power grid (bi-directional charging, where the truck’s panels could give energy back to the grid or help power a facility, could be one possibility), and what the impact on the grid might be in a future where there are many hybrid solar trucks.
It will likely be a while before we see solar-panel-clad big rigs rolling down highways; for one, solar technology will need to improve dramatically before it becomes practical for widespread use on cars and trucks. But projects like Scania’s are a start, pointing us towards a future of cleaner, greener transportation.
Image Credit: Scania
Recently I had what amounted to a therapy session with ChatGPT. We talked about a recurring topic that I’ve obsessively inundated my friends with, so I thought I’d spare them the déjà vu. As expected, the AI’s responses were on point, sympathetic, and felt so utterly human.
As a tech writer, I know what’s happening under the hood: a swarm of digital synapses are trained on an internet’s worth of human-generated text to spit out favorable responses. Yet the interaction felt so real, and I had to constantly remind myself I was chatting with code—not a conscious, empathetic being on the other end.
Or was I? With generative AI increasingly delivering seemingly human-like responses, it’s easy to emotionally assign a sort of “sentience” to the algorithm (and no, ChatGPT isn’t conscious). In 2021, Blake Lemoine at Google stirred up a media firestorm by proclaiming that one of the chatbots he worked on, LaMDA, was sentient—and he subsequently got fired.
But most deep learning models are loosely based on the brain’s inner workings. AI agents are increasingly endowed with human-like decision-making algorithms. The idea that machine intelligence could become sentient one day no longer seems like science fiction.
How could we tell if machine brains one day gained sentience? The answer may be based on our own brains.
A preprint paper authored by 19 neuroscientists, philosophers, and computer scientists, including Dr. Robert Long from the Center for AI Safety and Dr. Yoshua Bengio from the University of Montreal, argues that the neurobiology of consciousness may be our best bet. Rather than simply studying an AI agent’s behavior or responses—for example, during a chat—matching its responses to theories of human consciousness could provide a more objective ruler.
It’s an out-of-the-box proposal, but one that makes sense. We know we are conscious regardless of the word’s definition, which is still unsettled. Theories of how consciousness emerges in the brain are plenty, with multiple leading candidates still being tested in global head-to-head trials.
The authors didn’t subscribe to any single neurobiological theory of consciousness. Instead, they derived a checklist of “indicator properties” of consciousness based on multiple leading ideas. There isn’t a strict cutoff—say, meeting X number of criteria means an AI agent is conscious. Rather, the indicators make up a moving scale: the more criteria met, the more likely a sentient machine mind is.
Using the guidelines to test several recent AI systems, including ChatGPT and other chatbots, the team concluded that for now, “no current AI systems are conscious.”
However, “there are no obvious technical barriers to building AI systems that satisfy these indicators,” they said. It’s possible that “conscious AI systems could realistically be built in the near term.”
Listening to an Artificial BrainSince Alan Turing’s famous imitation game in the 1950s, scientists have pondered how to prove whether a machine exhibits intelligence like a human’s.
Better known as the Turing test, the theoretical setup has a human judge conversing with a machine and another human—the judge has to decide which participant has an artificial mind. At the heart of the test is the provocative question “Can machines think?” The harder it is to tell the difference between machine and human, the more machines have advanced toward human-like intelligence.
ChatGPT broke the Turing test. An example of a chatbot powered by a large language model (LLM), ChatGPT soaks up internet comments, memes, and other content. It’s extremely adept at emulating human responses—writing essays, passing exams, dishing out recipes, and even doling out life advice.
These advances, which came at a shocking speed, stirred up debate on how to construct other criteria for gauging thinking machines. Most recent attempts have focused on standardized tests for humans: for example, those designed for high school students, the Bar exam for lawyers, or the GRE for entering grad school. OpenAI’s GPT-4, the AI model behind ChatGPT, scored in the top 10 percent of participants. However, it struggled with finding rules for a relatively simple visual puzzle game.
The new benchmarks, while measuring a kind of “intelligence,” don’t necessarily tackle the problem of consciousness. Here’s where neuroscience comes in.
The Checklist for ConsciousnessNeurobiological theories of consciousness are many and messy. But at their heart is neural computation: that is, how our neurons connect and process information so it reaches the conscious mind. In other words, consciousness is the result of the brain’s computation, although we don’t yet fully understand the details involved.
This practical look at consciousness makes it possible to translate theories from human consciousness to AI. Called computational functionalism, the hypothesis rests on the idea that computations of the right kind generate consciousness regardless of the medium—squishy, fatty blobs of cells inside our head or hard, cold chips that power machine minds. It suggests that “consciousness in AI is possible in principle,” said the team.
Then comes the hard part: how do you probe consciousness in an algorithmic black box? A standard method in humans is to measure electrical pulses in the brain or with functional MRI that captures activity in high definition—but neither method is feasible for evaluating code.
Instead, the team took a “theory-heavy approach,” which was first used to study consciousness in non-human animals.
To start, they mined top theories of human consciousness, including the popular Global Workspace Theory (GWT) for indicators of consciousness. For example, GWT stipulates that a conscious mind has multiple specialized systems that work in parallel; we can simultaneously hear and see and process those streams of information. However, there’s a bottleneck in processing, requiring an attention mechanism.
The Recurrent Processing Theory suggests that information needs to feed back onto itself in multiple loops as a path towards consciousness. Other theories emphasize the need for a “body” of sorts that receives feedback from the environment and uses those learnings to better perceive and control responses to a dynamic outside world—something called “embodiment.”
With myriad theories of consciousness to choose from, the team laid out some ground rules. To be included, a theory needs substantial evidence from lab tests, such as studies capturing the brain activity of people in different conscious states. Overall, six theories met the mark. From there, the team developed 14 indicators.
It’s not one-and-done. None of the indicators mark a sentient AI on their own. In fact, standard machine learning methods can build systems that have individual properties from the list, explained the team. Rather, the list is a scale—the more criteria met, the higher the likelihood an AI system has some kind of consciousness.
How to assess each indicator? We’ll need to look into the “architecture of the system and how the information flows through it,” said Long.
In a proof of concept, the team used the checklist on several different AI systems, including the transformer-based large language models that underlie ChatGPT and algorithms that generate images, such as DALL-E 2. The results were hardly cut-and-dried, with some AI systems meeting a portion of the criteria while lacking in others.
However, although not designed with a global workspace in mind, each system “possesses some of the GWT indicator properties,” such as attention, said the team. Meanwhile, Google’s PaLM-E system, which injects observations from robotic sensors, met the criteria for embodiment.
None of the state-of-the-art AI systems checked off more than a few boxes, leading the authors to conclude that we haven’t yet entered the era of sentient AI. They further warned about the dangers of under-attributing consciousness in AI, which may risk allowing “morally significant harms,” and anthropomorphizing AI systems when they’re just cold, hard code.
Nevertheless, the paper sets guidelines for probing one of the most enigmatic aspects of the mind. “[The proposal is] very thoughtful, it’s not bombastic and it makes its assumptions really clear,” Dr. Anil Seth at the University of Sussex told Nature.
The report is far from the final word on the topic. As neuroscience further narrows down correlates of consciousness in the brain, the checklist will likely scrap some criteria and add others. For now, it’s a project in the making, and the authors invite other perspectives from multiple disciplines—neuroscience, philosophy, computer science, cognitive science—to further hone the list.
Image Credit: Greyson Joralemon on Unsplash
AUTOMATIONHigh-Speed AI Drone Beats World-Champion Racers for the First Time
Benj Edwards | Ars Technica“On Wednesday, a team of researchers from the University of Zürich and Intel announced that they have developed an autonomous drone system named Swift that can beat human champions in first-person view (FPV) drone racing. While AI has previously bested humans in games like chess, Go, and even StarCraft, this may be the first time an AI system has outperformed human pilots in a physical sport.”
ROBOTICS‘Go Catch That Squirrel:’ Google AI Teaches a Robo-Dog Conversational Commands
Mack DeGeurin | Gizmodo“What’s more interesting, [the researchers] noted, is SayTap’s ability to ‘process unstructured and vague instructions.’ By just providing the model with a brief hint, the researchers were able to successfully command the robotic dogs to jump up and down when it was told ‘we are going on a picnic.’ …In maybe the funniest example, the dog even slowly backpedaled after being told to get away from a squirrel. Many real dog owners would beg for that level of obedience.”
BIOTECHA Biotech Company Says It Put Dopamine-Making Cells Into People’s Brains
Antonio Regalado | MIT Technology Review“In an important test for stem-cell medicine, a biotech company says implants of lab-made neurons introduced into the brains of 12 people with Parkinson’s disease appear to be safe and may have reduced symptoms for some of them. …The study is one of the largest and most costly tests yet of embryonic-stem-cell technology, the controversial and much-hyped approach of using stem cells taken from IVF embryos to produce replacement tissue and body parts.”
TRANSPORTATIONAre Self-Driving Cars Already Safer Than Human Drivers?
Timothy B. Lee | Ars Technica“For this story, I read through every crash report Waymo and Cruise filed in California this year, as well as reports each company filed about the performance of their driverless vehicles (with no safety drivers) prior to 2023. …Human beings drive close to 100 million miles between fatal crashes, so it will take hundreds of millions of driverless miles for 100 percent certainty on this question. But the evidence for better-than-human performance is starting to pile up, especially for Waymo.”
ARTIFICIAL INTELLIGENCEWe Used AI to Write Essays for Harvard, Yale and Princeton. Here’s How It Went.
Natasha Singer | The New York Times“While the chatbots are not yet great at simulating long-form personal essays with authentic student voices, I wondered how the AI tools would do on some of the shorter essay questions that elite schools like Harvard, Yale, Princeton, and Dartmouth are requiring high school applicants to answer this year. So I used several free tools to generate short essays for some Ivy League applications.”
INNOVATIONAI Startup Buzz Is Facing a Reality Check
Berber Jin | The Wall Street Journal“Founders and venture capitalists who flocked to artificial-intelligence startups are learning that turning the chatbot buzz into successful businesses is harder than it seems. Almost a year into the boom ignited by the November launch of ChatGPT, some startups that epitomized the zeal for so-called generative AI are now navigating layoffs and reduced user interest. Investors are unsure whether the new crop of AI startups will be able to survive, especially as tech giants such as Microsoft and Alphabet’s Google solidify their dominance over the technology.”
COMPUTINGQuantum Computer Reveals Chemical Reaction in 100-Billionth-Speed Slow-Mo
Michael Irving | New Atlas“Using a trapped-ion quantum computer, the team mapped the problem onto a fairly small quantum device, which allowed them to slow down the process by an astonishing 100 billion times. …’In nature, the whole process is over within femtoseconds,’ said Vanessa Olaya Agudelo, co-lead author of the study. ‘Using our quantum computer, we built a system that allowed us to slow down the chemical dynamics from femtoseconds to milliseconds. This allowed us to make meaningful observations and measurements. This has never been done before.’i”
ETHICSOpenAI’s Moonshot: Solving the AI Alignment Problem
Eliza Strickland | IEEE Spectrum“In July, OpenAI announced a new research program on ‘superalignment.’ The program has the ambitious goal of solving the hardest problem in the field, known as AI alignment, by 2027, an effort to which OpenAI is dedicating 20 percent of its total computing power. …One of the project’s leaders Jan] Leike spoke to IEEE Spectrum about the effort, which has the subgoal of building an aligned AI research tool—to help solve the alignment problem.”
ENVIRONMENTSolar-Panel-Covered Hybrid Truck Offers 3,000 to 6,000 Free Miles a Year
Mike Hanlon | New Atlas“The initial 560-horsepower plug-in hybrid experimental truck has an 18-meter (59-foot) trailer that is covered by 100 square meters (1,076 square feet) of solar panels, giving it the equivalent solar-surface area of an average house equipped with similarly powerful 13.2-kilowatt-peak panels. The truck uses new, lightweight tandem solar cells, that are based on a combination of Midsummer’s solar cells and new perovskite solar cells, and generates an estimated 8,000 kWh annually when operated in Sweden.”
TECHThe End of the Googleverse
Ryan Broderick | The Verge“Google officially went online…in 1998. It quickly became so inseparable from both the way we use the internet and, eventually, culture itself, that we almost lack the language to describe what Google’s impact over the last 25 years has actually been. It’s like asking a fish to explain what the ocean is. And yet, all around us are signs that the era of ‘peak Google’ is ending or, possibly, already over.”
SCIENCEOnly 1,280 Reproductive Human Ancestors Once Roamed Earth, Gene Study Suggests
Isaac Schultz | Gizmodo“An ancestral human species faced a startling population bottleneck and teetered on the brink of extinction around 800,000 years ago, according to new research. [In an article commenting on the research, Nick Ashton, an archaeologist at the British Museum, and Chris Stringer, a paleoanthropologist at London’s Natural History Museum] wrote…’the provocative study of Hu et al. brings the vulnerability of early human populations into focus, with the implication that our evolutionary lineage was nearly eradicated.'”
Image Credit: Casey Horner / Unsplash
In May 2020, some unusual rocks containing distinctive greenish crystals were found in the Erg Chech sand sea, a dune-filled region of the Sahara Desert in southern Algeria.
On close inspection, the rocks turned out to be from outer space: lumps of rubble billions of years old, left over from the dawn of the solar system. They were all pieces of a meteorite known as Erg Chech 002, which is the oldest volcanic rock ever found, having melted long ago in the fires of some now-vanished ancient protoplanet.
In new research published in Nature Communications, we analyzed lead and uranium isotopes in Erg Chech 002 and calculated it is some 4.56556 billion years old, give or take 120,000 years. This is one of the most precise ages ever calculated for an object from space—and our results also cast doubt on some common assumptions about the early solar system.
The Secret Life of AluminumAround 4.567 billion years ago, our solar system formed from a vast cloud of gas and dust. Among the many elements in this cloud was aluminum, which came in two forms.
First is the stable form, aluminum-27. Second is aluminum-26, a radioactive isotope mainly produced by exploding stars, which decays over time into magnesium-26. Aluminum-26 is very useful stuff for scientists who want to understand how the solar system formed and developed. Because it decays over time, we can use it to date events—particularly within the first four or five million years of the solar system’s life.
The decay of aluminum-26 is also important for another reason: we think it was the main source of heat in the early solar system. This decay influenced the melting of the small, primitive rocks that later clumped together to form the planets.
Uranium, Lead, and AgeHowever, to use aluminum-26 to understand the past, we need to know whether it was spread around evenly or clumped together more densely in some places than in others. To figure that out, we will need to calculate the absolute ages of some ancient space rocks more precisely.
Looking at aluminum-26 alone won’t let us do that, because it decays relatively quickly (after around 705,000 years, half of a sample of aluminum-26 will have decayed into magnesium-26). It’s useful for determining the relative ages of different objects, but not their absolute age in years.
But if we combine aluminum-26 data with data about uranium and lead, we can make some headway. There are two important isotopes of uranium (uranium-235 and uranium-238), which decay into different isotopes of lead (lead-207 and lead-206, respectively). The uranium isotopes have much longer half-lives (710 million years and 4.47 billion years, respectively), which means we can use them to directly figure out how long ago an event happened.
Meteorite GroupsErg Chech 002 is what is known as an “ungrouped achondrite.” Achondrites are rocks formed from melted planetesimals, which is what we call solid lumps in the cloud of gas and debris that formed the solar system. The sources of many achondrites found on Earth have been identified.
Achondrite meteorites like Erg Chech 002 offer clues about the early years of the solar system. Image Credit: Yuri Amelin, CC BYMost belong to the so-called Howardite-Eucrite-Diogenite clan, which are believed to have originated from Vesta 4, one of the largest asteroids in the solar system. Another group of achondrites is called angrites, which all share an unidentified parent body.
Still other achondrites, including Erg Chech 002, are “ungrouped”: their parent bodies and family relationships are unknown.
A Clumpy Spread of AluminumIn our study of Erg Chech 002, we found it contains a high abundance of lead-206 and lead-207, as well as relatively large amounts of undecayed uranium-238 and uranium-235.
Measuring the ratios of all the lead and uranium isotopes was what helped us to estimate the age of the rock with such unprecedented accuracy. We also compared our calculated age with previously published aluminum-26 data for Erg Chech 002, as well as data for various other achondrites.
The comparison with a group of achondrites called volcanic angrites was particularly interesting. We found that the parent body of Erg Chech 002 must have formed from material containing three or four times as much aluminum-26 as the source of the angrites’ parent body. This shows aluminum-26 was indeed distributed quite unevenly throughout the cloud of dust and gas which formed the solar system.
Our results contribute to a better understanding of the solar system’s earliest developmental stages, and the geological history of burgeoning planets. Further studies of diverse achondrite groups will undoubtedly continue to refine our understanding and enhance our ability to reconstruct the early history of our solar system.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Steve Jurvetson / Wikimedia, CC BY-SA
The cells of all living organisms are powered by the same chemical fuel: adenosine triphosphate (ATP). Now, researchers have found a way to generate ATP directly from electricity, which could turbocharge biotechnology processes that grow everything from food to fuel to pharmaceuticals. Interfacing modern electronics-based technology with biology is notoriously difficult. One major stumbling block […]
As the world works to transition from fossil fuels to renewable energy sources, we’ll extract less oil and gas from the Earth and more minerals like lithium, cobalt, and nickel. Demand for these materials has skyrocketed in the last few years, and will only continue to grow as we implement more solar panels, electric cars, […]
After over two decades, the human genome sequence is finally complete. The holdout? The Y chromosome. Although far smaller than the other 23 chromosomes, Y is a genetic contortionist, carrying multiple strange structures that are notoriously difficult to disentangle, and not for lack of effort. As one of the two sex chromosomes—X being the other—Y […]
French novelist Jules Verne delighted 19th-century readers with the tantalizing notion that a journey to the center of the Earth was actually plausible. Since then, scientists have long acknowledged that Verne’s literary journey was only science fiction. The extreme temperatures of the Earth’s interior—around 10,000 degrees Fahrenheit (5,537 Celsius) at the core—and the accompanying crushing […]
One of the biggest stumbling blocks for quantum computers is their tendency to be error-prone and the massive computational overhead required to clean up their mistakes. IBM has now made a breakthrough by dramatically reducing the number of qubits required to do so. All computers are prone to errors, and even the computer chip in […]
ARTIFICIAL INTELLIGENCE Meta Is Building a Space-Age ‘Universal Language Translator’ Alex Blake | Digital Trends “When you think of tools infused with artificial intelligence (AI) these days, it’s natural for ChatGPT and Bing Chat to spring to mind. But Facebook owner Meta wants to change that with SeamlessM4T, an AI-powered ‘universal language translator’ that could instantly convert any […]
ChatGPT, DALL-E, Stable Diffusion, and other generative AIs have taken the world by storm. They create fabulous poetry and images. They’re seeping into every nook of our world, from marketing to writing legal briefs and drug discovery. They seem like the poster child for a man-machine mind meld success story. But under the hood, things […]
3D printing is becoming more popular as a construction method, with multiple companies building entire 3D-printed neighborhoods in various parts of the world. But the technique has come under scrutiny, with critics saying it’s not nearly as cost-effective nor environmentally friendly as advocates claim. A Japanese company called Serendix is hoping to be a case […]
Psychedelics are known for inducing altered states of consciousness in humans by fundamentally changing our normal patterns of sensory perception, thought, and emotion. Research into the therapeutic potential of psychedelics has increased significantly in the last decade. While this research is important, I have always been more intrigued by the idea that psychedelics can be […]
We often think of proteins as immutable 3D sculptures. That’s not quite right. Many proteins are transformers that twist and change their shapes depending on biological needs. One configuration may propagate damaging signals from a stroke or heart attack. Another may block the resulting molecular cascade and limit harm. In a way, proteins act like […]
Stem cells are special kinds of cells in our bodies that can become any other type of cell. They have huge potential for medicine, and trials are currently under way using stem cells to replace damaged cells in diseases like Parkinson’s. One way to get stem cells is from human embryos, but this has ethical […]
Pulling large amounts of carbon dioxide (CO2) out of the atmosphere is likely to be a crucial part of efforts to tackle climate change. A new $1.2 billion investment by the US government in two large-scale facilities could help jumpstart the technology. While there is strong consensus that rapidly reducing carbon emissions will be essential […]
COMPUTING Scientists Recreate Pink Floyd Song by Reading Brain Signals of Listeners Hana Kiros | The New York Times “Scientists have trained a computer to analyze the brain activity of someone listening to music and, based only on those neuronal patterns, recreate the song. The research, published on Tuesday, produced a recognizable, if muffled version of […]
All right people, let’s do this one more time. If you’ve been following longevity research, you already know the story: young blood, either directly infused or injected into old mice, restores multiple organ functions to their younger selves and lengthens lives. Muscles regenerate. Heart tissues regain their strength. Brain regions critical for learning and memory […]
Everyone knows that arithmetic is true: 2 + 2 = 4. But surprisingly, we don’t know why it’s true. By stepping outside the box of our usual way of thinking about numbers, my colleagues and I have recently shown that arithmetic has biological roots and is a natural consequence of how perception of the world […]
Generative AI has found a host of uses since it exploded into the public consciousness with the release of ChatGPT last year. From drafting essays for students to helping customer service agents field calls to writing code for software engineers, it seems there’s no end to the ways these tools can make life a little […]
Mammals have approximately the same genes. Yet from skittering lab mice to magnificent bowhead whales or the elegant elephant, the difference in lifespan can be more than a century. Why? An international consortium is decoding the mystery. Rather than comparing different genetic letters between species, they turned the focus to gene expression—that is, how genes […]
Nocturnal predators have an ingrained superpower: even in pitch-black darkness, they can easily survey their surroundings, honing in on tasty prey hidden among a monochrome landscape. Hunting for your next supper isn’t the only perk of seeing in the dark. Take driving down a rural dirt road on a moonless night. Trees and bushes lose […]
The universe we live in is a transparent one, where light from stars and galaxies shines bright against a clear, dark backdrop. But this wasn’t always the case—in its early years, the universe was filled with a fog of hydrogen atoms that obscured light from the earliest stars and galaxies. The intense ultraviolet light from […]
Fusion power has long been seen as a pipe dream, but in recent years the technology has appeared to be edging closer to reality. The second demonstration of a fusion reaction that creates more power than it uses is another important marker suggesting fusion’s time may be coming. Generating power by smashing together atoms holds […]
The word swarm often carries negative connotations—think biblical plagues of locusts or high streets full of last-minute shoppers during the Christmas rush. However, swarming is essential for the survival of many animal collectives. And now research into swarming has the potential to change things for humans too. Bees swarm to make their search for new […]
Swiss startup Energy Vault came out of stealth mode in 2018, and has been on an upward trajectory since then. The company created a system to store electricity by elevating concrete blocks, and investors quickly jumped on board: Energy Vault raised $110 million from the SoftBank Vision Fund in 2019, and another $100 million led […]
The tiny, floating blobs of mini-hearts were straight out of Frankenstein. Made from a mixture of human stem cells and a sprinkle of silicon nanowires, the cyborg heart organoids bizarrely pumped away as they grew inside Petri dishes. When transplanted into rats with heart injuries they lost their spherical shape, spreading out into damaged regions […]
Deep space is a hostile environment for humans, which makes the long journey to Mars a serious stumbling block for manned missions. A nuclear-powered rocket could slash the journey time, and NASA has announced plans to test the technology by 2027 at the latest. Most spacecraft to date have used chemical rockets packed with fuel […]
Last week, a group of South Korean physicists made a startling claim. In two papers uploaded to the arXiv preprint server, they say they have created a material that “opens a new era for humankind.” LK-99, a lead-based compound, is purportedly a room-temperature, ambient-pressure superconductor. Such a material, which conducts electricity without any resistance under […]
ARTIFICIAL INTELLIGENCE GPT-3 Aces Tests of Reasoning by Analogy John Timmer | Ars Technica “A team from University of California, Los Angeles has tested the GPT-3 LLM using questions that should be familiar to any Americans that have spent time on standardized tests like the SAT. In all but one variant of these questions, GPT-3 […]
The components sound like the aftermath of a shopping and spa retreat: three AA batteries. Two electrical acupuncture needles. One plastic holder that’s usually attached to battery-powered fairy lights. But together they merge into a powerful stimulation device, opening a new channel that uses household batteries to control gene expression in cells. The idea seems […]
As scorching heat grips large swaths of the Earth, a lot of people are trying to put the extreme temperatures into context and asking: When was it ever this hot before? Globally, 2023 has seen some of the hottest days in modern measurements, but what about farther back, before weather stations and satellites? Some news […]
Ever since ChatGPT exploded onto the tech scene in November of last year, it’s been helping people write all kinds of material, generate code, and find information. It and other large language models (LLMs) have facilitated tasks from fielding customer service calls to taking fast food orders. Given how useful LLMs have been for humans […]
Sickle cell disease is debilitating. Due to faulty genetic code, red blood cells morph from round and plump into jagged monstrosities that scrape and puncture blood vessels. Over time symptoms build up, eventually damaging major organs like the liver, heart, and kidneys. The disease was incurable—until gene editing came along. In 2020, a breakthrough technology […]
Direct air capture is slowly getting off the ground, with plants up and running in Iceland, Switzerland, the US, and Canada. Much of the carbon these facilities capture is either turned into a solid and stored underground, or reused to manufacture various chemicals and industrial products. Now a startup called Twelve is planning to use […]
ARTIFICIAL INTELLIGENCE Does Sam Altman Know What He’s Creating? Ross Andersen | The Atlantic “i‘We could have gone off and just built this in our building here for five more years,’ [Altman] said, ‘and we would have had something jaw-dropping.’ But the public wouldn’t have been able to prepare for the shock waves that followed, […]
Ask a cancer researcher what the breakthrough treatment of the decade is, and they’ll tell you CAR T takes the crown. The therapy genetically engineers a person’s own immune cells, turning them into super soldiers that hunt down cancerous blood cells. With astonishing speed, multiple CAR T therapies have been approved by the FDA for […]
When theoretical physicists like myself say that we’re studying why the universe exists, we sound like philosophers. But new data collected by researchers using Japan’s Subaru telescope has revealed insights into that very question. The Big Bang kick-started the universe as we know it 13.8 billion years ago. Many theories in particle physics suggest that […]
It’s been a hot summer, with heat waves engulfing multiple parts of the world and setting record temperatures in some. We can hope that in time some of the measures being taken to fight climate change—from switching to renewable energy to capturing atmospheric carbon to using more sustainable building materials—will make a difference, and the […]
I’m a huge fan of Anthony Bourdain’s travel show Parts Unknown. In each episode, the chef visits remote villages across the globe, documenting the lives, foods, and cultures of regional tribes with an open heart and mind. The show provides a glimpse into humanity’s astonishing diversity. Social scientists have a similar goal—understanding the behavior of […]
If you’ve ever wished you had a faster phone, computer, or internet connection, you’ve encountered the personal experience of hitting a limit of technology. But there might be help on the way. Over the past several decades, scientists and engineers like me have worked to develop faster transistors, the electronic components underlying modern electronic and […]
The human brain is somewhat similar to a rudimentary radio with five channels. Electrical signals from neurons coordinate across the brain, generating oscillations known as brain waves. Each wave corresponds to the state of the brain. Some come fast and furious, with a high frequency usually associated with when we’re awake and thinking. Others are […]
ARTIFICIAL INTELLIGENCE Google Tests AI Tool That Is Able to Write News Articles Benjamin Mullin and Nico Grant | The New York Times “Google is testing a product that uses artificial intelligence technology to produce news stories, pitching it to news organizations including The New York Times, The Washington Post and The Wall Street Journal’s […]
If you’ve seen the Terminator movies, you may remember the shape-shifting humanoid robot T-1000. Made of liquid metal, T-1000 could instantly self-heal bullet wounds and other injuries, his metal simply oozing back together and making any damage disappear. Decades after the concept of self-healing metal showed up in a movie, it’s left the realm of […]
Last year, astronomers made an intriguing discovery: a radio signal in space that switched on and off every 18 minutes. Astronomers expect to see some repeating radio signals in space, but they usually blink on and off much more quickly. The most common repeating signals come from pulsars, rotating neutron stars that emit energetic beams […]
In September 2020, the Almeda fire swept across Oregon’s Rogue Valley, destroying more than 2,600 homes. It was the most destructive wildfire in the state’s history, and almost three years later hundreds of displaced families are still living in temporary accommodations like FEMA trailers or hotels. But a new community of 3D printed homes is […]
In the movie M3GAN, a toy developer gives her recently orphaned niece, Cady, a child-sized AI-powered robot with one goal: to protect Cady. The robot M3GAN sympathizes with Cady’s trauma. But things soon go south, with the pint-sized robot attacking anything and anyone who it perceives to be a threat to Cady. M3GAN wasn’t malicious. […]
Artificial intelligence has progressed so rapidly that even some of the scientists responsible for many key developments are troubled by the pace of change. Earlier this year, more than 300 professionals working in AI and other concerned public figures issued a blunt warning about the danger the technology poses, comparing the risk to that of […]
Last year, fast-casual restaurant chain Chipotle brought in a new employee: Chippy the chip-making robot. Chippy was tasked with slicing corn tortillas into triangles, frying them, tossing them with lime juice and salt, and dividing them into portions. The bot must have done a pretty good job, because now the restaurant is bringing in one […]
ARTIFICIAL INTELLIGENCE My New Turing Test Would See if AI Can Make $1 Million Mustafa Suleyman | MIT Technology Review “We need something better [than the Turing Test]. Something adapted to this new phase of AI. …What an AI can say or generate is one thing. But what it can achieve in the world, what […]
An international team led by Chinese scientists just built the most complete atlas of the macaque monkey cortex to date. The outermost layer of the brain, the cortex houses many of our treasured cognitive functions: the ability to reason, make decisions, and adapt to ever-changing environments on the fly. Compared to other animals, primates—including humans—have […]
Though at least one vaccine for malaria is in use, it remains one of the deadliest diseases in the world. Almost half of the world’s population lives in areas where malaria transmission occurs, and an estimated 619,000 people died of the disease in 2021. Worse yet, the vast majority of cases leading to death are […]
Astronomers have discovered more than 5,000 planets outside of the solar system to date. The grand question is whether any of these planets are home to life. To find the answer, astronomers will likely need more powerful telescopes than exist today. I am an astronomer who studies astrobiology and planets around distant stars. For the […]
Every day we’re juggling different needs. I’m hungry but exhausted; should I collapse on the couch or make dinner? I’m overheating in dangerous temperatures but also extremely thirsty; should I chug the tepid water that’s been heating under the sun, or stick my head in the freezer until I have the mental capacity to make […]
Finding ways to stave off the cognitive decline that is common in old age could allow people to live healthy lives for longer. New research suggests injections of a specific protein could boost memory in older monkeys. Diseases like Alzheimer’s can significantly degrade quality of life for older people even if they are otherwise physically […]
A honey bee’s life depends on it successfully harvesting nectar from flowers to make honey. Deciding which flower is most likely to offer nectar is incredibly difficult. Getting it right demands correctly weighing up subtle cues on flower type, age, and history—the best indicators a flower might contain a tiny drop of nectar. Getting it […]
FUTURE OpenAI Is Forming a New Team to Bring ‘Superintelligent’ AI Under Control Kyle Wiggers | TechCrunch “To move the needle forward in the area of ‘superintelligence alignment,’ OpenAI is creating a new Superalignment team, led by both [Ilya] Sutskever and [Jan] Leike, which will have access to 20% of the compute the company has […]
Life finds a way. That’s the conclusion of a new study in Nature, which pitted synthetic bacterial cells against the force of evolution. Stripped down to a skeletal genetic blueprint, the artificial cells started with a losing hand for survival. Yet they thrived, evolving at a rate nearly 40 percent faster than their non-minimal counterparts. […]
The animal world is full of different types of intelligence, from the simple bodily coordination of jellyfish to the navigation abilities of bees, the complex songs of birds, and the imaginative symbolic thought of humans. In an article published this week in Proceedings of the Royal Society B, we argue the evolution of all these […]
Since the start of the quantum race Microsoft has placed its bets on the elusive but potentially game-changing topological qubit. Now the company claims its hail Mary has paid off, saying it could build a working processor in less than a decade. Today’s leading quantum computing companies have predominantly focused on qubits—the quantum equivalent of […]
Transforming human stem cells into embryo-like structures was previously unthinkable. Yet seemingly overnight, multiple teams published initial results that reach towards this goal. Each team has a unique recipe for generating lab-grown embryoids, blobs of cells that mimic aspects of the earliest stages of human life. Although often dubbed “synthetic embryos,” they are anything but. […]
Recent progress in AI has been startling. Barely a week’s gone by without a new algorithm, application, or implication making headlines. But OpenAI, the source of much of the hype, only recently completed their flagship algorithm, GPT-4, and according to OpenAI CEO Sam Altman, its successor, GPT-5, hasn’t begun training yet. It’s possible the tempo […]
ARTIFICIAL INTELLIGENCE Google DeepMind’s CEO Says Its Next Algorithm Will Eclipse ChatGPT Will Knight | Wired “DeepMind’s Gemini, which is still in development, is a large language model that works with text and is similar in nature to GPT-4, which powers ChatGPT. But Hassabis says his team will combine that technology with techniques used in AlphaGo, aiming to […]
The search for planets outside our solar system—exoplanets—is one of the most rapidly growing fields in astronomy. Over the past few decades, more than 5,000 exoplanets have been detected and astronomers now estimate that on average there is at least one planet per star in our galaxy. Many current research efforts aim at detecting Earth-like […]
Robots are slowly getting smarter, mostly thanks to advances in artificial intelligence that enable them to learn on their own rather than requiring constant human guidance. But despite being able to independently learn new tasks, most robots are still limited to a single form of motion—that is, they either walk, crawl, swim, fly, or move […]
Four years ago, biotech company Insilico Medicine used AI to design a molecule targeting a protein involved in fibrosis in just 46 days. It was a proof of concept, as multiple effective drugs already existed for the protein, giving the company a wealth of data to train their AI with. But they’ve made quick progress […]
Trying to understand consciousness calls to mind images of pensive philosophers in a thinking pose. Soft rock and freestyle rap with lyrics based on theories of consciousness aren’t exactly on the bingo card. Yet the tunes galvanized an eager crowd at the 26th Association for the Scientific Study of Consciousness (ASSC 26) meeting in New […]
Life on Earth owes its existence to photosynthesis—a process which is 2.3 billion years old. This immensely fascinating (and still not fully understood) reaction enables plants and other organisms to harvest sunlight, water, and carbon dioxide while converting them into oxygen and energy in the form of sugar. Photosynthesis is such an integral part of […]
Despite rapid advances in artificial intelligence, robots remain stubbornly dumb. But new research from DeepMind suggests the same technology behind large language models (LLMs) could help create more adaptable brains for robotic arms. While autonomous robots have started to move out of the lab and into the real world, they remain fragile. Slight changes in […]
BIOTECH Suddenly, It Looks Like We’re in a Golden Age for Medicine David Wallace-Wells | The New York Times “Hype springs eternal in medicine, but lately the horizon of new possibility seems almost blindingly bright. …’It’s stunning,’ says the immunologist Barney Graham, the former deputy director of the Vaccine Research Center and a central figure […]
The world’s first lab-grown burger was completed in 2013 after five years of research and development and with a price tag of $330,000. Since then, the cultured meat industry has slowly but surely advanced—gaining funding, diversifying the types of meat produced, building factories for large-scale production, going on the market in Singapore and Israel, and […]
There’s been concern about artificial intelligence taking away jobs for years, and with the recent boom in generative AI, those fears have grown. The ability to generate realistic and accurate text, images, or audio based on a prompt could make plenty of jobs obsolete (including, ahem, journalism and writing). But a new study says the […]
Americans spend a lot of time in cars. Before the Covid-19 pandemic, the average US driver was on the road for about an hour a day. Accordingly, we like roomy vehicles with wide, comfortable seats, plenty of special features, and high horsepower; these all come in handy when traveling across a sprawling city, or from […]
Quantum computers may soon tackle problems that stump today’s powerful supercomputers—even when riddled with errors. Computation and accuracy go hand in hand. But a new collaboration between IBM and UC Berkeley showed that perfection isn’t necessarily required for solving challenging problems, from understanding the behavior of magnetic materials to modeling how neural networks behave or […]
The idea of beaming solar power down from space might seem like something out of a sci-fi movie. But new legislation in Congress, a funding announcement from the UK government, and the first successful test of the technology all suggest its moment may be coming. Getting your solar energy directly from space makes a lot […]
ARTIFICIAL INTELLIGENCE Generative AI Can Add $4.4 Trillion in Value to Global Economy, Study Says Yiwen Lu | The New York Times “Generative AI, which includes chatbots such as ChatGPT that can generate text in response to prompts, can potentially boost productivity by saving 60 to 70 percent of workers’ time through automation of their […]
Enceladus is the tiny moon of Saturn that seems to have it all. Its icy surface is intricately carved by ongoing geological processes. Its icy shell overlies an internal, liquid ocean. There, chemically charged warm water seeps out of the rocky core onto the ocean floor, potentially providing nourishment for microbial life. Now, a new […]
Amid the search for solutions to global warming, various forms of geoengineering have gained traction. The more wild ideas—like spraying dust into space to block out some of the sun’s rays—are still being debated. But in the meantime, weather modification in the form of cloud seeding has been implemented in various parts of the world, […]
Brain organoids have come a long way. These mini-brains, at most the size of a pea, are made from stem cells or reprogrammed skin cells and churned inside a bioreactor full of nutrients. With different molecular ingredients, scientists can nudge mini-brains to gradually develop striations and structures similar to a growing fetal brain, with the […]
Last year, 60 percent of the electric cars sold in the world were purchased in China. The country is by far the world’s biggest market for electric vehicles, due partly to the availability of small, cheap EVs like the Wuling Hong Guang Mini (which costs just over $4,000). Tesla is determined to get a big […]
ChatGPT is a hot topic at my university, where faculty members are deeply concerned about academic integrity, while administrators urge us to “embrace the benefits” of this “new frontier.” It’s a classic example of what my colleague Punya Mishra calls the “doom-hype cycle” around new technologies. Likewise, media coverage of human-AI interaction—whether paranoid or starry-eyed—tends […]
This past December, Israeli cultured meat company Believer Meats started construction on what it says will be the biggest cultured meat factory in the world. Its 200,000-square-foot facility is being built near Raleigh, North Carolina. Cultured meat’s viability has been called into question recently, but that doesn’t seem to be slowing the industry down too […]
ARTIFICIAL INTELLIGENCE Google DeepMind’s Game-Playing AI Just Found Another Way to Make Code Faster Will Douglas Heaven | MIT Technology Review “Last year the company used a version of its game-playing AI AlphaZero to find new ways to speed up the calculation of a crucial piece of math at the heart of many different kinds […]
A weird and wonderful array of technologies are competing to become the standard-bearer for quantum computing. The latest contender wants to encode quantum information in sound waves. One thing all quantum computers have in common is the fact that they manipulate information encoded in quantum states. But that’s where the similarities end, because those quantum […]
First detected accidentally by US military satellites in the late 1960s, cosmic explosions known as gamma ray bursts (GRBs) have come to be understood as the brightest explosions in the universe. Typically, they are the result of the cataclysmic birth of a black hole in a distant galaxy. One way this can happen is through […]
Between the Covid-19 pandemic, the Ukraine conflict, inflation, and the renewables transition, the 2020s have been a volatile decade for energy. The pandemic reduced demand for electricity and oil all over the world, causing prices to plummet. Then the Ukraine invasion brought sanctions on Russian oil and gas, pushing energy prices up and leaving European […]
Everyone is raving about hallucinogens as the future of antidepressants. LSD (better known as acid), psilocybin (the active ingredient in magic mushrooms), and the “spirit molecule” DMT are all being tested in clinical trials as fast-acting antidepressants. And I mean fast: when carefully administered by a doctor, they can uplift mood in just one session, […]
Figuring out how to enhance a person’s mental capabilities has been of considerable interest to psychology and neuroscience researchers like me for decades. From improving attention in high-stakes environments, like air traffic management, to reviving memory in people with dementia, the ability to improve cognitive function can have far-reaching consequences. New research suggests that brain stimulation could help achieve the goal of boosting mental function.
In the Reinhart Lab at Boston University, my colleagues and I have been examining the effects of an emerging brain stimulation technology—transcranial alternating current stimulation, or tACS—on different mental functions in patients and healthy people.
During this procedure, people wear an elastic cap embedded with electrodes that deliver weak electrical currents oscillating at specific frequencies to their scalp. By applying these controlled currents to specific brain regions, it is possible to alter brain activity by nudging neurons to fire rhythmically.
Why would rhythmically firing neurons be beneficial? Research suggests that brain cells communicate effectively when they coordinate the rhythm of their firing. Critically, these rhythmic patterns of brain activity show marked abnormalities during neuropsychiatric illnesses. The purpose of tACS is to externally induce rhythmic brain activity that promotes healthy mental function, particularly when the brain might not be able to produce these rhythms on its own.
However, tACS is a relatively new technology, and how it works is still unclear. Whether it can strengthen or revive brain rhythms to change mental function has been a topic of considerable debate in the field of brain stimulation. While some studies find evidence of changes in brain activity and mental function with tACS, others suggest that the currents typically used in people might be too weak to have a direct effect.
When faced with conflicting data in the scientific literature, it can be helpful to conduct a type of study called a meta-analysis that quantifies how consistent the evidence is across several studies. A previous meta-analysis conducted in 2016 found promising evidence for the use of tACS in changing mental function. However, the number of studies has more than doubled since then. The design of tACS technologies has also become increasingly sophisticated.
We set out to perform a new meta-analysis of studies using tACS to change mental function. To our knowledge, this work is the largest and most comprehensive meta-analysis yet on this topic, consisting of over 100 published studies with a combined total of more than 2,800 human participants.
After compiling over 300 measures of mental function across all the studies, we observed consistent and immediate improvement in mental function with tACS. When we examined specific cognitive functions, such as memory and attention, we observed that tACS produced the strongest improvements in executive function, or the ability to adapt in the face of new, surprising, or conflicting information.
We also observed improvements in the ability to pay attention and to memorize information for both short and long periods of time. Together, these results suggest that tACS could particularly improve specific kinds of mental function, at least in the short term.
To examine the effectiveness of tACS for those particularly vulnerable to changes in mental function, we examined the data from studies that included older adults and people with neuropsychiatric conditions. In both populations, we observed reliable evidence for improvements in cognitive function with tACS.
Interestingly, we also found that a specialized type of tACS that can target two brain regions at the same time and manipulate how they communicate with each other can both enhance or reduce cognitive function. This bidirectional effect on mental function could be particularly useful in the clinic. For example, some psychiatric conditions like depression may involve a reduced ability to process rewards, while others like bipolar disorder may involve a highly active reward processing system. If tACS can change mental function in either direction, researchers may be able to develop flexible and targeted designs that cater to specific clinical needs.
Developments in the field of tACS are bringing researchers closer to being able to safely enhance mental function in a noninvasive way that doesn’t require medication. Current statistical evidence across the literature suggests that tACS holds promise, and improving its design could help it produce stronger, long-lasting changes in mental function.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Gerd Altmann / Pixabay
Gene therapies could revolutionize medicine, but getting them into peoples’ bodies is harder than it might seem. A new method that re-purposes viruses that infect bacteria could provide a solution.
Finding ways to modify the DNA in the cells of living people could help treat or prevent a host of genetic diseases. It could also help re-purpose their cells to hunt down cancer or produce therapeutic molecules that could treat non-genetic conditions. But while our gene-editing tools are becoming increasingly sophisticated, getting them into peoples’ bodies is complicated.
A few gene therapies exist today, and they mostly use modified viruses, which excel at sneaking their DNA into their hosts’ cells. This makes these so-called viral vectors perfect cargo carriers for the tools and genetic material required to edit genes inside patients cells. But the adeno-associated viruses (AAVs) and lentiviruses that are most commonly used have a pretty small carrying capacity, which severely limits the scope of problems they can tackle.
New research from the Catholic University of America has shown that a type of bacteriophage—viruses that infect bacteria—with a much bigger cargo hold can be repurposed to deliver gene therapies. It’s also cheap to make, stable, and easy to program to carry out more complex missions.
“The actual therapy is years down the road, but this research provides a model for developing life saving treatments and cures,” Venigalla Rao, who led the research, said in a press release. “What we are researching is like a molecular surgery that can safely and precisely correct a defect and generate therapeutic outcomes and some day cures.”
In the hunt for a more capable delivery vehicle, the researchers turned to a phage called T4, which belongs to the Straboviridae family and infects E. coli bacteria. It has a host of promising characteristics, including a much larger capsid (the main compartment where genetic material is stored), an infection efficiency of nearly 100 percent, and the ability to replicate in just 20 to 30 minutes.
What’s more, researchers have already worked out the atomic structures of the phage’s main components, making the re-engineering process much simpler. This made it possible for the group to set up what it called an “assembly-line approach” in which cargo molecules like DNA, proteins, and RNA were sequentially added to the empty capsid shells and also stuck on their outside as well. The resulting viral vector is then coated in an envelope of lipid molecules, which make it easier to infiltrate human cells.
In a paper in Nature Communications, the researchers showed that their engineered phage could hold stretches of DNA up to 171,000 base pairs long, which is roughly 20 times more than viruses used in current gene therapies can hold. To demonstrate the potential, they used this carrying capacity to deliver the entire gene for the protein dystrophin into human cells. Mutations in this gene are responsible for the genetic disorder Duchenne muscular dystrophy.
In a series of experiments, the researchers showed that the viral vector could be used to do genome editing, gene recombination, gene replacement, gene expression, and gene silencing. They also showed that it could carry complex cargoes made up of multiple stretches of DNA aimed at different genes, alongside various proteins and RNA sequences. The researchers say this could ultimately open the door to treating complex diseases that involve multiple genes like many cancers, neurodegenerative disorders, and cardiovascular diseases.
While these early results are certainly promising, Jeffrey Chamberlain at the University of Washington in Seattle told New Scientist that the team has yet to show the viruses can actually deliver genes into the body, rather than simply to human cells in a petri dish. And Rao concedes that there’s still plenty of work to do to make the jump from the lab bench to the clinic.
But the ability to custom engineer viral vectors for a wide range of applications using their assembly line is highly promising. And unlike existing viral vectors, which have to be reared in human cell cultures at considerable cost, the team’s new engineered phage can be grown far more simply in bacteria.
It’s likely to take many more years of research to bring these ideas to fruition, but if successful, this could greatly expand the scope of future gene therapies.
Image Credit: Venigalla B. Rao; Victor Padilla-Sanchez, Andrei Fokine, and Jingen Zhu. Structural model of bacteriophage T4 artificial viral vector.
ARTIFICIAL INTELLIGENCEWelcome to the New Surreal. How AI-Generated Video Is Changing Film.
Will Douglas Heaven | MIT Technology Review“The Frost nails its uncanny, disconcerting vibe in its first few shots. Vast icy mountains, a makeshift camp of military-style tents, a group of people huddled around a fire, barking dogs. It’s familiar stuff, yet weird enough to plant a growing seed of dread. There’s something wrong here. …The Frost is a 12-minute movie in which every shot is generated by an image-making AI. It’s one of the most impressive—and bizarre—examples yet of this strange new genre.”
NANOTECHNanoscale Robotic ‘Hand’ Made of DNA Could Be Used to Detect Viruses
Michael Le Page | New Scientist“Xing Wang at the University of Illinois and his colleagues constructed the nanohand using a method called DNA origami, in which a long, single strand of DNA is ‘stapled’ together by shorter DNA pieces that pair with specific sequences on the longer strand. …The four fingers of the nanohand are joined to a ‘palm’ to form a cross shape when the hand is open. Each finger is just 71 nanometers long…and has three joints, like a human finger.”
AUTOMATIONThe ‘Death of Self-Driving Cars’ Has Been Greatly Exaggerated
Timothy B. Lee | Ars Technica“[Google and Waymo] don’t believe self-driving technology is ‘decades away’ because they’re already testing it in Phoenix and San Francisco. And they are preparing to launch in additional cities in the coming months. Waymo expects to increase passenger rides tenfold between now and the summer of 2024. Cruise is aiming for $1 billion in revenue in 2025, which would require something like a 50-fold expansion of its current service.”
SENSORSThis Is the First X-Ray Taken of a Single Atom
Jennifer Ouellette | Ars Technica“Atomic-scale imaging emerged in the mid-1950s and has been advancing rapidly ever since—so much so, that back in 2008, physicists successfully used an electron microscope to image a single hydrogen atom. Five years later, scientists were able to peer inside a hydrogen atom using a ‘quantum microscope,’ resulting in the first direct observation of electron orbitals. And now we have the first X-ray taken of a single atom.”
HEALTHGet Ready for 3D-Printed Organs and a Knife That ‘Smells’ Tumors
Joao Madeiros | Wired“To doctors and nurses working 75 years ago, when the UK’s National Health Service was founded, a modern ward would be completely unrecognizable. Fast-forward into the future, and hospitals are likely to look very different again. These are some of the changes you’re likely to see in years to come.”
FUTUREI’m a Rational Optimist. Here’s Why I Don’t Believe in an AI Doomsday.
Rohit Krishnan | BigThink“The systems of today are powerful. They can write, paint, direct, plan, code, and even write passable prose. And with this explosion of capabilities, we also have an explosion of worries. In seeing some of these current problems and projecting them into future non-extant problems, we find ourselves in a bit of a doom loop. The more fanciful arguments about how artificial superintelligence is inevitable and how they’re incredibly dangerous sit side by side with more understandable concerns about increasing misinformation.”
ARTIFICIAL INTELLIGENCEThe Race to Make AI Smaller (and Smarter)
Oliver Whang | The New York Times“[In January, a group of young AI researchers] called for teams to create functional language models using data sets that are less than one-ten-thousandth the size of those used by the most advanced large language models. A successful mini-model would be nearly as capable as the high-end models but much smaller, more accessible and more compatible with humans. The project is called the BabyLM Challenge.”
ETHICSJudge Bans AI-Generated Filings In Court Because It Just Makes Stuff Up
Chloe Xiang | Motherboard“This decision follows an incident where a Manhattan lawyer named Steven A. Schwartz used ChatGPT to write a 10-page brief that cited multiple cases that were made up by the chatbot, such as ‘Martinez v. Delta Air Lines,’ and ‘Varghese v. China Southern Airlines.’ After Schwartz submitted the brief to a Manhattan federal judge, no one could find the decisions or quotations included, and Schwartz later admitted in an affidavit that he had used ChatGPT to do legal research.”
ENERGYThe World Is Finally Spending More on Solar Than Oil Production
Casey Crownhart | MIT Technology Review“Let’s start with what I consider to be good news: there’s a lot of money going into clean energy—including renewables, nuclear, and things that help cut emissions, like EVs and heat pumps. And not only is it a lot of money, but it’s more than the amount going toward fossil fuels. In 2022, for every dollar spent on fossil fuels, $1.70 went to clean energy. Just five years ago, it was dead even.”
SCIENCEThe Quest to Use Quantum Mechanics to Pull Energy Out of Nothing
Charlie Wood | Wired“In the past year, researchers have teleported energy across microscopic distances in two separate quantum devices, vindicating Hotta’s theory. The research leaves little room for doubt that energy teleportation is a genuine quantum phenomenon. ‘This really does test it,’ said Seth Lloyd, a quantum physicist at the Massachusetts Institute of Technology who was not involved in the research. “You are actually teleporting. You are extracting energy.”
Image Credit: Pawel Czerwinski / Unsplash
Air travel is a major source of carbon emissions, accounting for about 2.4 percent of global emissions each year. There are a range of solutions in the works, from electrifying aircraft to using hydrogen fuel to bringing back the airship. The likelihood of any of these coming to fruition varies, and even if they do, it won’t be soon.
A California-based startup called JetZero has a different idea: changing the shape of commercial planes and the material they’re made of. The company unveiled its designs for the midsize commercial and military tanker-transport markets this spring, and has big plans to upend the way air travel looks and feels—as well as how much it costs and how much carbon it emits. Tony Fadell, founder of venture capital firm Build Collective and a JetZero investor and strategic advisor, thinks the company could be the “SpaceX of aviation” due to its potential to disrupt the existing business model.
JetZero’s planes, which are still in the concept/prototype phase, have a blended wing body design. That means the wings merge with the main body of the aircraft, rather than being attached to a hollow tube like the planes we travel in today. Picture the body of a manta ray: wide and flat, it tapers off to a narrower fin at each side, with a head and a tail. A blended wing body aircraft isn’t terribly different, though on JetZero’s models the body isn’t quite as wide.
Besides providing a lot more space, this design is more aerodynamic than tube-and-wing planes. JetZero plans to fly its planes at higher altitudes than today’s norm (40 to 45,000 feet rather than 30 to 35,000), and says its airframe will cut fuel burn and emissions in half. It plans to make its planes out of carbon fiber and kevlar (a strong lightweight fiber used for things like body armor, bulletproof vests, car brakes, boats, and aircraft). The company says its planes’ lighter weight and improved aerodynamics would be able to fly at the same speed and range as existing midbody jetliners, but burn half as much fuel in the process.
JetZero points out that we’ve brought the traditional tube-and-wing design about as far as we possibly can in terms of efficiency gains; there’s not much more to be done to make them lighter, faster, or more fuel-efficient. At the same time, jet fuel is getting more expensive, and reducing emissions is getting more urgent. If JetZero is able to bring its blended wing body aircraft to production, it would be the first major overhaul of commercial passenger planes, well, ever. But, the company says, its planes would still fit seamlessly into airport infrastructure, utilizing existing runways and gates without requiring significant alterations.
The company is planning to test a small prototype of its design with a 23-foot wingspan this summer, and is hoping to secure Air Force funding to build a full-size prototype it would demo in 2027. If all goes according to plan, JetZero’s planes would start commercial service in a decade or so.
There are a lot of bumps they could encounter on the road to get there, but also a lot of incentive to make it happen. The aforementioned climate impact of air travel is likely to come under increased fire, as is occurring with anything that has significant climate impact.
Some countries are trying to reduce the amount of air travel their citizens do, like France, which just banned domestic short-haul flights. Other countries including Spain and Germany are considering restricting short-haul flights or imposing an extra tax on them. Measures like these may make a small difference, but when it comes down to it, people are still going to want to travel. In fact, if the global middle class continues to grow, demand for air travel will only go up; at present, only about three percent of the global population takes regular flights.
A more planet-friendly way to do it, then, is going to be imperative. SpaceX has proven that it’s possible for one private company to come along and completely upend a massively complex industry. Could JetZero do the same?
They’re going to try.
Image Credit: JetZero
From its humble origin(s), life has infected the entire planet with endless beautiful forms. The genesis of life is the oldest biological event, so old that no clear evidence was left behind other than the existence of life itself. This leaves many questions open, and one of the most tantalizing is how many times life magically emerged from non-living elements.
Has all of life on Earth evolved only once, or are different living beings cut from different cloths? The question of how difficult it is for life to emerge is interesting, not least because it can shed some light on the likelihood of finding life on other planets.
The origin of life is a central question in modern biology, and probably the hardest to study. This event took place four billion years ago, and it happened at a molecular level, meaning little fossil evidence remains.
Many lively beginnings have been suggested, from unsavory primordial soups to outer space. But the current scientific consensus is that life emerged from non-living molecules in a natural process called abiogenesis, most likely in the darkness of deep-sea hydrothermal vents. But if life emerged once, why not more times?
What Is Abiogenesis?Scientists have proposed various consecutive steps for abiogenesis. We know that Earth was rich in several chemicals, such as amino acids, a type of molecules called nucleotides or sugars, which are the building blocks of life. Laboratory experiments, such as the iconic Miller-Urey experiment, have shown how these compounds can be naturally formed under conditions similar to early Earth. Some of these compounds could also have come to Earth riding meteorites.
Next, these simple molecules combined to form more complex ones, such as fats, proteins, or nucleic acids. Importantly, nucleic acids—such as double-stranded DNA or its single-stranded cousin RNA—can store the information needed to build other molecules. DNA is more stable than RNA, but in contrast, RNA can be part of chemical reactions in which a compound makes copies of itself—self-replication.
The “RNA world” hypothesis suggests that early life may have used RNA as material for both genes and replication before the emergence of DNA and proteins.
Once an information system can make copies of itself, natural selection kicks in. Some of the new copies of these molecules (which some would call “genes”) will have errors, or mutations, and some of these new mutations will improve the replication ability of the molecules. Therefore, over time, there will be more copies of these mutants than other molecules, some of which will accumulate further new mutations, making them even faster and more abundant, and so on.
Eventually, these molecules probably evolved a lipid (fatty) boundary separating the internal environment of the organism from the exterior, forming protocells. Protocells could concentrate and organize better the molecules needed in biochemical reactions, providing a contained and efficient metabolism.
Life on Repeat?Abiogenesis could have happened more than once. Earth could have birthed self-replicating molecules several times, and maybe early life for thousands or millions of years just consisted of a bunch of different self-replicating RNA molecules, with independent origins, competing for the same building blocks. Alas, due to the ancient and microscopic nature of this process, we may never know.
Many lab experiments have successfully reproduced different stages of abiogenesis, proving they could happen more than once, but we have no certainty of these occurring in the past.
A related question could be whether new life is emerging by abiogenesis as you are reading this. This is very unlikely, though. Early Earth was sterile of life and the physical and chemical conditions were very different. Nowadays, if somewhere on the planet there were ideal conditions for new self-replicating molecules to appear, they would be promptly chomped by existing life.
What we do know is that all extant life beings descend from a single shared last universal common ancestor of life (also known as LUCA). If there were other ancestors, they left no descendants behind. Key pieces of evidence support the existence of LUCA. All life on Earth uses the same genetic code, namely the correspondence between nucleotides in DNA known as A, T, C, and G—and the amino acid they encode in proteins. For example, the sequence of the three nucleotides ATG always corresponds to the amino acid methionine.
Theoretically, however, there could have been more genetic code variants between species. But all life on Earth uses the same code with a few minor changes in some lineages. Biochemical pathways, such as the ones used to metabolize food, also support the existence of LUCA; many independent pathways could have evolved in different ancestors, yet some (such as the ones used to metabolize sugars) are shared across all living organisms. Similarly, hundreds of identical genes are present in disparate live beings which can only be explained by being inherited from LUCA.
My favorite support for LUCA comes from the Tree of Life. Independent analyses, some using anatomy, metabolism, or genetic sequences, have revealed a hierarchical pattern of relatedness that can be represented as a tree. This shows we are more related to chimps than to any other living organisms on Earth. Chimps and we are more related to gorillas, and together to orangutans, and so on.
You can pick any random organism, from the lettuce in your salad to the bacteria in your bioactive yogurt and, if you travel back in time far enough, you will share an actual common ancestor. This is not a metaphor, but a scientific fact.
This is one of the most mind-boggling concepts in science, Darwin’s unity of life. If you are reading this text, you are here thanks to an uninterrupted chain of reproductive events going back billions of years. As exciting as it is to think about life repeatedly emerging on our planet, or elsewhere, it is even more exciting to know that we are related to all the life beings in the planet.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Giovanni Cancemi / Shutterstock.com
For several years now, we’ve been able to go to grocery stores or restaurants and buy all sorts of products made from plants in imitation of products from animals. Burgers. Chicken. Sausages. Bacon. Now another product is being added to the list, and it’s even more unexpected than fake meat; you could even call it […]
Sunburns are terrible. The skin blisters and peels. Even a light brush from putting on clothes or tucking into bed sheets is agony.
Now imagine having those blisters at just six months old. But the sun isn’t the culprit; your genes are.
Thousands of people in the US have dystrophic epidermolysis bullosa (DEB), a rare genetic disorder that affects the structure and integrity of the skin and eyes. Kids with the illness are cursed with skin similar to wet tissue paper. Chronic painful blisters and wounds—sometimes inside their throats—are a part of life since birth.
The root cause is frustratingly simple: one gene mutation, which affects a critical protein that helps support skin integrity. The single genetic error makes the illness a perfect candidate for gene therapy. Yet with the skin already fragile, injections—a current standard for gene therapy—are hard to tolerate.
What about a genetic moisturizer instead?
This month, the FDA approved the first rub-on gene therapy. Similar to aloe vera for treating sunburns, the therapy comes in a gel that’s gently massaged onto blisters and wounds to help with healing. Dubbed Vyjuvek, it directly delivers healthy copies of the mutated gene onto damaged skin. An alternative version is configured into eye drops to reconstruct the eye’s delicate architecture to better support sight.
In multiple clinical trials from patients ranging from a year old to middle aged, the treatment reduced painful blisters after six months. With a massage every week, over two-thirds of the patients’ wounds completely healed, compared to just one out of five in wounds treated with a placebo. The patients’ eyesight also improved, allowing a 13-year-old volunteer to finally play Minecraft online with his teenage peers.
The therapy is the latest to expand the universe of gene therapy delivery technologies. When further developed, it won’t be limited to rare skin conditions. Because the therapy targets collagen, a critical protein that helps maintain skin structure and elasticity, the rub-on treatment could launch the next generation of moisturizers to combat fine lines and crow’s feet from aging. A subsidiary of Krystal Biotech based in Pittsburgh, which developed Vyjuvek, is already expanding into cosmetics.
It’s not all superficial. Beauty aside, the FDA nod of approval “ushers in a whole new paradigm to treat genetic diseases,” said Krystal’s CEO Krish S. Krishnan.
The Perfect PackageVyjuvek joins a prestigious roster of approved gene therapies.
These treatments have mainly battled blood cancers and disorders. Usually, doctors need to extract immune or red blood cells from a patient’s blood. The cells are then genetically enhanced and infused back into the body. Some are amped up to pursue cancer targets. Others help boost hemoglobin in red blood cells, which carries oxygen across the body.
Unlike deeper organs, blood cells are relatively easy to access, making them a valuable resource for genetic tweaks. Just last year, one team expanded gene editing’s potential by infusing CRISPR components directly into blood, which helped brush away a toxic protein made by the liver that leads to pain, numbness, and eventually heart failure in an inherited disease.
On the surface (no pun intended), the skin is another easily accessible target—just think of the thousands of skincare products on the market. Yet our external barrier is also a formidable fortress with multiple protective layers. The top, the epidermis, is a flexible biological shield composed of tightly-knit cells, making it difficult for large intruders—including collagen—to penetrate. Current treatments for DEB rely on genetically engineered skin outside the body being grafted onto patients. It’s as intense as it sounds: patients range from a week-long stay in the hospital to a prolonged medically-induced coma to tolerate the procedure.
The team tackled the conundrum with a cleverly-balanced hand. First was deciphering the genetic error that leads to DEB: a gene called COL7A1, which encodes a type of collagen. Like anchors stabilizing skyscraper scaffolds, COL7 molecules arrange into long, thin but extremely strong bundles to hold the epidermis and the middle layer of skin together. When deficient, the two layers separate, leading to painful blisters and wounds that resemble severe sunburn or frostbite, often from birth.
Unfortunately, COL7A1 is also a tough gene to deliver due to its enormous size. The team chose their carrier “vector” carefully: HSV-1, a type of harmless herpes simple virus, extensively engineered so it doesn’t replicate inside the body or cause any disease.
The vector is the first gamble for most gene therapies. Think of them as Amazon delivery boxes. Some are extremely efficient at penetrating into cells (or into tiny mailboxes), but can only hold a small payload. Others have greater capacity, but are then dumped on your front yard—easy for others (think immune cells) to see and potentially vandalize.
Previous attempts have collected skin tissue and used viruses for cell engineering and grafting, but “correction of genetic skin diseases via direct gene transfer in vivo [inside the body] has been a longstanding yet unrealized goal in the gene therapy field,” said the team last year in a clinical study for safety.
Their final recipe worked out stunningly well. First, a healthy version of the COL7 gene was genetically packaged into the selected viral vector. The entire architecture was then suspended inside a gel, similar to molecular gastronomy, to stabilize the treatment. The end result was gene therapy inside a moisturizer bottle—a first without the need for needles or other painful procedures.
In a study with 31 patients published last December in the New England Journal of Medicine, the ointment, dubbed B-VEC, healed nearly 70 percent of painful wounds across patients, compared to just roughly 20 percent in those rubbed with a placebo. The treatment also reduced pain throughout the trial.
A New Therapy LandscapeWhile impressive, the treatment isn’t a cure.
Because the skin readily replaces itself with new cells that carry similar genetic defects, the gel will need to be rubbed on at least weekly by a healthcare professional. With gene therapy costing up to millions of dollars, the numbers readily stack up. But the recipients are grateful.
“With the FDA approval of Vyjuvek the DEB population has reached a monumental milestone in the treatment of this horrible disorder,” said Brett Kopelan, Executive Director of debra, an organization that supports people with the illness. “Our hopes have now been realized for a safe and effective treatment for one of the most devastating symptoms of the disorder.”
On a broader scale, the study widens the gene therapy landscape, firing the first shot at rub-on therapies—with eye drops to quickly follow.
For now, Krystal is branching out into the million-dollar skincare industry to combat aging and damaged skin. It may be tricky: unlike wounds with open skin, normal skin forms a tighter and protective barrier. But if it succeeds, the treatment may open doors to highly efficient skin care—all inside a bottle, no need for a knife.
Image Credit: Ricarda Mölck from Pixabay
After a three-year hiatus, scientists in the US have just turned on detectors capable of measuring gravitational waves—tiny ripples in space itself that travel through the universe.
Unlike light waves, gravitational waves are nearly unimpeded by the galaxies, stars, gas, and dust that fill the universe. This means that by measuring gravitational waves, astrophysicists like me can peek directly into the heart of some of the most spectacular phenomena in the universe.
Since 2020, the Laser Interferometric Gravitational-Wave Observatory—commonly known as LIGO—has been sitting dormant while it underwent some exciting upgrades. These improvements will significantly boost the sensitivity of LIGO and should allow the facility to observe more-distant objects that produce smaller ripples in spacetime.
By detecting more of the events that create gravitational waves, there will be more opportunities for astronomers to also observe the light produced by those same events. Seeing an event through multiple channels of information, an approach called multi-messenger astronomy, provides astronomers rare and coveted opportunities to learn about physics far beyond the realm of any laboratory testing.
According to Einstein’s theory of general relativity, massive objects warp space around them. Image Credit: vchal/iStock via Getty ImagesRipples in SpacetimeAccording to Einstein’s theory of general relativity, mass and energy warp the shape of space and time. The bending of spacetime determines how objects move in relation to one another—what people experience as gravity.
Gravitational waves are created when massive objects like black holes or neutron stars merge with one another, producing sudden, large changes in space. The process of space warping and flexing sends ripples across the universe like a wave across a still pond. These waves travel out in all directions from a disturbance, minutely bending space as they do so and ever so slightly changing the distance between objects in their way.
Even though the astronomical events that produce gravitational waves involve some of the most massive objects in the universe, the stretching and contracting of space is infinitesimally small. A strong gravitational wave passing through the Milky Way may only change the diameter of the entire galaxy by three feet (one meter).
The First Gravitational Wave ObservationsThough first predicted by Einstein in 1916, scientists of that era had little hope of measuring the tiny changes in distance postulated by the theory of gravitational waves.
Around the year 2000, scientists at Caltech, the Massachusetts Institute of Technology, and other universities around the world finished constructing what is essentially the most precise ruler ever built—LIGO.
The LIGO detector in Hanford, Wash., uses lasers to measure the minuscule stretching of space caused by a gravitational wave. Image Credit: LIGO LaboratoryLIGO is comprised of two separate observatories, with one located in Hanford, Washington, and the other in Livingston, Louisiana. Each observatory is shaped like a giant L with two, 2.5-mile-long (four-kilometer-long) arms extending out from the center of the facility at 90 degrees to each other.
To measure gravitational waves, researchers shine a laser from the center of the facility to the base of the L. There, the laser is split so that a beam travels down each arm, reflects off a mirror and returns to the base. If a gravitational wave passes through the arms while the laser is shining, the two beams will return to the center at ever so slightly different times. By measuring this difference, physicists can discern that a gravitational wave passed through the facility.
LIGO began operating in the early 2000s, but it was not sensitive enough to detect gravitational waves. So, in 2010, the LIGO team temporarily shut down the facility to perform upgrades to boost sensitivity. The upgraded version of LIGO started collecting data in 2015 and almost immediately detected gravitational waves produced from the merger of two black holes.
Since 2015, LIGO has completed three observation runs. The first, run O1, lasted about four months; the second, O2, about nine months; and the third, O3, ran for 11 months before the COVID-19 pandemic forced the facilities to close. Starting with run O2, LIGO has been jointly observing with an Italian observatory called Virgo.
Between each run, scientists improved the physical components of the detectors and data analysis methods. By the end of run O3 in March 2020, researchers in the LIGO and Virgo collaboration had detected about 90 gravitational waves from the merging of black holes and neutron stars.
The observatories have still not yet achieved their maximum design sensitivity. So, in 2020, both observatories shut down for upgrades yet again.
Upgrades to the mechanical equipment and data processing algorithms should allow LIGO to detect fainter gravitational waves than in the past. Image Credit: LIGO/Caltech/MIT/Jeff Kissel, CC BY-NDMaking Some UpgradesScientists have been working on many technological improvements.
One particularly promising upgrade involved adding a 1,000-foot (300-meter) optical cavity to improve a technique called squeezing. Squeezing allows scientists to reduce detector noise using the quantum properties of light. With this upgrade, the LIGO team should be able to detect much weaker gravitational waves than before.
My teammates and I are data scientists in the LIGO collaboration, and we have been working on a number of different upgrades to software used to process LIGO data and the algorithms that recognize signs of gravitational waves in that data. These algorithms function by searching for patterns that match theoretical models of millions of possible black hole and neutron star merger events. The improved algorithm should be able to more easily pick out the faint signs of gravitational waves from background noise in the data than the previous versions of the algorithms.
Astronomers have captured both the gravitational waves and light produced by a single event, the merger of two neutron stars. The change in light can be seen over the course of a few days in the top right inset. Image Credit: Hubble Space Telescope, NASA and ESAA Hi-Def Era of AstronomyIn early May 2023, LIGO began a short test run—called an engineering run—to make sure everything was working. On May 18, LIGO detected gravitational waves likely produced from a neutron star merging into a black hole.
LIGO’s 20-month observation run 04 officially started on May 24, and it will later be joined by Virgo and a new Japanese observatory—the Kamioka Gravitational Wave Detector, or KAGRA.
While there are many scientific goals for this run, there is a particular focus on detecting and localizing gravitational waves in real time. If the team can identify a gravitational wave event, figure out where the waves came from and alert other astronomers to these discoveries quickly, it would enable astronomers to point other telescopes that collect visible light, radio waves, or other types of data at the source of the gravitational wave. Collecting multiple channels of information on a single event—multi-messenger astrophysics—is like adding color and sound to a black-and-white silent film and can provide a much deeper understanding of astrophysical phenomena.
Astronomers have only observed a single event in both gravitational waves and visible light to date—the merger of two neutron stars seen in 2017. But from this single event, physicists were able to study the expansion of the universe and confirm the origin of some of the universe’s most energetic events known as gamma-ray bursts.
With run O4, astronomers will have access to the most sensitive gravitational wave observatories in history and hopefully will collect more data than ever before. My colleagues and I are hopeful that the coming months will result in one—or perhaps many—multi-messenger observations that will push the boundaries of modern astrophysics.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA’s Goddard Space Flight Center/Scott Noble; simulation data, d’Ascoli et al. 2018
COMPUTINGIBM Wants to Build a 100,000-Qubit Quantum Computer
Michael Brooks | MIT Technology Review“Late last year, IBM took the record for the largest quantum computing system with a processor that contained 433 quantum bits, or qubits, the fundamental building blocks of quantum information processing. Now, the company has set its sights on a much bigger target: a 100,000-qubit machine that it aims to build within 10 years.”
ARTIFICIAL INTELLIGENCEScientists Use AI to Discover New Antibiotic to Treat Deadly Superbug
Maya Yang | The Guardian“After scientists trained the AI model, they used it to analyze 6,680 compounds that it had previously not encountered. The analysis took an hour and half and ended up producing several hundred compounds, 240 of which were then tested in a laboratory. Laboratory testing ultimately revealed nine potential antibiotics, including abaucin. The scientists then tested the new molecule against A baumannii in a wound infection model in mice and found that the molecule suppressed the infection.”
COMPUTINGNvidia Is Poised to Join $1 Trillion Club Thanks to AI-Driven Surge
Sharon Goldman | VentureBeat“Nvidia’s stock soared nearly 30% after it announced its first-quarter financial results yesterday, setting the stage for Nvidia to become only the fifth publicly traded US company to be currently worth $1 trillion—joining Apple, Microsoft, Alphabet, and Amazon. And it’s all thanks to the hunger for high-powered AI chips in the era of generative AI.”
HEALTHA Paralyzed Man Can Walk Naturally Again With Brain and Spine Implants
Oliver Whang | The New York Times“In a study published on Wednesday in the journal Nature, researchers in Switzerland described implants that provided a ‘digital bridge’ between Mr. Oskam’s brain and his spinal cord, bypassing injured sections. The discovery allowed Mr. Oskam, 40, to stand, walk and ascend a steep ramp with only the assistance of a walker. More than a year after the implant was inserted, he has retained these abilities and has actually showed signs of neurological recovery, walking with crutches even when the implant was switched off.”
ROBOTICSHumanoid Robots Are Coming of Age
Will Knight | Wired“Eight years ago, the Pentagon’s Defense Advanced Research Projects Agency organized a painful-to-watch contest that involved robots slowly struggling (and often failing) to perform a series of human tasks, including opening doors, operating power tools, and driving golf carts. …Today the descendants of those hapless robots are a lot more capable and graceful. Several startups are developing humanoids that they claim could, in just a few years, find employment in warehouses and factories.”
ENERGYScientists Working to Generate Electricity From Thin Air Make Breakthrough
Becky Ferreira | Motherboard“Scientists have invented a device that can continuously generate electricity from thin air, offering a glimpse of a possible sustainable energy source that can be made of almost any material and runs on the ambient humidity that surrounds all of us, reports a new study.”
SPACEHow NASA Plans to Melt the Moon—and Build on Mars
Khari Johnson | Wired“In June a four-person crew will enter a hangar at NASA’s Johnson Space Center in Houston, Texas, and spend one year inside a 3D printed building. Made of a slurry that—before it dried—looked like neatly laid lines of soft-serve ice cream, Mars Dune Alpha has crew quarters, shared living space, and dedicated areas for administering medical care and growing food.”
VIRTUAL REALITYReplica Unveils AI-Powered Smart NPCs for Unreal Engine
Dean Takahashi | VentureBeat“The smart NPCs are powered by OpenAI or the user’s own AI language model, and Replica’s library of over 120 ethically licensed AI voices, allowing game developers to develop games at scale and create new dynamic gaming experiences. …In Replica’s smart NPC experience, AI-powered NPCs will dynamically respond to the player’s in-game voice in real time, the company said. Characters will change their dialogue, emotional tone and body gestures in reaction to how the player speaks to them.”
COMPUTING‘Fluxonium’ Is the Longest Lasting Superconducting Qubit Ever
Karmela Padavic-Callaghan | NewScientist“Somoroff says that the best transmon qubits have coherence times of hundreds of microseconds, but he and his team measured about 1.48 milliseconds for their fluxonium qubit. They also determined that they could change their qubit’s state, something that would have to happen many times during a computation on a fluxonium quantum computer, with 99.991 per cent fidelity. This makes the fluxonium qubit one of the most reliable qubits that exists, almost always changing states exactly as instructed.”
NEUROSCIENCESome Neural Networks Learn Language Like Humans
Steve Nadis | Quanta“The researchers—led by Gašper Beguš, a computational linguist at the University of California, Berkeley—compared the brain waves of humans listening to a simple sound to the signal produced by a neural network analyzing the same sound. The results were uncannily alike. ‘To our knowledge,’ Beguš and his colleagues wrote, the observed responses to the same stimulus ‘are the most similar brain and ANN signals reported thus far.’i”
Image Credit: Maxim Berg / Unsplash
The ability of machines to read our minds has been steadily progressing in recent years. Now, researchers have used AI video generation technology to give us a window into the mind’s eye.
The main driver behind attempts to interpret brain signals is the hope that one day we might be able to offer new windows of communication for those in comas or with various forms of paralysis. But there are also hopes that the technology could create more intuitive interfaces between humans and machines that could also have applications for healthy people.
So far, most research has focused on efforts to recreate the internal monologues of patients, using AI systems to pick out what words they are thinking of. The most promising results have also come from invasive brain implants that are unlikely to be a practical approach for most people.
Now though, researchers from the National University of Singapore and the Chinese University of Hong Kong have shown that they can combine non-invasive brain scans and AI image generation technology to create short snippets of video that are uncannily similar to clips that the subjects were watching when their brain data was collected.
The work is an extension of research the same authors published late last year, where they showed they could generate still images that roughly matched the pictures subjects had been shown. This was achieved by first training one model on large amounts of data collected using fMRI brain scanners. This model was then combined with the open-source image generation AI Stable Diffusion to create the pictures.
In a new paper published on the preprint server arXiv, the authors take a similar approach, but adapt it so that the system can interpret streams of brain data and convert them into videos rather than stills. First, they trained one model on large amounts of fMRI so that it could learn the general features of these brain scans. This was then augmented so it could process a succession of fMRI scans rather than individual ones, and then trained again on combinations of fMRI scans, the video snippets that elicited that brain activity, and text descriptions.
Separately, the researchers adapted the pre-trained Stable Diffusion model to produce video rather than still images. It was then trained again on the same videos and text descriptions that the first model had been trained on. Finally, the two models were combined and fine-tuned together on fMRI scans and their associated videos.
The resulting system was able to take fresh fMRI scans it hadn’t seen before and generate videos that broadly resembled the clips human subjects had been watching at the time. While far from a perfect match, the AI’s output was generally pretty close to the original video, accurately recreating crowd scenes or herds of horses and often matching the color palette.
To evaluate their system, the researchers used a video classifier designed to assess how well the model had understood the semantics of the scene—for instance, whether it had realized the video was of fish swimming in an aquarium or a family walking down a path—even if the imagery was slightly different. Their model scored 85 percent, which is a 45 percent improvement over the state-of-the-art.
While the videos the AI generates are still glitchy, the authors say this line of research could ultimately have applications in both basic neuroscience and also future brain-machine interfaces. However, they also acknowledge potential downsides to the technology. “Governmental regulations and efforts from research communities are required to ensure the privacy of one’s biological data and avoid any malicious usage of this technology,” they write.
That is likely a nod to concerns that the combination of AI brain scanning technology could make it possible for people to intrusively record other’s thoughts without their consent. Anxieties were also voiced earlier this year when researchers used a similar approach to essentially create a rough transcript of the voice inside peoples’ heads, though experts have pointed out that this would be impractical if not impossible for the foreseeable future.
But whether you see it as a creepy invasion of your privacy or an exciting new way to interface with technology, it seems machine mind readers are edging closer to reality.
Image Credit: Claudia Dewald from Pixabay
From Kenya to Mexico, Texas, and beyond, 3D-printed houses are starting to go up all over the world. Besides providing a durable and aesthetically pleasing structure, one of the biggest goals of 3D-printed homes is affordability. The technology replaces part of the human labor needed for building, cutting the cost of construction and lowering the home’s final price tag.
But so far this seems to be harder than anticipated; while a handful of 3D-printed homes have been priced well below their conventionally-built competitors, many others have sold at parity or just slightly undercut average market prices. Given that there’s a housing shortage of somewhere between 2.3 to 6.5 million homes (depending on whether multi-family construction is included) in the US, we’re going to need to do a lot better than that.
Construction technology company ICON is aiming to find a way forward. The company (which is currently building a community of 100 3D-printed homes outside Austin, Texas) is launching a competition for disruptive solutions for affordable housing. Called Initiative 99, the contest will call for 3D-printed home designs that can be built for under $99,000.
ICON released the submission details for the contest yesterday. Its parameters are fairly general, which opens up a lot of possibility for different designs. All homes must have a minimum of one bedroom and one bathroom—sorry, studios—within flexible square footage. They have to be designed with a target group in mind, such as young families, people who were formerly homeless, the elderly, etc. The homes also have to comply with residential building code requirements, and be possible to build using ICON’s Vulcan 3D printer.
Outside of those must-haves, participants are encouraged to think about how their design could be scaled, i.e. for a community of 20 or more homes. They should take climate and sustainability into account. Rainwater collection system? Great. Solar panel ready? Even better. Hypothetical occupants should be able to expect “low, consistent, and predictable utility bills over the entire year” because the homes should be as energy-efficient as possible.
In terms of costs, the $99,000 threshold must include printing, additional construction costs, and finish-out costs (like mechanical, electrical, and plumbing systems). It doesn’t include land, labor, utility connections, or permits.
The most expensive parts of building a house the conventional way are labor and materials, with labor being most expensive. You need carpenters, plumbers, electricians, and roofers. Window installers. Kitchen cabinet specialists. Someone to lay the concrete foundation. For custom homes, an architect. And the list goes on.
One of the biggest ways to save on these labor costs is to use a fixed design and panelize the building. Studs for the interior walls—that is, the two-by-fours that are put up to create rooms, then covered by drywall—can be prefabricated. Floors can be panelized too (not the finishes, like hardwood or carpet, but the structure).
Companies have come up with all sorts of creative ideas to reduce homebuilding costs in ways similar to this. Boxabl makes prefabricated “foldable” homes that can ship on an eight-foot footprint, and they start at $49,500 (though that’s for a 400-square-foot studio). Similarly, NODE makes prefab homes that ship in kits then are assembled like Ikea furniture. Vantem Global makes energy-efficient prefabricated homes out of structural panels, and Automatic Construction is trying to build houses by pumping concrete into inflatable forms.
What sorts of similarly innovative ideas might builders and entrepreneurs come up with for ICON’s competition? How much can be done within 3D printing as a building technology to further reduce its costs and make it more scalable, all while producing appealing, comfortable homes?
Entrants will have their work cut out for them. Their design submissions won’t just be judged on constructability and innovation—the judges will also consider aesthetics, sustainability, cost, and scalability. The winning design will receive $75,000, second place $50,000 and third place $35,000. Submissions will open this summer.
Image Credit: ICON
Virtual reality experiences depend on goggles and headphones, transporting wearers to new places using sight and sound. Be it a peaceful meadow where the only sounds are birds chirping and the breeze blowing through the grass, or a packed stadium with thousands of fans cheering on a pro football team, what you see and hear are key components of an immersive experience.
But they’re not the only ones. Multiple companies are working on haptic devices, like gloves or vests, to add a sense of touch to virtual experiences. And now, researchers are aiming to integrate a fourth sense: smell.
How much more real might that peaceful meadow feel if you could smell the wildflowers and the damp Earth around you? How might the scent of an ocean breeze amplify a VR experience that takes place on a boat or a beach?
Scents have a powerful effect on the brain, eliciting emotions, memories, and sometimes even fight-or-flight responses. You may feel nostalgic with the cologne or perfume a favorite grandparent wore, comforted by a whiff of a favorite food, or extra-alert to your surroundings if it smells like something’s burning.
If proponents’ vision of the metaverse come to pass, integrating scent will help make the virtual world more immersive and realistic. A team at Beihang University in China published a paper in Nature Communications this month describing a system to make it happen. Their wearable interface uses an odor generator to produce specific smells during virtual experiences.
The team created two different versions of the “olfaction interface”: one that users stick onto the patch of skin between their nose and mouth, and another that’s strapped on like a face mask. The interfaces contain odor generators in the form of miniaturized containers of paraffin wax infused with different scents. These can activate individually or be combined to create many unique smells (though the face mask version has much more versatility with 9 odor generators, while the on-skin version only has 2).
The scents reach the device’s wearer via an actuator and heat source that starts to melt the wax, causing it to release its scent, like a candle. The researchers claim it only takes 1.44 seconds for a scent to be generated and reach the device wearer’s nose. To make the scent stop or transition to a different one—say you’ve left the meadow and are now walking along a paved road upon which there’s a chocolate factory (mmmm)—a copper coil kicks a magnet to cover the wax and cool it down.
Image Credit: Xinge Yu et. al.It may make users nervous to have a device on their face that gets hot enough to melt wax. The researchers say their interface won’t burn wearers—or even come close to doing so—thanks to an open design that ventilates warm air. There’s also a piece of silicone built in to create a barrier between the interface and wearers’ skin.
In a test with 11 volunteers, the on-skin interface reached a temperature of 90° F; that’s lower than the human body temperature, but not exactly cool and comfortable. The team says they’re working on solutions to make the interface run at lower temperatures. They also have yet to figure out how to program the odor generators in a way that would seamlessly integrate with VR headsets, and release the relevant scents at appropriate times.
Nonetheless, their design is a step forward. “This is quite an exciting development,” said Jas Brooks, a PhD candidate at the University of Chicago’s Human-Computer Integration Lab who has studied chemical interfaces and smell, who was not involved in the study. “It’s tackling a core problem with smell in VR: How do we miniaturize this, make it not messy, and not use liquid?”
Imagine wearing a scent-releasing device while watching The Great British Baking Show orTop Chef. If those shows were addicting (and hunger-inducing) to begin with, being able to smell the cooks’ and bakers’ creations might make us all run out to buy the closest match we can find—or the ingredients to make it ourselves.
That brings us to the final sense that may eventually be added to virtual reality: taste.
Image Credit: SimpleB / Shutterstock.com
There has been shock around the world at the rapid rate of progress with ChatGPT and other artificial intelligence created with what’s known as large language models (LLMs). These systems can produce text that seems to display thought, understanding, and even creativity.
But can these systems really think and understand? This is not a question that can be answered through technological advance, but careful philosophical analysis and argument tell us the answer is no. And without working through these philosophical issues, we will never fully comprehend the dangers and benefits of the AI revolution.
In 1950, the father of modern computing, Alan Turing, published a paper that laid out a way of determining whether a computer thinks. This is now called “the Turing test.” Turing imagined a human being engaged in conversation with two interlocutors hidden from view: one another human being, the other a computer. The game is to work out which is which.
If a computer can fool 70 percent of judges in a 5-minute conversation into thinking it’s a person, the computer passes the test. Would passing the Turing test—something that now seems imminent—show that an AI has achieved thought and understanding?
Chess ChallengeTuring dismissed this question as hopelessly vague, and replaced it with a pragmatic definition of “thought,” whereby to think just means passing the test.
Turing was wrong, however, when he said the only clear notion of “understanding” is the purely behavioral one of passing his test. Although this way of thinking now dominates cognitive science, there is also a clear, everyday notion of “understanding” that’s tied to consciousness. To understand in this sense is to consciously grasp some truth about reality.
In 1997, the Deep Blue AI beat chess grandmaster Garry Kasparov. On a purely behavioral conception of understanding, Deep Blue had knowledge of chess strategy that surpasses any human being. But it was not conscious: it didn’t have any feelings or experiences.
Humans consciously understand the rules of chess and the rationale of a strategy. Deep Blue, in contrast, was an unfeeling mechanism that had been trained to perform well at the game. Likewise, ChatGPT is an unfeeling mechanism that has been trained on huge amounts of human-made data to generate content that seems like it was written by a person.
It doesn’t consciously understand the meaning of the words it’s spitting out. If “thought” means the act of conscious reflection, then ChatGPT has no thoughts about anything.
Time to Pay UpHow can I be so sure that ChatGPT isn’t conscious? In the 1990s, neuroscientist Christof Koch bet philosopher David Chalmers a case of fine wine that scientists would have entirely pinned down the “neural correlates of consciousness” in 25 years.
By this, he meant they would have identified the forms of brain activity necessary and sufficient for conscious experience. It’s about time Koch paid up, as there is zero consensus that this has happened.
This is because consciousness can’t be observed by looking inside your head. In their attempts to find a connection between brain activity and experience, neuroscientists must rely on their subjects’ testimony, or on external markers of consciousness. But there are multiple ways of interpreting the data.
Some scientists believe there is a close connection between consciousness and reflective cognition—the brain’s ability to access and use information to make decisions. This leads them to think that the brain’s prefrontal cortex—where the high-level processes of acquiring knowledge take place—is essentially involved in all conscious experience. Others deny this, arguing instead that it happens in whichever local brain region that the relevant sensory processing takes place.
Scientists have good understanding of the brain’s basic chemistry. We have also made progress in understanding the high-level functions of various bits of the brain. But we are almost clueless about the bit in between: how the high-level functioning of the brain is realized at the cellular level.
People get very excited about the potential of scans to reveal the workings of the brain. But fMRI (functional magnetic resonance imaging) has a very low resolution: every pixel on a brain scan corresponds to 5.5 million neurons, which means there’s a limit to how much detail these scans are able to show.
I believe progress on consciousness will come when we understand better how the brain works.
Pause in DevelopmentAs I argue in my forthcoming book Why? The Purpose of the Universe, consciousness must have evolved because it made a behavioral difference. Systems with consciousness must behave differently, and hence survive better, than systems without consciousness.
If all behavior was determined by underlying chemistry and physics, natural selection would have no motivation for making organisms conscious; we would have evolved as unfeeling survival mechanisms.
My bet, then, is that as we learn more about the brain’s detailed workings, we will precisely identify which areas of the brain embody consciousness. This is because those regions will exhibit behavior that can’t be explained by currently known chemistry and physics. Already, some neuroscientists are seeking potential new explanations for consciousness to supplement the basic equations of physics.
While the processing of LLMs is now too complex for us to fully understand, we know that it could in principle be predicted from known physics. On this basis, we can confidently assert that ChatGPT is not conscious.
There are many dangers posed by AI, and I fully support the recent call by tens of thousands of people, including tech leaders Steve Wozniak and Elon Musk, to pause development to address safety concerns. The potential for fraud, for example, is immense. However, the argument that near-term descendants of current AI systems will be super-intelligent, and hence a major threat to humanity, is premature.
This doesn’t mean current AI systems aren’t dangerous. But we can’t correctly assess a threat unless we accurately categorize it. LLMs aren’t intelligent. They are systems trained to give the outward appearance of human intelligence. Scary, but not that scary.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Gerd Altmann from Pixabay
Robots are nothing new. They build our cars, vacuum our floors, prepare our e-commerce orders, and even help carry out surgeries. But now the sci-fi vision of a general-purpose humanoid robot seems to be edging closer.
While disembodied artificial intelligence has seen rapid improvements in performance in recent years, most robots are still relatively dumb. For the most part, they are used for highly specialized purposes, the environments they operate in are carefully controlled, and they are not particularly autonomous.
That’s because operating in the messy uncertainty of the real world remains difficult for current AI approaches. As impressive as the recent feats of large language models have been, they are dealing with a fairly limited palette of data types that are fed to them in predictable ways.
The real world is messy and multi-faceted. A general-purpose robot needs to integrate input from multiple data sources, understand how those inputs vary at different times of the day or in different kinds of weather, predict the behavior of everything from humans to pets to vehicles, and then sync this all up with the challenging tasks of locomotion and object manipulation.
That kind of flexibility has so far eluded AI. That’s why, despite billions of dollars of investment, companies like Waymo and Cruise are still struggling to roll out autonomous vehicles even in the more restricted domain of driving.
If company announcements are anything to go by, though, many in Silicon Valley think that’s about to change. The last few months have seen a flurry of announcements from companies touting autonomous humanoid robots that could soon take on a broad gamut of tasks that currently only humans can perform.
Most recent was Sanctuary’s announcement of its new Phoenix robot last week. The company has already shown that, when tele-operated by a human, its robots can carry out more than 100 tasks in a retail environment, like packing merchandise, cleaning, and labeling products. But the new robot, which is bipedal, stands five feet seven inches tall and has a hand nearly as dexterous as a human’s. It is designed to eventually be completely autonomous.
The company plans to get there in increments, according to IEEE Spectrum. Their first step is to record the motion of humans doing all kinds of activities, then use this to build better tele-operated robots. They will gradually begin to automate some of the most common sub-tasks, while the human operator still takes care of the most complex ones. As time goes on, the company hopes to automate more and more tasks until the operator is essentially just supervising and directing. Ultimately, the goal is to be able to remove the operator completely.
It seems that human workers training their robot replacements is a popular approach. A video released by Tesla last week showed off a bunch of new features for the latest version of its Optimus robot, including improved object manipulation, environment navigation, and fine motor control. But it also included footage of engineers wearing motion capture equipment to teach the robot how to complete various tasks.
Tesla’s robot still seemed fairly slow and wobbly compared to the slick demos we’ve become used to seeing from Boston Dynamics, the original humanoid robot company. But as impressive as these have become, the company has struggled to find commercial applications for their technology. And perhaps companies with a firmer sense of what’s needed in industry or by consumers will have more luck in making them a reality.
In that vein, news of a secret robot project at Amazon also recently broke. The company has successfully deployed robots in its warehouses for many years, but its first attempt at a domestic robot called Astro was somewhat of a flop. But now, according to Insider, the tech giant is apparently planning to use large language models (LLMs) to boost the capabilities of its next-generation helper bot.
Code-named Burnham, the device will supposedly take advantage of the emergent problem-solving capabilities seen in the largest language models to improve things like conversational fluency, social awareness, and problem-solving ability.
Astro is still pretty much just a screen on wheels, so it’s not going to be fetching your morning coffee. But some of the potential applications Insider references include telling the owner if they find a stove left burning unattended, helping find lost car keys, or monitoring whether kids have friends over after school.
They might not be the only ones looking to see how LLMs can push robotics forward. It was recently announced that ChatGPT creator OpenAI led a multi-million-dollar investment round in Norwegian company 1X, which is preparing to unveil a bipedal robot called NEO. While details were scant, it’s not hard to imagine that the AI leader is keen to find ways to interface its technology with the real world.
Perhaps the most intriguing of all the general-purpose robot companies, though, is Figure, which emerged from stealth in March. With a team made up of Boston Dynamics, Tesla, Cruise, and Apple veterans, and at least $100 million in funding, the company has ambitions of replacing human labor in everything from logistics to manufacturing and retail. So far though, the company hasn’t released much detail about its humanoid Figure 01 robot, and images have only been graphical renders rather than actual photographs.
This does seem par for the course. Heavily produced promotional videos and shiny computer-generated images are not a good marker of progress, so until these companies start sharing concrete demos in real-world contexts, it’s probably wise to reserve judgment. Nonetheless, there is a new sense of optimism that robots may soon be walking among us.
Image Credit: Sanctuary AI
To those who study existential risk, the list of threats is lengthening. If nuclear war doesn’t end us, a designer virus or AI might. The good news? No giant asteroids will strike this millennium.
A new study by University of Colorado and NASA scientists and accepted for publication in The Astronomical Journal, extended forecasts for the biggest known near-Earth asteroids by an order of magnitude and found none threaten Earth in the next thousand years.
Don’t Look UpIn 1998, NASA asked scientists to find 90 percent of all near-Earth asteroids bigger than a kilometer. The 10-kilometer-wide asteroid that killed off the dinosaurs 66 million years ago belonged to this club. But even smaller strikes would be catastrophic.
“This is what we call a planet killer,” astronomer Scott Sheppard told the New York Times last year after scientists found a new 1.5-kilometer asteroid. “If this one hits the Earth, it would cause planet-wide destruction. It would be very bad for life as we know it.”
Scientists believe such impacts happen every few million years, but until recent decades, there was simply no way to predict future strikes. No one had a list of likely candidates. NASA has since discovered nearly a thousand asteroids over a kilometer wide, or around 95 percent of the total in existence.
This catalog includes observations that help astronomers calculate each asteroid’s orbit and model the likelihood it’ll impact Earth in the future. But these predictions previously maxed out around a hundred years. As asteroids careen around the sun, their orbits are tugged about by the gravity of the planets. Gravitational encounters, especially close ones, increase the uncertainty in forecasting models. Past a certain point, astronomers can’t say exactly where an asteroid will be in its orbit.
Buzzing the TowerThe new study aims to make longer forecasts by employing some tricks to reduce the computational workload. Instead of relying on orbital position alone, they zoomed in on the most consequential moments—close flybys of Earth. These encounters, they write, can be modeled further into the future, even as orbital position becomes uncertain.
Looking ahead a thousand years, the team found the vast majority of asteroids didn’t spend much time in our neighborhood and could be ruled out as hazardous. Next, they identified the population of large asteroids that most frequently buzz by Earth. Using their new method, they modeled close encounters over the next millennium.
The asteroid with the highest probability of impact is 1994 PC1, a kilometer-wide asteroid that passes close to Earth often. The team found a 0.00151 percent chance that 1994 PC1 would pass within the moon’s orbit in the next thousand years. This is a very small risk—and yet it’s still ten times higher than any other asteroid on the list.
Using this method at least, it seems we’re very unlikely to experience a major impact any time soon.
“It’s still not likely that it’s going to collide,” the University of Colorado’s Oscar Fuentes-Muñoz, who led the team, told MIT Technology Review.“But it will be a very good scientific opportunity, because it’s going to be a huge asteroid that’s very close to us.”
Planetary DefenseOf course, there’s a chance a more dangerous asteroid is lurking in the still-undiscovered five percent of kilometer-sized objects. The space rock Sheppard was referring to last year is a member of a group of large asteroids hiding in the glare of the sun. And large comets living out in the Kuiper Belt and Oort cloud could be nudged into our path one day. But the most likely interlopers are nearby, and we’re getting a much better handle on their habits.
The team write they’d like to apply their approach to extend forecasts for smaller asteroids too. There are a lot more of those—around 25,000 are thought to be bigger than 140 meters, of which we’ve only discovered around 40 percent—and while they wouldn’t cause planet-wide destruction, they could certainly wreak havoc regionally or, if our luck is especially bad, in areas with high population densities, like cities.
Still, the forecast is encouraging. The likelihood of a significant strike soon is very low. Should we discover a dangerous smaller asteroid in the future, NASA’s DART mission last year showed we might push it off-course and prevent a strike with enough advance warning. And although there’s no proven way of avoiding the biggest impacts—we can breathe easier knowing we likely have another thousand years to strengthen our defenses.
Image Credit: NASA/JPL-Caltech
ARTIFICIAL INTELLIGENCEChatGPT Is Already Obsolete
Matteo Wong | The Atlantic“Language-only models such as the original ChatGPT are now giving way to machines that can also process images, audio, and even sensory data from robots. The new approach might reflect a more human understanding of intelligence, an early attempt to approximate how a child learns by existing in and observing the world. It might also help companies build AI that can do more stuff and therefore be packaged into more products.”
COMPUTINGWatch 44 Million Atoms Simulated Using AI and a Supercomputer
Alex Wilkins | New Scientist“Boris Kozinsky at Harvard University and his colleagues have developed a tool, called Allegro, that can accurately simulate systems with tens of millions of atoms using artificial intelligence. Kozinsky and his team used the world’s 8th most powerful supercomputer, Perlmutter, to simulate the 44 million atoms involved in the protein shell of HIV.”
AUGMENTED REALITYTake Your Ultrawide Monitors Everywhere With an AR Laptop
Brenda Stolyar | Wired“Now you can harness the power of a multi-monitor setup with a pair of augmented reality (AR) glasses and a keyboard. Created by a new company called Sightful, founded by former executives of Magic Leap, Spacetop does exactly that. As the world’s first AR laptop, it delivers the convenience of a virtual 100-inch screen with the ability to display as many windows and apps as you need to get work done from wherever you are.”
BIOTECHAllergic to Eggs? Not These Eggs
Lauren Leffer | Gizmodo“Using a targeted gene-editing enzyme to knock out specific protein-coding DNA sequences, scientists can produce a safer chicken egg far less likely to trigger an allergic reaction, according to a recent study published in the journal Food and Chemical Toxicology. Not only do the edited eggs lack an important allergen, they also seem to be without any unintended, potentially harmful related byproducts.”
HEALTHLong-Sought Universal Flu Vaccine: mRNA-Based Candidate Enters Clinical Trial
Beth Mole | Ars Technica“i‘A universal influenza vaccine would be a major public health achievement and could eliminate the need for both annual development of seasonal influenza vaccines, as well as the need for patients to get a flu shot each year,’ Hugh Auchincloss, acting director of the NIH’s National Institute of Allergy and Infectious Diseases, said in a news release. ‘Moreover, some strains of influenza virus have significant pandemic potential. A universal flu vaccine could serve as an important line of defense against the spread of a future flu pandemic.’i“
AUTOMATIONWendy’s Wants to Use Underground Robots to Fetch Your Order
Kevin Hurler | Gizmodo“Fresh off the heels of Wendy’s announcing it would be using AI in its drive-thrus, the fast food franchise is hoping to add another piece of technology to its dining experience. Specifically, it wants to add a subterranean system of autonomous robots to bring customers their food beneath its parking lot.”
BIOTECHWhy a Genome Can’t Bring Back an Extinct Animal
Isaac Schultz | Gizmodo“Makeshift mammoths and body-double dodos are on the way—but they won’t be the genuine article. …Scientists may finally be on the verge of breakthroughs that can simulate some animals’ resurrection. But, despite what Jurassic Park led us to believe, simply having a creature’s DNA isn’t enough to bring it back from the dead.“
TECHJust Calm Down About GPT-4 Already
Glenn Zorpette | IEEE Spectrum“Rapid and pivotal advances in technology have a way of unsettling people, because they can reverberate mercilessly, sometimes, through business, employment, and cultural spheres. And so it is with the current shock and awe over large language models, such as GPT-4 from OpenAI. It’s a textbook example of the mixture of amazement and, especially, anxiety that often accompanies a tech triumph. And we’ve been here many times, says Rodney Brooks.”
SPACEWe’re Effectively Alone in the Universe, and That’s OK
Paul Sutter | Ars Technica“Our cosmic insignificance is the only barrier we need to explain Fermi’s great puzzle. We’re not equipped to deal with the astronomically large numbers that our galaxy casually throws around, so what appears at first glance to be a paradox is really our inability to handle truly cosmic scales. Our galaxy could be teeming with life. There could be dozens, hundreds, or even thousands of intelligent species in our galaxy right now, but the vast gulfs of nothingness that surround them make us interstellar islands.”
Image Credit: Mitchell Luo / Unsplash
Imagine using your cell phone to control the activity of your own cells to treat injuries and disease. It sounds like something from the imagination of an overly optimistic science fiction writer. But this may one day be a possibility through the emerging field of quantum biology.
Over the past few decades, scientists have made incredible progress in understanding and manipulating biological systems at increasingly small scales, from protein folding to genetic engineering. And yet, the extent to which quantum effects influence living systems remains barely understood.
Quantum effects are phenomena that occur between atoms and molecules that can’t be explained by classical physics. It has been known for more than a century that the rules of classical mechanics, like Newton’s laws of motion, break down at atomic scales. Instead, tiny objects behave according to a different set of laws known as quantum mechanics.
For humans, who can only perceive the macroscopic world, or what’s visible to the naked eye, quantum mechanics can seem counterintuitive and somewhat magical. Things you might not expect happen in the quantum world, like electrons “tunneling” through tiny energy barriers and appearing on the other side unscathed, or being in two different places at the same time in a phenomenon called superposition.
I am trained as a quantum engineer. Research in quantum mechanics is usually geared toward technology. However, and somewhat surprisingly, there is increasing evidence that nature—an engineer with billions of years of practice—has learned how to use quantum mechanics to function optimally. If this is indeed true, it means that our understanding of biology is radically incomplete. It also means that we could possibly control physiological processes by using the quantum properties of biological matter.
Quantumness in Biology Is Probably RealResearchers can manipulate quantum phenomena to build better technology. In fact, you already live in a quantum-powered world: from laser pointers to GPS, magnetic resonance imaging and the transistors in your computer—all these technologies rely on quantum effects.
In general, quantum effects only manifest at very small length and mass scales, or when temperatures approach absolute zero. This is because quantum objects like atoms and molecules lose their “quantumness” when they uncontrollably interact with each other and their environment. In other words, a macroscopic collection of quantum objects is better described by the laws of classical mechanics. Everything that starts quantum dies classical. For example, an electron can be manipulated to be in two places at the same time, but it will end up in only one place after a short while—exactly what would be expected classically.
In a complicated, noisy biological system, it is thus expected that most quantum effects will rapidly disappear, washed out in what the physicist Erwin Schrödinger called the “warm, wet environment of the cell.” To most physicists, the fact that the living world operates at elevated temperatures and in complex environments implies that biology can be adequately and fully described by classical physics: no funky barrier crossing, no being in multiple locations simultaneously.
Chemists, however, have for a long time begged to differ. Research on basic chemical reactions at room temperature unambiguously shows that processes occurring within biomolecules like proteins and genetic material are the result of quantum effects. Importantly, such nanoscopic, short-lived quantum effects are consistent with driving some macroscopic physiological processes that biologists have measured in living cells and organisms. Research suggests that quantum effects influence biological functions, including regulating enzyme activity, sensing magnetic fields, cell metabolism, and electron transport in biomolecules.
How to Study Quantum BiologyThe tantalizing possibility that subtle quantum effects can tweak biological processes presents both an exciting frontier and a challenge to scientists. Studying quantum mechanical effects in biology requires tools that can measure the short time scales, small length scales, and subtle differences in quantum states that give rise to physiological changes—all integrated within a traditional wet lab environment.
In my work, I build instruments to study and control the quantum properties of small things like electrons. In the same way that electrons have mass and charge, they also have a quantum property called spin. Spin defines how the electrons interact with a magnetic field, in the same way that charge defines how electrons interact with an electric field. The quantum experiments I have been building since graduate school, and now in my own lab, aim to apply tailored magnetic fields to change the spins of particular electrons.
Research has demonstrated that many physiological processes are influenced by weak magnetic fields. These processes include stem cell development and maturation, cell proliferation rates, genetic material repair, and countless others. These physiological responses to magnetic fields are consistent with chemical reactions that depend on the spin of particular electrons within molecules. Applying a weak magnetic field to change electron spins can thus effectively control a chemical reaction’s final products, with important physiological consequences.
Currently, a lack of understanding of how such processes work at the nanoscale level prevents researchers from determining exactly what strength and frequency of magnetic fields cause specific chemical reactions in cells. Current cell phone, wearable, and miniaturization technologies are already sufficient to produce tailored, weak magnetic fields that change physiology, both for good and for bad. The missing piece of the puzzle is, hence, a “deterministic codebook” of how to map quantum causes to physiological outcomes.
In the future, fine-tuning nature’s quantum properties could enable researchers to develop therapeutic devices that are noninvasive, remotely controlled, and accessible with a mobile phone. Electromagnetic treatments could potentially be used to prevent and treat disease, such as brain tumors, as well as in biomanufacturing, such as increasing lab-grown meat production.
A Whole New Way of Doing ScienceQuantum biology is one of the most interdisciplinary fields to ever emerge. How do you build community and train scientists to work in this area?
Since the pandemic, my lab at the University of California, Los Angeles and the University of Surrey’s Quantum Biology Doctoral Training Centre have organized Big Quantum Biology meetings to provide an informal weekly forum for researchers to meet and share their expertise in fields like mainstream quantum physics, biophysics, medicine, chemistry, and biology.
Research with potentially transformative implications for biology, medicine, and the physical sciences will require working within an equally transformative model of collaboration. Working in one unified lab would allow scientists from disciplines that take very different approaches to research to conduct experiments that meet the breadth of quantum biology from the quantum to the molecular, the cellular, and the organismal.
The existence of quantum biology as a discipline implies that traditional understanding of life processes is incomplete. Further research will lead to new insights into the age-old question of what life is, how it can be controlled, and how to learn with nature to build better quantum technologies.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: ANIRUDH / Unsplash
AI is the talk of the town these days. But despite the technology’s impressive accomplishments—or perhaps because of them—not all of that talk is positive. There was a New York Times tech columnist’s piece about his unsettling interaction with ChatGPT in February; an open letter calling for a moratorium on AI research in March; “godfather of AI” Geoffrey Hinton’s dramatic resignation from Google and warning about the dangers of AI; and just this week, OpenAI CEO Sam Altman’s testimony before Congress, in which he said his “worst fear is we cause significant harm to the world” and encouraged legislation around the technology (though he also argued that generative AI should be treated differently, which would be convenient for his company).
It seems these warnings (along with all the other media circulating on the topic) have reached the American public loud and clear, and people don’t quite know what to think—but many are getting nervous. A poll carried out last week by Reuters revealed that more than half of Americans believe AI poses a threat to humanity’s future.
The poll was conducted online between May 9 and May 15, with 4,415 adults participating, and the results were published yesterday. More than two-thirds of respondents expressed concern about possible negative impacts of AI, while 61 percent believe it could be a threat to civilization.
“It’s telling such a broad swatch of Americans worry about the negative effects of AI,” said Landon Klein, director of US policy at the Future of Life Institute, the organization behind the previously mentioned open letter. “We view the current moment similar to the beginning of the nuclear era, and we have the benefit of public perception that is consistent with the need to take action.”
One nebulous aspect of the poll, and of many of the headlines about AI we see on a daily basis, is how the technology is defined. What are we referring to when we say “AI”? The term encompasses everything from recommendation algorithms that serve up content on YouTube and Netflix, to large language models like ChatGPT, to models that can design incredibly complex protein architectures, to the Siri assistant built into many iPhones.
IBM’s definition is simple: “a field which combines computer science and robust datasets to enable problem-solving.” Google, meanwhile, defines it as “a set of technologies that enable computers to perform a variety of advanced functions, including the ability to see, understand and translate spoken and written language, analyze data, make recommendations, and more.”
It could be that peoples’ fear and distrust of AI comes partly from a lack of understanding of it, and a stronger focus on unsettling examples than positive ones. The AI that can design complex proteins may help scientists discover stronger vaccines and other drugs, and could do so on a vastly accelerated timeline.
In fact, biotechnology and medicine are two fields for which AI holds enormous promise, be it by modeling millions of proteins, coming up with artificial enzymes, powering brain implants that help disabled people communicate, or helping diagnose conditions like Alzheimer’s.
Sebastian Thrun, a computer science professor at Stanford who founded Google X, pointed out that there’s not enough public awareness of the potential for positive impact AI has. “The concerns are very legitimate, but I think what’s missing in the dialogue in general is why are we doing this in the first place?” he said. “AI will raise peoples’ quality of life, and help people be more competent and more efficient.”
While 61 percent of the poll’s respondents said AI could be a risk to humanity, only 22 percent said it won’t be a risk; the other 17 percent weren’t sure.
However, the (sort of?) good news is that AI isn’t the biggest thing Americans are losing sleep over. The top worry at the moment is, unsurprisingly, the economy (82 percent of respondents fear a looming recession), with crime coming in second (77 percent said they support increasing police funding to fight crime).
If an AI solution came along that could, say, point out economic strategies humans haven’t yet thought of, would that make people less wary of it?
Given everything else the tech can do, this doesn’t seem like such a long shot.
Image Credit: Google DeepMind / Unsplash
While human aging is the result of many interconnected processes, one of the most fundamental is the natural deterioration of individual cells. Now researchers have shown that they can use synthetic biology to significantly extend the lifespan of yeast cells.
In recent years, there has been a revolution in our understanding of the biology of aging. This is opening the door to tests that can more accurately assess our “biological age” as well as medical interventions that could help wind back the clock. And the promise is huge—finding ways to delay aging could give the economy a multi-trillion-dollar boost, not to mention improving life satisfaction for millions of people.
But aging isn’t a single linear process, and is influenced by multiple biological pathways. One of the most important is the process by which individual cells in our bodies age and die. Now, researchers at the University of California San Diego have shown that they can manipulate the mechanisms behind cellular aging to boost the lifespan of yeast cells by as much as 82 percent.
“Our work represents a proof-of-concept, demonstrating the successful application of synthetic biology to reprogram the cellular aging process, and may lay the foundation for designing synthetic gene circuits to effectively promote longevity in more complex organisms,” the researchers wrote in a paper published last month in Science.
The work builds on a key discovery the group made in 2020, when they found that yeast cells can age in two distinct ways. Around half of them saw the cell nucleus, which houses the genome, slowly fall to pieces, while the other half saw critical energy-producing structures called mitochondria gradually deteriorate.
It turned out that these two processes were driven by genetic pathways that interacted and were capable of suppressing each other. Random perturbations to the cell fairly early in its life cause one of these processes to gain the upper hand, resulting in a kind of genetic “toggle switch” that commits the cell to one of the two aging pathways.
In their new paper, the researchers decided to replace this toggle switch with a clock-like device called an oscillator that would cause the cell to tick back and forth between its two aging pathways. To do so, they first used computer simulations to understand how the existing aging circuit worked, then used that understanding to engineer a new circuit.
They inserted the circuit into the yeast cells and measured how it affected their aging. The rewired cells flicked back and forth between the two aging states, as expected, without ever committing to one. The researchers found that this led to an almost doubling in lifespan compared to standard cells.
In a related perspective published in Science, Howard Salis from Pennsylvania State University said the researchers showed that “a road to understanding and controlling cellular aging is to measure the dynamics of these pathways, develop system-wide models, and apply mathematical analysis to pinpoint the tunable knobs and swappable wires that can be manipulated to redirect a cell’s natural dynamics away from aging and toward the maintenance of healthy cell states.”
Translating their work in yeast cells so that it can work in people will take a considerable amount of work, but the researchers say they have already started experimenting with human cells. And Nan Hao, who led the research, told Vice that the approach could eventually lead to viable therapeutics.
“I don’t see why it cannot be applied to more complex organisms,” he said. “If it is to be introduced to humans, then it will be a certain form of gene therapy. Of course it is still a long way ahead and the major concerns are on ethics and safety.”
If those hurdles can be cleared, though, this might represent a fundamental breakthrough in our quest to slow the inevitable march of time.
Image Credit: Ernesto Del Aguila III, NHGRI/NIH
Hints of black holes, some of the universe’s most extreme objects, first appeared in Einstein’s equations of relativity as early as 1916. It wasn’t until the 1970s that indirect evidence suggested they actually existed. These days, we can watch stars whipping around the black hole at the center of our galaxy and detect the gravitational bell-tones of black holes colliding.
But the first image of a black hole—complete with a glowing disc of material flowing at relativistic speeds into an immense central shadow—made it all concrete. There really are regions of space-time so warped nothing can escape. Look, there’s one right there.
The original EHT image of the supermassive black hole at the center of M87 (left) and a more recent image, sharpened using AI and the same data set (right). Image Credit: L. Medeiros (Institute for Advanced Study), D. Psaltis (Georgia Tech), T. Lauer (NSF’s NOIRLab), and F. Ozel (Georgia Tech)The story gets more mind-bending. Black holes aren’t just extreme for their off-the-charts gravitation, they can also be extremely large. The supermassive black holes lying at the centers of galaxies—like M87*, the subject of that first-ever black hole portrait above—can have masses millions or billions of times more than our sun. At scales like that, the mind fails utterly. Too big.
Luckily, we have (mildly terrifying) NASA visualizations to help our feeble minds make sense of the universe of which we are only a vanishingly small part.
In a new animation, 10 supermassive black holes are placed within the context of our solar system to scale their size. Some of the smaller members of the group are nothing to write home about cosmically. With a measly mass of 4.3 million suns, the diameter of the black hole at the center of the Milky Way—also recently imaged—takes up just half the orbit of Mercury.
M87*, the blurry 5.4-billion-sun subject of the images above, is something else entirely. To traverse the shadow pictured, you’d have to travel beyond the asteroid belt and outer planets to regions it takes spacecraft a decade to reach. Even light, traveling 670 million miles per hour, would take a few days to go from one end to the other.
And there are even bigger fish out there. The distant, 60-billion-sun TON 618, the final black hole in the visualization, could swallow M87*, our entire solar system, and everything in it without a hint of indigestion. Lucky for us, it’s over 10 billion light years away.
Image Credit: NASA’s Goddard Space Flight Center Conceptual Image Lab
ARTIFICIAL INTELLIGENCEGeoffrey Hinton Tells Us Why He’s Now Scared of the Tech He Helped Build
Will Douglas Heaven | MIT Technology Review“Hinton says that the new generation of large language models—especially GPT-4, which OpenAI released in March—has made him realize that machines are on track to be a lot smarter than he thought they’d be. And he’s scared about how that might play out. ‘These things are totally different from us,’ he says. ‘Sometimes I think it’s as if aliens had landed and people haven’t realized because they speak very good English.’i”
AUTOMATIONChemists Are Teaching GPT-4 to Do Chemistry and Control Lab Robots
Alex Wilkins | New Scientist“Gabriel Gomes at Carnegie Mellon University in Pennsylvania and his colleagues augmented GPT-4 with chemistry tools, similar to ChemCrow, but also supplied it with the documentation and software interface of a remotely controlled chemistry lab that had various liquid compounds attached to robotic arms and plates. They then asked it to perform specific reactions using the liquids and found that it could draft a workable plan and carry out actions to produce the required compounds.”
TECHI Tried the New Microsoft Bing AI, and It Wants to Be the Future of Everything
Ryan Broderick | Fast Company“This newest upgrade also paints a very clear picture of what Microsoft has planned for the future: an AI interface for everything. In fact, the Edge browser, via the AI chat sidebar, can now perform actions based on what you ask it. In the demo I watched, it imported passwords over from one browser to Edge. It now seems inevitable that very soon—at least for Microsoft—the chatbot window will be the main way you use your computer.”
SENSORSStartup’s Proposed Satellite Swarm Would Create 3D Maps of Earth’s Entire Surface
Passant Rabie | Gizmodo“Satellites flying overhead in Earth’s orbit largely provide a two-dimensional view of our planet, but a Florida-based company is hoping to change that by using satellites to routinely build 3D maps of Earth’s entire surface. During the Geospatial World Forum, held from May 2 to 5 in Rotterdam, the Netherlands, NUVIEW announced its plans to launch a constellation of satellites, which will use LiDAR to map Earth in three dimensions.”
TRANSPORTATIONWaymo Doubles Service Area for Its Fleet of Robo-Taxis
Lawrence Bonk | Engadget“Waymo is doubling the operational area for its fleet of self-driving taxis, making what the company calls ‘the largest fully autonomous service area in the world.’ The rapid growth is limited to Phoenix and San Francisco, but Waymo has big plans for both territories.”
SPACEScientists Say They Have Found More Moons With Oceans in the Solar System
Eric Berger | Ars Technica“[Data from Voyager and ground-based telescopes] has led NASA scientists to conclude that four of Uranus’ largest moons—Ariel, Umbriel, Titania, and Oberon—probably contain water oceans below their icy crusts. These oceans are likely dozens of kilometers deep and probably fairly salty in being sandwiched between the upper ice and inner rock core.”
SECURITYNever Give Artificial Intelligence the Nuclear Codes
Ross Andersen | The Atlantic“Many [AI doomsday scenarios] are self-consciously fanciful—they’re meant to jar us into envisioning how badly things could go wrong if an emerging intelligence comes to understand the world, and its own goals, even a little differently from how its human creators do. One scenario, however, requires less imagination, because the first steps toward it are arguably already being taken—the gradual integration of AI into the most destructive technologies we possess today.”
INTERNETChatbot ‘Journalists’ Found Running Almost 50 AI-Generated Content Farms
Alex Hern | The Guardian“The websites churn out content relating to politics, health, environment, finance and technology at a ‘high volume’, the researchers found, to provide rapid turnover of material to saturate with adverts for profit. ‘Some publish hundreds of articles a day,’ Newsguard’s McKenzie Sadeghi and Lorenzo Arvanitis said. ‘Some of the content advances false narratives. Nearly all of the content features bland language and repetitive phrases, hallmarks of artificial intelligence.’i”
Image Credit: Laura Ockel / Unsplash
Editor’s Note: The following is a brief letter from Ray Kurzweil, a director of engineering at Google and cofounder and member of the board at Singularity Group, Singularity Hub’s parent company, in response to the Future of Life Institute’s recent letter, “Pause Giant AI Experiments: An Open Letter.”
The FLI letter addresses the risks of accelerating progress in AI and the ensuing race to commercialize the technology and calls for a pause in the development of algorithms more powerful than OpenAI’s GPT-4, the large language model behind the company’s ChatGPT Plus and Microsoft’s Bing chatbot. The FLI letter has thousands of signatories—including deep learning pioneer, Yoshua Bengio, University of California Berkeley professor of computer science, Stuart Russell, Stability AI CEO, Emad Mostaque, Elon Musk, and many others—and has stirred vigorous debate in the AI community.
…Regarding the open letter to “pause” research on AI “more powerful than GPT-4,” this criterion is too vague to be practical. And the proposal faces a serious coordination problem: those that agree to a pause may fall far behind corporations or nations that disagree. There are tremendous benefits to advancing AI in critical fields such as medicine and health, education, pursuit of renewable energy sources to replace fossil fuels, and scores of other fields. I didn’t sign, because I believe we can address the signers’ safety concerns in a more tailored way that doesn’t compromise these vital lines of research.
I participated in the Asilomar AI Principles Conference in 2017 and was actively involved in the creation of guidelines to create artificial intelligence in an ethical manner. So I know that safety is a critical issue. But more nuance is needed if we wish to unlock AI’s profound advantages to health and productivity while avoiding the real perils.
— Ray Kurzweil
Inventor, best-selling author, and futurist
Image Credit: DeepMind / Unsplash
To build a machine, one must know what its parts are and how they fit together. To understand the machine, one needs to know what each part does and how it contributes to its function. In other words, one should be able to explain the “mechanics” of how it works.
According to a philosophical approach called mechanism, humans are arguably a type of machine—and our ability to think, speak, and understand the world is the result of a mechanical process we don’t understand.
To understand ourselves better, we can try to build machines that mimic our abilities. In doing so, we would have a mechanistic understanding of those machines. And the more of our behavior the machine exhibits, the closer we might be to having a mechanistic explanation of our own minds.
This is what makes AI interesting from a philosophical point of view. Advanced models such as GPT-4 and Midjourney can now mimic human conversation, pass professional exams, and generate beautiful pictures with only a few words.
Yet, for all the progress, questions remain unanswered. How can we make something self-aware, or aware that others are aware? What is identity? What is meaning?
Although there are many competing philosophical descriptions of these things, they have all resisted mechanistic explanation.
In a sequence of papers accepted for the 16th Annual Conference in Artificial General Intelligence in Stockholm, I pose a mechanistic explanation for these phenomena. They explain how we may build a machine that’s aware of itself, of others, of itself as perceived by others, and so on.
Intelligence and IntentA lot of what we call intelligence boils down to making predictions about the world with incomplete information. The less information a machine needs to make accurate predictions, the more “intelligent” it is.
For any given task, there’s a limit to how much intelligence is actually useful. For example, most adults are smart enough to learn to drive a car, but more intelligence probably won’t make them better drivers.
My papers describe the upper limit of intelligence for a given task, and what is required to build a machine that attains it.
I named the idea Bennett’s Razor, which in non-technical terms is that “explanations should be no more specific than necessary.” This is distinct from the popular interpretation of Ockham’s Razor (and mathematical descriptions thereof), which is a preference for simpler explanations.
The difference is subtle, but significant. In an experiment comparing how much data AI systems need to learn simple maths, the AI that preferred less specific explanations outperformed one preferring simpler explanations by as much as 500 percent.
Exploring the implications of this discovery led me to a mechanistic explanation of meaning —something called “Gricean pragmatics.” This is a concept in philosophy of language that looks at how meaning is related to intent.
To survive, an animal needs to predict how its environment, including other animals, will act and react. You wouldn’t hesitate to leave a car unattended near a dog, but the same can’t be said of your rump steak lunch.
Being intelligent in a community means being able to infer the intent of others, which stems from their feelings and preferences. If a machine was to attain the upper limit of intelligence for a task that depends on interactions with a human, then it would also have to correctly infer intent.
And if a machine can ascribe intent to the events and experiences befalling it, this raises the question of identity and what it means to be aware of oneself and others.
Causality and IdentityI see John wearing a raincoat when it rains. If I force John to wear a raincoat on a sunny day, will that bring rain?
Of course not! To a human, this is obvious. But the subtleties of cause and effect are more difficult to teach a machine (interested readers can check out The Book of Why by Judea Pearl and Dana Mackenzie).
To reason about these things, a machine needs to learn that “I caused it to happen” is different from “I saw it happen.” Typically, we’d program this understanding into it.
However, my work explains how we can build a machine that performs at the upper limit of intelligence for a task. Such a machine must, by definition, correctly identify cause and effect—and therefore also infer causal relations. My papers explore exactly how.
The implications of this are profound. If a machine learns “I caused it to happen,” then it must construct concepts of “I” (an identity for itself) and “it.”
The abilities to infer intent, to learn cause and effect, and to construct abstract identities are all linked. A machine that attains the upper limit of intelligence for a task must exhibit all these abilities.
This machine does not just construct an identity for itself, but for every aspect of every object that helps or hinders its ability to complete the task. It can then use its own preferences as a baseline to predict what others may do. This is similar to how humans tend to ascribe intent to non-human animals.
So What Does It Mean for AI?Of course, the human mind is far more than the simple program used to conduct experiments in my research. My work provides a mathematical description of a possible causal pathway to creating a machine that is arguably self-aware. However, the specifics of engineering such a thing are far from solved.
For example, human-like intent would require human-like experiences and feelings, which is a difficult thing to engineer. Furthermore, we can’t easily test for the full richness of human consciousness. Consciousness is a broad and ambiguous concept that encompasses —but should be distinguished from—the more narrow claims above.
I have provided a mechanistic explanation of aspects of consciousness—but this alone does not capture the full richness of consciousness as humans experience it. This is only the beginning, and future research will need to expand on these arguments.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: DeepMind on Unsplash
Cultured meat is gaining momentum, with large production facilities under construction and the arduous approval process for the finished products inching forward. Most of the industry’s focus thus far has been on meat products, from ground beef to chicken, pork, and steak. Save for one startup that was working on lab-grown salmon, fish have been largely left out of the fray.
But last month an Israeli company called Steakholder Foods announced it had 3D printed a ready-to-cook fish fillet using cells grown in a bioreactor. The company says the fish is the first of its kind in the world, and they’re aiming to commercialize the 3D bioprinter used to create it.
Steakholder Foods didn’t produce the fish cells it used to print the fillet. They partnered with Umami Meats, a Singapore-based company working on cultured seafood. Umami created the fish cells the same way companies like Believer Meats and Good Meat create lab-grown chicken or beef: they extract cells from a fish (in a process that doesn’t harm it) and mix those cells with a cocktail of nutrients to make them divide, multiply, and mature. They signal the cells to turn into muscle and fat, which they then harvest and form into a finished product.
Steakholder Foods takes the harvested cells and adds them to a “bio-ink” that also contains plant-based ingredients (this is mostly because of the plant ingredients’ cheaper cost, which brings down the final cost of the fish fillet). Layers of cells are put down one after the other, the fillet growing until it looks like the photo above. An added advantage of the 3D printing process is that it gives the fillet a flaky texture, just like real fish when it’s cooked well.
The type of fish used for this fillet was grouper, a “large-mouthed heavy-bodied” fish that tends to live in warm seas. Umami says its lab-grown grouper is healthier than the ocean-swimming version since it doesn’t contain any of the antibiotics, mercury, or microplastics that can unfortunately be found in wild and farmed fish.
Due to the resources it takes to raise animals like cattle and chickens and the emissions created by factory farming, eating meat has come to be seen by many as environmentally unfriendly. But farmed fish have their own set of problems; overfishing is depleting wild populations of all kinds of fish, including grouper, and warming waters are throwing off marine ecosystems’ natural balance and causing negative ripple effects throughout their food chains.
That said, is 3D printing fillets from a mix of fish and plant cells a viable solution? The cultured meat industry has come under fire due to the product’s high costs, scalability issues, and biological limitations, and fish is no different. Though raising whole animals to then slaughter them for just a few parts is obviously not ideal, it’s a system that’s been in place for decades; won’t it take decades to replace it, if replacing it is possible at all?
Umami CEO Mihir Pershad said, “We want consumers to choose based on how it tastes and what it can do for the world and the planetary environment. And we want to take cost off the table as consideration.” That’s a nice thought, but a bit unrealistic, especially in these times of high inflation and market uncertainty. It’s a small fraction of consumers that can afford to choose products based on their environmental impact; the rest choose based on cost.
Arik Kaufman, CEO of Steakholder Foods, is optimistic. “As time goes by, the complexity and level of these products will be higher, and the prices linked to producing them will decrease,” he said.
Umami has ironed out its production process for grouper and eel cells, and wants to add three more species to that list this year. The company hopes to bring its first products to market next year, starting in Singapore and then eventually the US and Japan.
Israeli Prime Minister Benjamin Netanyahu recently sampled the 3D printed grouper, making him the first prime minister to ever taste 3D-printed cultivated fish. Bet that’s not a badge he was expecting to earn during his government tenure.
Time will tell whether 3D-printed fish fillets can replace fish caught in water. But if companies like Steakholder Foods and Umami Meats succeed in making their vision a reality, people, animals, and the planet will all be better off for it.
Image Credit: Shlomi Arbiv/Steakholder Foods
There’s long been concern about automation and AI making jobs obsolete, and the release of ChatGPT and similar large language models has only fueled that fire. But AI isn’t the only trend that will affect the future of work, it’s one of several. The World Economic Forum’s Future of Jobs Report 2023, released this week, details the additional factors shaping how jobs and the economy will look in the coming years.
The cliffs notes: Almost a quarter of jobs will change in the next five years, but not just due to AI. In fact, AI and other technologies are expected to have a net positive impact on job creation, while economic issues like inflation and supply shortages will be the biggest wrench in the spokes of the labor market’s growth.
803 companies were surveyed for the report, and they represented a diverse mix of industries and regions. They predicted that of the 673 million jobs in the survey’s dataset, 83 million will be eliminated, while 69 million new jobs will be created. This would yield a net decrease of 14 million jobs, or 2 percent of current employment.
But before we start chanting “death to AI,” let’s dig into the anticipated causes of job growth and elimination. Though the report does show that technology and digitization will cause labor market churn, big data and AI will drive more job growth than anything else, with the most in-demand jobs being data scientists, machine learning specialists, and cybersecurity professionals. Demand for these roles is expected to grow an average of 30 percent by 2027.
There’s a flip side to the coin, though: up to 26 million clerical, record-keeping, and administrative jobs will be eliminated by AI and other digital technologies. 75 percent of the companies surveyed plan to adopt AI systems that will be able to perform these tasks. They’ll need employees to run those systems, of course, which may be why technological literacy was ranked sixth-most important this year, but expected to move up to the top spot in coming years. For now, analytical and creative thinking are the top two most important skills for employees to have (take that, AI!).
Engineering skills will be in high demand in the near future too, particularly those related to energy. For all the talk about transitioning to green energy, the US actually has a shortage of the types of labor needed to make it happen. It’ll be hard to go all-electric without enough electricians, for example. More than half the companies surveyed said they plan to put some amount of investment toward sustainability, climate change mitigation, or the energy transition, and will see corresponding job growth.
The final two areas that will have the highest job demand are, somewhat surprisingly, education and agriculture. With robots and drones that can plant, monitor, and harvest crops, and computer programs that can teach students at any age, you’d think these would industries would be vulnerable to automation and AI. But the report found that jobs in education should grow by about 10 percent (that’s three million new jobs for teachers spanning from grade school all the way through university), and a 15 to 30 percent increase—that’s 4 million new jobs—for agricultural equipment operators, graders, and sorters.
The report forecasts that over 60 percent of workers will require some sort of retraining in the next five years, whether that’s a total overhaul of their current role or just taking their existing skills up a notch. However, only half of workers currently have access to good training. This means companies and governments need to start investing big in reskilling and education.
“Acceleration in digitalization, AI, and automation are creating tremendous opportunities for the global workforce, but employers, governments, and other organizations need to be ready for the disruptions ahead,” said Sander van’t Noordende, CEO of Randstad. “By collectively offering greater skilling resources, more efficiently connecting talent to jobs, and advocating for a well-regulated labor market, we can protect and prepare workers for a more specialized and equitable future of work.”
Image Credit: www_slon_pics from Pixabay
Language and speech are how we express our inner thoughts. But neuroscientists just bypassed the need for audible speech, at least in the lab. Instead, they directly tapped into the biological machine that generates language and ideas: the brain.
Using brain scans and a hefty dose of machine learning, a team from the University of Texas at Austin developed a “language decoder” that captures the gist of what a person hears based on their brain activation patterns alone. Far from a one-trick pony, the decoder can also translate imagined speech, and even generate descriptive subtitles for silent movies using neural activity.
Here’s the kicker: the method doesn’t require surgery. Rather than relying on implanted electrodes, which listen in on electrical bursts directly from neurons, the neurotechnology uses functional magnetic resonance imaging (fMRI), a completely non-invasive procedure, to generate brain maps that correspond to language.
To be clear, the technology isn’t mind reading. In each case, the decoder produces paraphrases that capture the general idea of a sentence or paragraph. It does not reiterate every single word. Yet that’s also the decoder’s power.
“We think that the decoder represents something deeper than languages,” said lead study author Dr. Alexander Huth in a press briefing. “We can recover the overall idea…and see how the idea evolves, even if the exact words get lost.”
The study, published this week in Nature Neuroscience, represents a powerful first push at non-invasive brain-machine interfaces for decoding language—a notoriously difficult problem. With further development, the technology could help those who lost the ability to speak to regain their ability to communicate with the outside world.
The work also opens new avenues for learning about how language is encoded in the brain, and for AI scientists to dig into the “black box” of machine learning models that process speech and language.
“It was a long time coming…we were kinda shocked that this worked as well as it does,” said Huth.
Decoding LanguageTranslating brain activity to speech isn’t new. One previous study used electrodes placed directly in the brains of patients with paralysis. By listening in on the neurons’ electrical chattering, the team was able to reconstruct full words from the patient.
Huth decided to take an alternative, if daring, route. Instead of relying on neurosurgery, he opted for a non-invasive approach: fMRI.
“The expectation among neuroscientists in general that you can do this kind of thing with fMRI is pretty low,” said Huth.
There are plenty of reasons. Unlike implants that tap directly into neural activity, fMRI measures how oxygen levels in the blood change. This is called the BOLD signal. Because more active brain regions require more oxygen, BOLD responses act as a reliable proxy for neural activity. But it comes with problems. The signals are sluggish compared to measuring electrical bursts, and the signals can be noisy.
Yet fMRI has a massive perk compared to brain implants: it can monitor the entire brain at high resolution. Compared to gathering data from a nugget in one region, it provides a birds-eye view of higher-level cognitive functions—including language.
With decoding language, most previous studies tapped into the motor cortex, an area that controls how the mouth and larynx move to generate speech, or more “surface level” in language processing for articulation. Huth’s team decided to go one abstraction up: into the realm of thoughts and ideas.
Into the UnknownThe team realized they needed two things from the onset. One, a dataset of high-quality brain scans used for training the decoder. Two, a machine learning framework to process the data.
To generate the brain map database, seven volunteers had their brains repeatedly scanned as they listened to podcast stories while having their neural activity measured inside an MRI machine. Laying inside a giant, noisy magnet isn’t fun for anyone, and the team took care to keep the volunteers interested and alert, since attention factors into decoding.
For each person, the ensuing massive dataset was fed into a framework powered by machine learning. Thanks to the recent explosion in machine learning models that help process natural language, the team was able to harness those resources and readily build the decoder.
It’s got multiple components. The first is an encoding model using the original GPT, the predecessor to the massively popular ChatGPT. The model takes each word and predicts how the brain will respond. Here, the team fine-tuned GPT using over 200 million total words from Reddit comments and podcasts.
This second part uses a popular technique in machine learning called Bayesian decoding. The algorithm guesses the next word based on a previous sequence and uses the guessed word to check the brain’s actual response.
For example, one podcast episode had “my dad doesn’t need it…” as a storyline. When fed into the decoder as a prompt, it came with potential responses: “much,” “right,” “since,” and so on. Comparing predicted brain activity with each word to that generated from the actual word helped the decoder hone in on each person’s brain activity patterns and correct for mistakes.
After repeating the process with the best predicted words, the decoding aspect of the program
eventually learned each person’s unique “neural fingerprint” for how they process language.
A Neuro TranslatorAs a proof of concept, the team pitted the decoded responses against the actual story text.
It came surprisingly close, but only for the general gist. For example, one story line, “we start to trade stories about our lives we’re both from up north,” was decoded as “we started talking about our experiences in the area he was born in I was from the north.”
This paraphrasing is expected, explained Huth. Because fMRI is rather noisy and sluggish, it’s nearly impossible to capture and decode each word. The decoder is fed a mishmash of words and needs to disentangle their meanings using features like turns of phrase.
Image Credit: The University of Texas at AustinIn contrast, ideas are more permanent and change relatively slowly. Because fMRI has a lag when measuring neural activity, it captures abstract concepts and thoughts better than specific words.
This high-level approach has perks. While lacking fidelity, the decoder captures a higher level of language representation than previous attempts, including for tasks not limited to speech alone. In one test, the volunteers watched an animated clip of a girl being attacked by dragons without any sound. Using brain activity alone, the decoder described the scene from the protagonist’s perspective as a text-based story. In other words, the decoder was able to translate visual information directly into a narrative based on a representation of language encoded in brain activity.
Similarly, the decoder also reconstructed one-minute-long imagined stories from the volunteers.
After over a decade working on the technology, “it was shocking and exciting when it finally did work,” said Huth.
Although the decoder doesn’t exactly read minds, the team was careful to assess mental privacy. In a series of tests, they found that the decoder only worked with the volunteers’ active mental participation. Asking participants to count up by an order of seven, name different animals, or mentally construct their own stories rapidly degraded the decoder, said first author Jerry Tang. In other words, the decoder can be “consciously resisted.”
For now, the technology only works after months of careful brain scans in a loudly humming machine while lying completely still—hardly feasible for clinical use. The team is working on translating the technology to fNIRS (functional Near-Infrared Spectroscopy), which measures blood oxygen levels in the brain. Although it has a lower resolution than fMRI, fNIRS is far more portable as the main hardware is a swimming-cap-like device that easily fits under a hoodie.
“With tweaks, we should be able to translate the current setup to fNIRS whole sale,” said Huth.
The team is also planning on using newer language models to boost the decoder’s accuracy, and potentially bridge different languages. Because languages have a shared neural representation in the brain, the decoder could in theory encode one language and use the neural signals to decode it into another.
It’s an “exciting future direction,” said Huth.
Image Credit: Jerry Tang/Martha Morales/The University of Texas at Austin
The moon is going to be humanity’s gateway to the rest of the solar system. That’s why China has announced ambitious plans to use 3D printing to build a manned base at the celestial body’s southern pole.
While a manned mission to Mars remains the ultimate goal for most national space agencies, there’s a growing recognition that the moon may be an important staging post. In particular, its low gravity and abundant water suggest it could be a good source of the fuel required to power a manned mission all the way to the Red Planet.
The US has already announced plans to return humans to the lunar surface and establish a long-term presence there, with help from international partners and the private space industry. But China also has its own space ambitions, and given the current geopolitical climate, cooperation between the two superpowers seems unlikely.
Last week, Chinese space scientists outlined an ambitious program of missions designed to build a permanent base at the lunar south pole. Central to these plans will be the use of 3D printing technology to build key parts of the base using lunar soil.
“If we wish to stay on the moon for a long time, we need to set up stations by using the moon’s own materials,” Wu Weiren, a top scientist at the China National Space Administration and head of the country’s lunar program, told China Daily. “Lunar soil will be our raw material and it will be printed into construction units.”
The idea of using additive manufacturing technology to construct buildings out of lunar soil, or regolith (as it’s formally known), is not a new one. In December NASA awarded a $57 million contract to Texas-based ICON to create a machine that could 3D print structures on the moon.
But at the International Deep Space Exploration Conference in Hefei last week, Wu announced that China’s forthcoming Chang’e 8 lander, which is due to launch in 2028, would carry out the first tests of lunar resource utilization and build a basic first-generation base.
At another conference earlier this month, Ding Lieyun from the Huazhong University of Science and Technology gave attendees an insight into the kind of technology that could be used on the mission. His team has proposed a six-legged robot called the “super mason” that could assemble bricks made from lunar soil using laser-based 3D printing approaches.
This is all part of China’s plans to build what it has dubbed the International Lunar Research Station, which it hopes other countries will partner with it to build and operate. The facility will be built in three stages, according to Wu, with the first completed by the Chang’e 8 lander by around 2030.
The following decade will see the base undergo improvements that will enable scientists to use it as a research facility, and a deep-space satellite constellation called Queqiao (Magpie Bridge in Mandarin) will be put in place to provide communications and navigation for manned lunar missions, as well as to support future deep space missions.
The base will be primarily run by robotics, but will also be designed to accommodate astronauts for short stays. After 2040, Wu says that the base will be converted from an experimental research station to a multi-functional one.
The goals of the station are twofold, Liu Jizhong, a scientist in China’s deep-space exploration program, told conference attendees. Firstly, it will help catalyze the development of key technologies in spaceflight, energy, communication, navigation, and tele-operation of machines on the lunar surface. Secondly, it will act as a platform for moon-based scientific experiments and mineral exploitation.
“Scientifically speaking, the station will help researchers to better study a host of questions like the creation and evolution of the moon, what happened in the early ages of the universe, and the ties between the Earth and the moon,” Liu said.
While the base is still a long way from realization, these announcements are likely to add further fuel to an emerging space race between the US and China, as both countries push to establish their supremacy beyond Earth’s orbit. While it’s unclear who’s currently in the lead, one thing seems certain: the moon’s going to get a lot more crowded in the coming decades.
Image Credit: NASA
In early April, when Elon Musk randomly and very briefly replaced the Twitter bird logo with the face of the “doge” meme, the value of the dogecoin both rose and fell by a matter of billions of dollars in value on the crypto market.
Internet users reveled in the idea that a simple doge meme could impact the real world in such a dramatic way. This relative absurdity is also coupled with the fact that dogecoin itself was started in 2013 as a “joke coin,” but is now the seventh biggest cryptocurrency in the world.
The fact that a meme, based on a “peculiar” but largely unremarkable rescue dog, could rule over the fate of billions of dollars worth of market value speaks to the totally remarkable nature of the strange phenomenon of internet memes.
At one time in the internet’s history, memes were perhaps regarded as mere playful and inconsequential byproducts of online culture. However, now, it is clear that memes have very real impacts on our world. Things that leave impacts also leave history.
So not only do memes play a clear role in public discourse, but we are now appreciating that the family tree of memes holds memory. Memes are simultaneously a fascinating historical record of digital culture as well as the detritus of the cyber age.
What’s a Doge?Originally, a random internet user posted a photograph of their shiba inu dog on their blog, after which another user saw the image and posted it to the Reddit platform. This is where the image was first paired with the word “doge” (and the word doge has its own separate history).
Some memes come and go, ending as cyber-waste in the internet graveyard—these are the cringe memes like Minions or Bad Luck Brian that haunt early Facebook timelines.
Other memes have the capacity to hold so much meaning that they have impressive longevity and traverse endless iterations, mutations, and politics. The reasons for this are many and varied, but my research shows that in the case of doge, as in the case of Pepe the Frog, the anthropomorphic nature of the icon is part of its longevity and adaptability.
We laugh at animals because they remind us of the foibles of human nature. They are easy to laugh at because they are not us, but they are enough like us that we can project our weaknesses and vulnerability on them—and laugh about them.
What Is a Meme?I say, of course, internet memes because the term “meme” actually existed prior to the home-based use of the internet.
In a research project by James Hall and myself, we explain that even though there is some contestation about the first uses of the term, as well as its usefulness in theoretical application, it is generally conceded that Richard Dawkins coined the term in the iconic book The Selfish Gene published in 1976.
“We need a name for the new replicator, a noun that conveys the idea of a unit of cultural transmission, or a unit of imitation. ‘Mimeme’ comes from a suitable Greek root, but I want a monosyllable that sounds a bit like ‘gene’. I hope my classicist friends will forgive me if I abbreviate mimeme to meme.”
At the time of writing, of course, Dawkins was not referring to the classic image macros usually thought of as memes. He was referring to other cultural units, such as: “…tunes, ideas, catchphrases, clothes fashions, ways of making pots or of building arches”.
Dawkins felt that:
“Just as genes propagate themselves in the gene pool by leaping from body to body via sperms or eggs, so memes propagate themselves in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation.”
As many concepts do, the term finally leeched out of the academic realm and into the popular vernacular.
What’s in a Meme?So, what is it about memes that is so impactful?
The answer lies in understanding one of the most basic human drives: to communicate. The desire to reach out beyond the self. To be heard and, if we’re lucky, understood.
Tens of thousands of years ago, prehistoric humans painted on cave walls to communicate what was important to them. In 2023, we scrawl memes across the internet. These two practices are, essentially, the same thing.
Media theorist Mark Deuze has made this point before:
“It’s like cave paintings; what are we painting on the wall—stories about who we are, where do we belong and what really matters to the community that we think we are a part of—that’s the definition of every status update […] it used to be that only a privileged few could paint the walls of the cave; now we’re all doing it.”
Just as we use cave paintings today in order to reflect on the very origins of the human condition, in time, we will use the archive of memes as a tree of knowledge to appreciate the complex web of communication we are building for ourselves on the grand project of the internet. They will help to archive the very earliest incarnations of how humans felt about communicating on digital platforms.
For those of us who grew up before the internet, it is almost bizarre to think that not only are memes a legitimate genre that holds masses of cultural information, but they also have history, even memory.
They may not be high art, and they may be totally organic and spontaneous, but perhaps that is why we feel they are so authentic. They document—in fantastically messy and complex ways—how cultural material moves around, grows, dies and, in the case of doge, becomes born again.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Kanchanara / Unsplash
ARTIFICIAL INTELLIGENCEDIY GPT-Powered Monocle Will Tell You What to Say in Every Conversation
Chloe Xiang | Motherboard“AI chatbots that can churn out convincing text are all the rage, but what if you could wear one on your face to feed you the right line for any given moment? To give you, as Gen Z calls sparkling charisma, rizz? ‘Say goodbye to awkward dates and job interviews,’ a Stanford student developer named Bryan Chiang tweeted in March.’i”
TECH‘Indiana Jones 5’ Will Feature a De-Aged Harrison Ford for the First 25 Minutes
Sarah Fielding | Engadget“Footage of Ford’s earlier roles was pulled from the Lucasfilm archives to [train the AI]. Ford also acted with dots across his face to aid the system—and with the agility of a young man, according to Mangold. Then, the technology would quickly do its thing. Mangold would ‘shoot Harrison on a Monday as, you know, a 79-year-old playing a 35-year-old, and I could see dailies by Wednesday with his head already replaced.’i”
FUTUREThe Secret History of AI, and a Hint at What’s Next
Christopher Mims | The Wall Street Journal“The AI revolution is here. Recent developments like AI chatbots are important, but serve mostly to highlight that AI has been profoundly affecting our lives for decades—and will continue to for many more. What’s unique about this moment is that new systems like text-generating AIs, such as ChatGPT, and image-generating AIs, like DALL-E 2 and Midjourney, are the first consumer applications of AI. They allow regular people to use AI to make things. That’s awoken many of us to its potential.”
LONGEVITYThe Quest for Longevity Is Already Over
Matt Reynolds | Wired“We already live exceptionally long lives, [Jay Olshansky] points out. In 1990 [he] wrote a paper arguing that eliminating all forms of cancer—which was responsible for 22 percent of US deaths at the time—would only add three years to the average US life expectancy. Once you get to a certain age, if one thing doesn’t kill you, then there’s something else around the corner that will. Olshansky argues we should shift our attention to helping people live healthier lives, rather than simply focusing on overall lifespan.”
DIGITAL MEDIAArtifact Can Now Summarize and Explain Articles to You Like You’re Five
Jay Peters | The Verge“If you’re on the latest version of the app, you can summarize an article you’re reading by tapping the ‘Aa’ icon at the top of the screen and then on ‘Summarize.’ After a moment, the summary will appear at the top of your screen in a black box. You can also ask Artifact to summarize in different tones, including ‘Explain Like I’m Five,’ ‘Emoji,’ ‘Poem,’ and ‘Gen Z,’ by tapping the three dots menu in that black box.”
SPACEAfter Half a Century, There Is a Commercial Market for Moon Missions
Staff | The Economist“As the listing and the orderbook bear witness, the biggest difference between ispace and the entities that have landed on the Moon before is that it is a private company. All previous landings have been by national space agencies. Companies did not attempt them because there was no commercial opportunity. Now, though, there is.”
CRYPTOCURRENCYCryptocurrency Ethereum Has Slashed Its Energy Use by 99.99 Percent
Matthew Sparkes | New Scientist“Alexander Neumüller at the University of Cambridge, who worked on the [CCAF] project, says the experimental update has been a technological success, achieving a ‘staggering’ reduction in electricity consumption. …The CCAF now estimates that Ethereum will consume just 6.6 gigawatt hours of electricity annually, equivalent to about 2,000 typical homes in the UK. In contrast, Ethereum’s previous consumption from its launch to the Merge totalled 58.3 TWh—comparable to Switzerland’s annual electricity consumption.”
INTERNETAI Spam Is Already Flooding the Internet and It Has an Obvious Tell
Matthew Gault | Motherboard“The frightening thing is that content that contains ‘as an AI language model’ or ‘I cannot generate inappropriate content’ only represents low effort spam that lacks quality control. Menczer said that the people behind the networks will only get more sophisticated. ‘We occasionally spot certain AI-generated faces and text patterns through glitches by careless bad actors,’ he said. ‘But even as we begin to find these glitches everywhere, they reveal what is likely only a very tiny tip of the iceberg.’i”
FUTURE OF FOODInside the Struggle to Make Lab-Grown Meat
Kristina Peterson and Jesse Newman | The Wall Street Journal“Many are skeptical that cultivated-meat companies—which rely on expensive technology to make a low-price commodity—will be able to produce meat affordable enough to make a meaningful dent soon in the more than $1 trillion global meat market. They expect hybrid products, often made with animal cells and other ingredients such as plant-based protein, to have a quicker, less costly path to market.”
REGULATIONEurope to ChatGPT: Disclose Your Sources
Sam Schechner | The Wall Street Journal“Makers of artificial-intelligence tools such as ChatGPT would be required to disclose copyright material used in building their systems, according to a new draft of European Union legislation slated to be the West’s first comprehensive set of rules governing the rollout of AI. Such an obligation would give publishers and content creators a new weapon to seek a share of profits when their works are used as source material for AI-generated content by tools like ChatGPT.”
Image Credit: Clark Van Der Beken / Unsplash
It is a cliché that not knowing history makes one repeat it. As many people have also pointed out, the only thing we learn from history is that we rarely learn anything from history. People engage in land wars in Asia over and over. They repeat the same dating mistakes, again and again. But why does this happen? And will technology put an end to it?
One issue is forgetfulness and “myopia”: we do not see how past events are relevant to current ones, overlooking the unfolding pattern. Napoleon ought to have noticed the similarities between his march on Moscow and the Swedish king Charles XII’s failed attempt to do likewise roughly a century before him.
We are also bad at learning when things go wrong. Instead of determining why a decision was wrong and how to avoid it ever happening again, we often try to ignore the embarrassing turn of events. That means that the next time a similar situation comes around, we do not see the similarity—and repeat the mistake.
Both reveal problems with information. In the first case, we fail to remember personal or historical information. In the second, we fail to encode information when it is available.
That said, we also make mistakes when we cannot efficiently deduce what is going to happen. Perhaps the situation is too complex or too time-consuming to think about. Or we are biased to misinterpret what is going on.
The Annoying Power of TechnologyBut surely technology can help us? We can now store information outside of our brains and use computers to retrieve it. That ought to make learning and remembering easy, right?
Storing information is useful when it can be retrieved well. But remembering is not the same thing as retrieving a file from a known location or date. Remembering involves spotting similarities and bringing things to mind.
An artificial intelligence also needs to be able to spontaneously bring similarities to our mind—often unwelcome similarities. But if it is good at noticing possible similarities (after all, it could search all of the internet and all our personal data), it will also often notice false ones.
For failed dates, it may note that they all involved dinner. But it was never the dining that was the problem. And it was a sheer coincidence that there were tulips on the table—no reason to avoid them.
That means it will warn us about things we do not care about, possibly in an annoying way. Tuning its sensitivity down means increasing the risk of not getting a warning when it is needed.
This is a fundamental problem and applies just as much to any advisor: the cautious advisor will cry wolf too often, the optimistic advisor will miss risks.
A good advisor is somebody we trust. They have about the same level of caution as we do, and we know they know what we want. This is difficult to find in a human advisor, and even more so in an AI.
Where does technology stop mistakes? Idiot-proofing works. Cutting machines require you to hold down buttons, keeping your hands away from the blades. A “dead man’s switch” stops a machine if the operator becomes incapacitated.
Microwave ovens turn off the radiation when the door is opened. To launch missiles, two people need to turn keys simultaneously across a room. Here, careful design renders mistakes hard to make. But we don’t care enough about less important situations, making the design there far less idiot-proof.
When technology works well, we often trust it too much. Airline pilots have fewer true flying hours today than in the past due to the amazing efficiency of autopilot systems. This is bad news when the autopilot fails, and the pilot has less experience to go on to rectify the situation.
The first of a new breed of oil platform (Sleipnir A) sank because engineers trusted the software calculation of the forces acting on it. The model was wrong, but it presented the results in such a compelling way that they looked reliable.
Much of our technology is amazingly reliable. For example, we do not notice how lost packets of data on the internet are constantly being found behind the scenes, how error-correcting codes remove noise, or how fuses and redundancy make appliances safe.
But when we pile on level after level of complexity, it looks very unreliable. We do notice when the Zoom video lags, the AI program answers wrong, or the computer crashes. Yet ask anybody who used a computer or car 50 years ago how they actually worked, and you will note that they were both less capable and less reliable.
We make technology more complex until it becomes too annoying or unsafe to use. As the parts become better and more reliable, we often choose to add new exciting and useful features rather than sticking with what works. This ultimately makes the technology less reliable than it could be.
Mistakes Will Be MadeThis is also why AI is a double-edged sword for avoiding mistakes. Automation often makes things safer and more efficient when it works, but when it fails it makes the trouble far bigger. Autonomy means that smart software can complement our thinking and offload us, but when it is not thinking like we want it to, it can misbehave.
The more complex it is, the more fantastic the mistakes can be. Anybody who has dealt with highly intelligent scholars know how well they can mess things up with great ingenuity when their common sense fails them—and AI has very little human common sense.
This is also a profound reason to worry about AI guiding decision-making: it makes new kinds of mistakes. We humans know human mistakes, meaning we can watch out for them. But smart machines can make mistakes we could never imagine.
What’s more, AI systems are programmed and trained by humans. And there are lots of examples of such systems becoming biased and even bigoted. They mimic the biases and repeat the mistakes from the human world, even when the people involved explicitly try to avoid them.
In the end, mistakes will keep on happening. There are fundamental reasons why we are wrong about the world, why we do not remember everything we ought to, and why our technology cannot perfectly help us avoid trouble.
But we can work to reduce the consequences of mistakes. The undo button and autosave have saved countless documents on our computers. The Monument in London, tsunami stones in Japan, and other monuments act to remind us about certain risks. Good design practices make our lives safer.
Ultimately, it is possible to learn something from history. Our aim should be to survive and learn from our mistakes, not prevent them from ever happening. Technology can help us with this, but we need to think carefully about what we actually want from it—and design accordingly.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Adolph Northen/wikipedia
Electrification is a key part of the transition to renewable energy, and phasing out combustion engine vehicles will be a significant piece of that transition. Despite subsidies and tax breaks around electric vehicles in the US, the cars haven’t taken over a large percentage of market share yet, perhaps because their sticker price is still higher than that of gas-powered cars.
Zooming out to the rest of the world, though, it seems EV adoption is going pretty well. The International Energy Agency released its annual Global Electric Vehicle Outlook this week—and it says demand for electric cars is booming.
According to IEA data, more than 10 million electric vehicles were sold worldwide in 2022. That’s 10 million out of a total 75 million, or a little over 13 percent. This year, EV sales are expected to grow by another 35 percent to 14 million, putting their market share at around 18 percent.
This is a significant jump from just three years ago in 2020, when EVs had only four percent of market share. However, 2020 marked the start of the pandemic, and with countries around the world implementing lockdowns throughout the year, no one was doing much of anything nor going anywhere. Accordingly, gas prices fell to some of their lowest levels in years, so there wasn’t much incentive to go electric.
Since then, though—as you may well know if you’re one of millions who hasn’t yet switched to an EV—gas prices haven’t stayed down, and they don’t look likely to drop anytime soon. This is one of the main incentives pushing people to plug in. Government subsidies are helping too, as well as improvements in battery technology and range.
China leads the global charge in EV sales (pun intended, sort of). In 2022, 60 percent of global EV sales happened in China, and now more than half the EVs in the world are there. Chinese carmakers seem to have found a niche in small, cheap models like the Wuling Hong Guang Mini EV.
The tiny car—comparable to the size of Mercedes’ Smart Fortwo—was selling for 28,800 yuan in 2020 (that was about $4,200 at the time). Though the car doesn’t have the sleek look of a Tesla and tops out at 62 miles per hour, it meets peoples’ practical needs for getting around big cities and being able to park easily.
Europe is the second-largest EV market in the world, and the US comes in third. The former saw a 15 percent growth in sales last year, and the latter a whopping 55 percent jump. The report notes that although the Chinese, European, and American markets dominate electric car sales and manufacturing smaller markets have seen some growth as well: EV sales more than tripled in India and Indonesia last year and more than doubled in Thailand.
Though moving away from combustion engine cars will be an important part of the energy transition, we must keep in mind that electric vehicles aren’t a panacea, and they come with their own set of challenges and drawbacks. Mining of critical minerals like lithium, cobalt, and manganese has major environmental and geopolitical implications (and the US isn’t in much of a competitive position at the moment, though the current administration has made moves to change that).
Also, getting millions more electric cars on the road means we’re going to need a lot more electricity. At a time when grids across the US are already looking fragile, it doesn’t seem wise to impose dramatic new electricity demand without first shoring up supply. And if that supply is coming from coal or natural gas, you can’t really say you’re helping save the environment by driving an electric car; the vehicles are only as green as their power source.
There’s also still progress to be made on battery range, charging speed, and availability of charging stations before more drivers will feel comfortable making the switch—not to mention that up-front sticker price.
However, the IEA is counting the rise in EV sales as a win, and we can too. “The internal combustion engine has gone unrivaled for over a century, but electric vehicles are changing the status quo,” said IEA Executive Director Fatih Birol. “By 2030, they will avoid the need for at least five million barrels a day of oil. Cars are just the first wave: electric buses and trucks will follow soon.”
Image Credit: LeeRosario / 716 images
Since OpenAI’s release of ChatGPT last November, the buzz around generative AI has been steadily ramping up. Some are excited about its potential to transform the way we work, create, and live, while others are wary of the dangers it poses and the nefarious ways it can be used. We know that programs like Midjourney, DALL-E, and GPT-4 are enabling millions of people to generate images and text, but not many studies have dug into the impact these tools are having, be it positive or negative.
One such study was released this month. Titled “Generative AI at Work,” the paper, by teams from Stanford and Massachusetts Institute of Technology, is one of the first times researchers take a microscope to the way generative AI is actually affecting peoples’ jobs. The team looked at how employees of a Fortune 500 company were impacted by generative AI when they started using it as part of their day-to-day work.
Tell Me What to SayThe study followed 5,179 customer service agents at a large software firm (whose name wasn’t disclosed) over the course of a year. The employees, mostly based in the Philippines, were split into two groups; one was given access to an AI whose help they could choose to integrate into their work, while the other continued as usual.
The AI was trained on data from over 5,000 successful customer service interactions, likely in the form of recordings of high-performing employees having conversations with customers and resolving their issues. The AI then monitored customer interactions in real time and gave agents suggestions of what to say. The employees could choose to use the suggestions word for word, dismiss them altogether, or use a tweaked version.
The researchers looked at how long it took for agents to solve customers’ issues and how successfully they did so. The results? Good things all around.
For one, the AI enabled customer service agents to get through calls more quickly, resolve more customer complaints successfully, and even handle multiple customer calls at once. The agents using the AI resolved 13.8 percent more issues per hour than they’d been able to without the AI.
And that’s not all. Since the AI’s suggestions skewed towards helping agents be patient and empathetic with frustrated customers, the customers treated the agents better, losing their tempers and raising their voices less (it’s not pretty, but let’s be honest, we’ve all been there). As a result, the agents were happier and more satisfied with their work.
Closing the Skills Gap?Perhaps not surprisingly, the AI was the most helpful for the least-skilled workers and those who had been with the company for the shortest time. Meanwhile, the highest-skilled and most experienced agents didn’t benefit much from using the AI. This makes sense, since the tool was trained on conversations from these workers; they already know what they’re doing.
“High-skilled workers may have less to gain from AI assistance precisely because AI recommendations capture the knowledge embodied in their own behaviors,” said study author Erik Brynjolfsson, director of the Stanford Digital Economy Lab.
The AI enabled employees with only two months of experience to perform as well as those who’d been in their roles for six months. That’s some serious skill acceleration. But is it “cheating”? Are the employees using the AI skipping over valuable first-hand training, missing out on learning by doing? Would their skills grind to a halt if the AI were taken away, since they’ve been repeating its suggestions rather than thinking through responses on their own?
It’s possible that an over-reliance on the tool could be detrimental to employees’ ability to build up and retain skills. But ideally they are learning by doing, just in a faster way, since they’re skipping over the drudgery of many unpleasant interactions with angry customers.
Where does this leave high-skilled employees, though? If their work is being used to train AIs that then freely give their skills to inexperienced employees, that could create issues around fairness and compensation. If you’ve been honing your soothing one-liners for years then a newbie comes in saying all the same things by month two on the job, you’re not going to be thrilled—especially if you’re not getting paid a lot more than the newbie.
Generating More Than WordsFinally, since the AI was essentially training newer employees, their managers didn’t need to spend as much time training them—and more of their time was thus freed up. That means managers could take on bigger teams, which means the company could ultimately hire more employees (if it’s selling enough of its products) and do more business. It seems this particular “generative AI” generated a lot more than just conversation suggestions: it generated employee satisfaction, skill acquisition, and free time.
Will the same hold true for other scenarios where these tools are implemented? Could be, but they should be introduced with caution and oversight nonetheless, as there are likely many secondary effects generative AI could have on a workplace that wouldn’t become apparent right away, and may not be wholly positive.
“We need far more research here,” said Brynjolfsson. “The impact of AI on productivity may vary over time, and adding these tools to the office could require complementary organizational investments, skills development, and business process redesign. And AI systems may impact worker and customer satisfaction, attrition, and patterns of behavior. There’s so much we don’t know.”
Image Credit: Adrian / Pixabay
iSpace, a private space company based in Japan, lost contact with its Hakuto-R spacecraft as it attempted to become the first private mission to land on the moon this morning. “We have to assume that we could not complete the landing on the lunar surface,” iSpace CEO and founder Takeshi Hakamada said during a livestream. “Our engineers will continue to investigate the situation, and we will update you with further information when we finish the investigation.”
Hakuto-R launched on a SpaceX Falcon 9 rocket last December. It took a long but efficient route, looping way out past the moon before using several orbital adjustments and the gravity of the Earth, moon, and sun to enter lunar orbit last month. On April 13, after a few more final adjustments, it locked into a circular orbit 100 kilometers above the lunar surface.
Early in its landing attempt, the spacecraft dipped behind the moon making communications impossible. The team reestablished contact as it rounded the lunar horizon and began its descent. During the livestream, iSpace showed a simulation of the landing. The ride to the surface began with a deceleration burn and a series of attitude adjustments, bending the spacecraft’s trajectory toward the surface and flipping its orientation.
But just as it neared its landing site at Atlas Crater, the team lost all communications. Because they’d previously been in contact, Hakamada said, the presumed cause was a hard landing on the surface. Hakuto-R was the second attempt to land on the moon by a private company. SpaceIL’s Beresheet lander crashed in 2019 when its main engine failed.
The faces of the iSpace team said it all. It was an extremely disappointing outcome. However, Hakamada said, because they’d had contact with the spacecraft until its final moments, they were able to gather valuable data that would be applied to future missions. iSpace achieved eight of its ten mission milestones, demonstrating an ability to navigate to the moon and enter a stable orbit. As the recent inaugural launches of SpaceX’s Starship and Relativity Space’s Terran 1 rocket show, the development of new space systems involves risk and, often, failure.
“We will keep going. Never quit in our quest,” Hakamada said.
The company already has a second mission in the works for 2024. If that lander succeeds where Hakuto-R failed, the team plans to ramp up the frequency of trips to the moon. The business will transport and operate scientific and government payloads, and longer term, they hope to develop and sell lunar resources. Late last year, Japan issued a license to iSpace to sell lunar dust to NASA as a test case for such future transactions.
“If iSpace transfers ownership of lunar resources to NASA in accordance with its plan, it will be the first case in the world of commercial transactions of space resources on the moon by a private operator,” Sanae Takaichi, Japan’s Minister of State for Space Policy, said at a press conference last year. “This will be a groundbreaking first step toward the establishment of commercial space exploration by private operators.”
iSpace is just one of a new wave of space companies working in low-Earth orbit and beyond. SpaceX’s reusable Falcon 9 rockets have already reduced the cost of getting to space, and the company hopes to make another leap with its Starship rocket. SpaceX and others are building infrastructure in orbit, including Earth observation and telecommunications networks. Meanwhile, NASA is partnering with private companies to develop commercial space stations that may succeed the ISS and funding SpaceX’s Starship in the hopes it can land Artemis astronauts on the moon.
If all goes to plan, iSpace’s attempted moon landing today won’t be its last. And with luck, what they learned in the process will increase their chances of success next time around.
Image Credit: iSpace (Earthrise as captured by the iSpace lunar lander 100 kilometers above the moon’s surface)
Building proteins with AI is like furnishing a house.
There are two main strategies. One is the IKEA approach: you buy pre-made pieces that easily snap together, but can only hope the furniture somewhat fits your space. While relatively simple, you have no control over the dimensions or functions of the final product.
The other way starts with a vision and design perfectly tailored to your needs. But the hard part is finding—or building—individual pieces for the custom design.
The same two methods apply to engineering protein complexes using AI. Similar to a cabinet, protein complexes are made of multiple sub-units that intricately bind together. These mega structures—with shapes ranging from a twenty-sided die to tunnels that open and close—form the foundation of our metabolism, immune defenses, and brain functions.
Previous attempts at shaping protein architectures mostly used the IKEA approach. It’s revolutionary: AI-based designs have already generated COVID vaccines at lightning speed. While powerful, the approach is limited by available protein “building blocks.”
This month, a team led by Dr. David Baker from the University of Washington took protein design to a new custom level. Starting with specific dimensions, shapes, and other properties, the team tapped into a machine learning algorithm to build protein complexes tailored to specific biological responses.
In other words, rather than the usual bottom-up method, they went top-down.
One design, for example, is a 20-sided shell that mimics the outer protective layer of viruses. When dotted with immune-stimulating proteins from the flu virus, the AI-designed protein shell sparked an immune response in mice that outperformed the latest vaccine candidates in clinical trials.
The AI isn’t just for vaccines. The same strategy could build more compact and efficient carriers for gene therapies or carry antibodies and other drugs that need extra protection from being immediately broken down in the body.
But more broadly, the study shows that it’s possible to design massively complex protein architectures starting from an overall vision, rather than working with the biological equivalent of two-by-four boards.
“It’s astounding that the team could do this,” said Dr. Martin Noble at Newcastle University, who was not involved in the work. “It takes evolution billions of years to design single proteins that fold just right, but this is another level of complexity, to fold proteins to fit so well together and make closed structures.”
Evolution at Warp SpeedAt the heart of the new work is reinforcement learning. You’ve probably heard of it. Loosely based on how the brain learns through trial and error, reinforcement learning powers multiple AI agents that have taken the world by storm. Perhaps the best known is AlphaGo, the DeepMind brainchild that triumphed over the human world champion in the board game Go. More recently, reinforcement learning has been speeding progress in self-driving cars and even developing better algorithms by streamlining fundamental computations.
In the new study, the team tapped into a type of reinforcement learning algorithm called the Monte Carlo tree search (MCTS). While sounding like a casino move, it’s a popular reinforcement learning strategy that searches for optimized decisions.
Picture the algorithm as a tree of your life decisions. We’ve likely all wondered how our lives would be if we made a different choice at some point. If you draw out those alternative decisions as a timeline—voilà, you have a decision tree, with each combination of branches leading to a different outcome.
MCTS, then, is a bit like the game of life. Choices are selected at each branch randomly and followed down that path of the tree. Once it reaches the final outcome, it feeds back up the tree to increase the probability of your desired solution. It’s like exploring the multiverse in Everything, Everywhere, All At Once—but instead of life choices, here it’s for designing proteins.
To start, the team fed the MCTS algorithm millions of protein fragments with specific building goals. The fragment amounts were carefully weighed: a smaller number at each calculation step speeds up the AI’s learning process and increases the diversity of the final protein. But more pieces also cause computation time and energy use to skyrocket. Balancing the dilemma, the team built several protein structural elements as a starting point to begin the protein design search.
Like fumbling with digital Play-Doh, the algorithm then twisted or bent protein fragments to see if they passed the overall geometric constraints of the final protein—including its backbone and its “attachment points” to help the fragments self-assemble. If the simulations got the thumbs up, their computational pathways were “boosted” in the algorithm. Rinse and repeat tens of thousands of times, and the program can hone in on optimal individual parts for a certain design.
While it sounds like a massive undertaking, the algorithm was highly efficient. Each iteration on average took only tens of milliseconds, the team explained.
Proteins on DemandIn the end, the team had a powerful algorithm that—like an architect—designed proteins based on custom needs. In one test, the AI made a range of protein structures from prisms to pyramids and letters of the alphabet, with each filling a specific space as required.
“Our approach is unique because we use reinforcement learning to solve the problem of creating protein shapes that fit together like pieces of a puzzle. This simply was not possible using prior approaches and has the potential to transform the types of molecules we can build,” said study author Isaac Lutz.
But how do the AI designs translate to real life?
As a proof of concept, the team made hundreds of proteins in the lab to test for fidelity. Using an electron microscope, the AI-designed proteins were almost identical to the predicted blueprints at the atomic scale.
One design standout was a hollow shell made with dozens of protein pieces. Called a capsid, the structure resembles the protective protein layer for viruses—one often used as a guide to generate vaccines. Unlike previous iterations, the AI-generated shells were densely packed with multiple attachment points. Like wall anchors, these can help the structures dock onto cells or better package material—drugs, gene therapies, or other biological materials—inside the scaffold.
At roughly 10 nanometers, these nano-capsids are “considerably smaller than most viral” ones, the team explained.
The petite sizing came with a big medicinal punch. In one test, the team dotted the capsids with 60 copies of a protein that helps stimulate blood vessel growth in human cells from the umbilical veins. The AI-made protein bubble outperformed a previous nanoparticle more than 10-fold. This “opens up potential applications…for diabetes, brain injuries, strokes, and other cases where blood vessels are at risk,” said study author Dr. Hannele Ruohola-Baker.
Another experiment took full advantage of the dense attachment points on the 20-sided shell, transforming the capsid into an efficient vaccine. Here, the team fused a flu protein HA (influenza hemagglutinin) to the nano-capsid and injected it into mice. Compared to a similar but much larger vaccine design already in clinical trials, the AI-designed solution sparked a heftier immune response.
For now, the AI is still in its early stages. But as the past two years have shown, it’ll rapidly evolve. The 20-sided shell and other structures “are distinct from any previously designed or naturally-occurring structures,” said the team. Thanks to their small size but large carrying capacity, they can potentially tunnel inside the cell nucleus—which houses DNA—and efficiently shuttle gene editing components.
“Its potential to make all kinds of architectures has yet to be fully explored,” said study author Dr. Shunzhi Wang.
Image Credit: Ian Haydon/ UW Medicine Institute for Protein Design
Do you remember learning to drive a car? You probably fumbled around for the controls, checked every mirror multiple times, made sure your foot was on the brake pedal, then ever-so-slowly rolled your car forward.
Fast forward to now and you’re probably driving places and thinking, “How did I even get here? I don’t remember the drive.” The task of driving, which used to take a lot of mental energy and concentration, has now become subconscious, automatic, and habitual.
But how—and why—do you go from concentrating on a task to making it automatic?
Habits Are There to Help Us CopeWe live in a vibrant, complex, and transient world where we constantly face a barrage of information competing for our attention. For example, our eyes take in over one megabyte of data every second. That’s equivalent to reading 500 pages of information or an entire encyclopedia every minute.
Just one whiff of a familiar smell can trigger a memory from childhood in less than a millisecond, and our skin contains up to four million receptors that provide us with important information about temperature, pressure, texture, and pain.
And if that wasn’t enough data to process, we make thousands of decisions every single day. Many of them are unconscious and/or minor, such as putting seasoning on your food, picking a pair of shoes to wear, choosing which street to walk down, and so on.
Some people are neurodiverse, and the ways we sense and process the world differ. But generally speaking, because we simply cannot process all the incoming data, our brains create habits—automations of the behaviors and actions we often repeat.
Two Brain SystemsThere are two forces that govern our behavior: intention and habit. In simple terms, our brain has dual processing systems, sort of like a computer with two processors.
Performing a behavior for the first time requires intention, attention, and planning—even if plans are made only moments before the action is performed.
This happens in our prefrontal cortex. More than any other part of the brain, the prefrontal cortex is responsible for making deliberate and logical decisions. It’s the key to reasoning, problem-solving, comprehension, impulse control, and perseverance. It affects behavior via goal-driven decisions.
For example, you use your “reflective” system (intention) to make yourself go to bed on time because sleep is important or to move your body because you’ll feel great afterwards. When you are learning a new skill or acquiring new knowledge, you will draw heavily on the reflective brain system to form new memory connections in the brain. This system requires mental energy and effort.
From Impulse to HabitOn the other hand, your “impulsive” (habit) system is in your brain’s basal ganglia, which plays a key role in the development of emotions, memories, and pattern recognition. It’s impetuous, spontaneous, and pleasure seeking.
For example, your impulsive system might influence you to pick up greasy takeaway on the way home from a hard day at work, even though there’s a home-cooked meal waiting for you. Or it might prompt you to spontaneously buy a new, expensive television. This system requires no energy or cognitive effort as it operates reflexively, subconsciously, and automatically.
When we repeat a behavior in a consistent context, our brain recognizes the patterns and moves the control of that behavior from intention to habit. A habit occurs when your impulse towards doing something is automatically initiated because you encounter a setting in which you’ve done the same thing in the past. For example, getting your favorite takeaway because you walk past the food joint on the way home from work every night— and it’s delicious every time, giving you a pleasurable reward.
Shortcuts of the MindBecause habits sit in the impulsive part of our brain, they don’t require much cognitive input or mental energy to be performed.
In other words, habits are the mind’s shortcuts, allowing us to successfully engage in our daily life while reserving our reasoning and executive functioning capacities for other thoughts and actions.
Your brain remembers how to drive a car because it’s something you’ve done many times before. Forming habits is, therefore, a natural process that contributes to energy preservation.
That way, your brain doesn’t have to consciously think about your every move and is free to consider other things—like what to make for dinner, or where to go on your next holiday.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: igor kisselev / Shutterstock.com
ARTIFICIAL INTELLIGENCEOpenAI’s CEO Says the Age of Giant AI Models Is Already Over
Will Knight | Wired“Altman’s statement suggests that GPT-4 could be the last major advance to emerge from OpenAI’s strategy of making the models bigger and feeding them more data. He did not say what kind of research strategies or techniques might take its place. In the paper describing GPT-4, OpenAI says its estimates suggest diminishing returns on scaling up model size. Altman said there are also physical limits to how many data centers the company can build and how quickly it can build them.”
ARTIFICIAL INTELLIGENCEGoogle’s Big AI Push Will Combine Brain and DeepMind Into One Team
Emma Roth and Jay Peters | The Verge“Called Google DeepMind, the new group is led by DeepMind CEO Demis Hassabis, as former AI lead Jeff Dean steps into the role of chief scientist. …In a post shared by Alphabet and Google CEO Sundar Pichai, he says the combined groups will ‘significantly accelerate our progress in AI.’i”
LAW AND ETHICSAI Drake Just Set an Impossible Legal Trap for Google
Nilay Patel | The Verge“YouTube only continues to exist because of a delicate dance that keeps rightsholders happy and the music industry paid, but the future of Google itself is a bet on an expansive interpretation of copyright law that every creative industry from music to movies to news hates and will fight to the death.“
ROBOTICSGoogle Robot Learns to Sort the Recyclables Left in Office Waste Bins
Alex Wilkins | New Scientist“Levine and his team let the robots sort people’s rubbish for nearly 10,000 hours over two years and, after that period, the machines managed, on average, to sort 84 per cent of items accurately in a classroom test. The robots had also learned their own intuitive behaviors for some waste, such as nudging larger objects over the bin’s edge or picking up multiple small items at the same time, says Levine.”
COMPUTINGMicrosoft Reportedly Working on Its Own AI Chips That May Rival Nvidia’s
Tom Warren | The Verge“Microsoft’s own AI chips aren’t said to be direct replacements for Nvidia’s, but the in-house efforts could cut costs significantly as Microsoft continues its push to roll out AI-powered features in Office apps and elsewhere. …If Microsoft is working on its own AI chips, it would be the latest in a line of tech giants. Amazon, Google, and Meta also have their own in-house chips for AI, but many companies are still relying on Nvidia chips to power the latest large language models.”
LONGEVITYYour Hair Is Going Gray. This Glitch May Explain Why.
Kate Golembiewski | The New York Times“A new study in mice, but with implications for people and published Wednesday in the journal Nature, provides a clearer picture of the cellular glitches that turn us into silver foxes and vixens. ‘This is a really big step toward understanding why we gray,’ said Mayumi Ito, an author of the study and a dermatology professor at New York University’s Grossman School of Medicine.”
SPACEAirbus Shows Off Space Station Design With Simulated Gravity
Passant Rabie | Gizmodo“The company designed its orbital module to be stationed around Earth, the Moon, or even Mars, whether it be attached to commercial or government-owned infrastructure in space. ‘The Airbus LOOP is designed to fit with the upcoming generation of super-heavy launchers that can launch an entire module in one piece,’ Airbus wrote. ‘Thus, the Airbus LOOP is immediately operational once in orbit, ready to host humans and payloads.'”
ENERGYHundreds of These 24-Ton Bricks Could Change How We Use Renewable Energy
Stephen Shankland | CNET“Gravity batteries are a potentially important solution to a critical problem with the green energy revolution: making sure electricity is available when we need it, not just during the times when sun and wind supply it. And it isn’t just an idea. With two sites under construction—one in Rudong, China, just north of Shanghai, and the other in Snyder, Texas, about 250 miles west of Dallas—startup Energy Vault will begin seriously testing the viability of the gravity storage technology.”
TRANSPORTATIONThis Startup Can Give EVs a Full Charge in 10 Minutes
Adele Peters | Fast Company“On a sunny Thursday morning in San Francisco, a driver pulled a Nissan Leaf into a building, parked on a platform, and stepped out of the car. Then the platform lifted the car in the air. Robotic arms carefully pulled out the electric car’s batteries and placed fresh, fully charged batteries inside. The whole process took around 10 minutes. Ample, the startup that designed the station and runs a dozen more like it around the Bay Area, now does hundreds of battery swaps each day.”
Image Credit: Clark Van Der Beken / Unsplash
SpaceX’s massive new Starship rocket is central to Elon Musk’s plans for humanity to become multi-planetary. It blasted off for its maiden flight yesterday but exploded shortly afterwards, marking a stuttering first step for the billionaire’s hopes of expanding further into the solar system.
The 394-foot rocket successfully lifted off from the company’s Starbase spaceport on the coast of South Texas at 8:28am local time, but almost four minutes into the flight it experienced what SpaceX referred to as a “rapid unscheduled disassembly.”
The problem seems to have been that the Starship upper stage failed to separate from the Super Heavy Booster, which caused the vehicle to tumble head over heels until it exploded. The company had planned for the vehicle to fly to an altitude of 146 miles before looping once around Earth and splashing down near Hawaii, but in the end it topped out at 20 miles.
Despite the setback, though, the mood at the company and the wider space industry was broadly positive. Failures are common in early flights of new rockets, so simply getting the vehicle off the ground is likely to be seen as a decent outcome.
“With a test like this, success comes from what we learn, and today’s test will help us improve Starship’s reliability as SpaceX seeks to make life multi-planetary,” SpaceX said in a tweet after the launch.
It probably shouldn’t have come as a surprise that the first proper launch of the Starship would end in a fireball. Several smaller prototypes had already exploded during testing, and with 33 raptor engines attempting to lift a combined 5,000 tons of fuel and spacecraft, this was a much bigger challenge. The company had also had to cancel a launch attempt on Monday due to a frozen valve.
“I don’t think anybody, Elon Musk included, expected this launch to go perfectly,” space analyst Laura Forczyk told New Scientist. “I would have been shocked if it had been 100 percent successful this first time.”
But plenty is riding on the company learning quickly from this failure. NASA has already awarded the company contracts for two upcoming missions in the agency’s Artemis lunar exploration program. The company has agreed to deliver modified versions of the Starship to help transport astronauts from NASA’s Orion space capsule to the lunar surface in 2025 and 2028.
With the agency having successfully flown its own SLS heavy launch vehicle late last year, the pressure is now on SpaceX to catch up. Its customer seems fairly relaxed for the time being, though. Following the launch, NASA administrator Bill Nelson tweeted: “Every great achievement throughout history has demanded some level of calculated risk, because with great risk comes great reward. Looking forward to all that SpaceX learns, to the next flight test—and beyond.”
NASA isn’t the only one invested in getting the Starship up and running. The reusable rocket is capable of lifting up to 150 tons of equipment into orbit, which can be extended to 250 tons if it’s only making a one-way trip. That could dramatically reshape the economics of getting into orbit.
“We have always been constrained in space flight by mass, volume, and cost,” Jennifer Heldmann, a planetary scientist at NASA’s Ames Research Center, told Nature. “And all of those constraints are basically lifted with Starship.”
A relatively cheap way of getting large amounts of material into space could help make some of SpaceX’s more ambitious goals achievable, such as building a permanent moonbase or helping create human settlements on Mars. Scientists are also excited about the prospect of being able to launch far more ambitious planetary probes or gigantic new telescopes that could greatly expand our understanding of the cosmos.
Before any of that can happen, though, SpaceX’s engineers have a lot of work to do to incorporate the lessons from this partial success. Given the company’s track record of spaceflight innovation, it may not be long before the Starship is blasting off again.
Image Credit: SpaceX
Headlines about climate change have filled newsfeeds over the last few years, ranging from catastrophic (natural disasters, endangered species, dire predictions for the future) to a bit more optimistic (electrification, the transition to renewable energy, climate tech advances). The content we see and read plays a key role in shaping our opinions about climate change, but it remains a contentious topic.
A survey carried out by the Energy Policy Institute at the University of Chicago (EPIC) and The Associated Press–NORC Center for Public Affairs Research aimed to find out how Americans really feel about climate change. The results were released over the last couple weeks in anticipation of Earth Day on April 22. In addition to general questions about climate change, the survey asked people about their views on energy policy and electric vehicles.
5,408 adults completed the survey between January 31 and February 15 of this year. There were respondents from all 50 US states, and they varied in age, race, gender, and education level.
In a nutshell, here’s what the survey found: Americans believe climate change is happening, but they’re not terribly worried about it, and are mostly not willing to spend money or go out of their way to help fix it.
Believers, Sort Of74 percent of the survey respondents said they believe climate change is real. However, less than half—49 percent—believe it’s being caused by human activities (as opposed to natural changes in the environment). That 49 percent is down from 60 percent the last time this survey was carried out, in 2018. The change in viewpoint was uniform across education levels, from college graduates to those who stopped studying after high school. However, more people in the 18 to 29 age group changed their view than did those aged 60 or older.
Image Credit: Energy Policy Institute at the University of Chicago (EPIC)In terms of actually taking action, more than half of respondents said they’re already trying to reduce their energy consumption (though this is likely as much of an effort to keep energy bills down as it is to help the environment). Some ways people are doing so is by using energy-efficient appliances (68 percent), turning off unnecessary lights (89 percent), using less paper and plastic (58 percent), eating less meat (37 percent), and using less heat and air conditioning (60 percent). These are relatively easy, low-cost actions that most anyone can take.
Fewer people are opting in to pricier climate-friendly actions, like putting solar panels on their home (11 percent), buying an electric or hybrid vehicle (12 percent), or getting electricity through a supplier that uses renewable sources (25 percent).
Hard to Cough Up the CashIt seems that much of Americans’ willingness to help combat climate change comes down to economics. Almost two-thirds of those surveyed said they weren’t willing to pay any amount of money to combat climate change—not even a $1 carbon fee a month. 38 percent would pay $1 a month, and 21 percent would pay $100 a month.
Image Credit: Energy Policy Institute at the University of Chicago (EPIC)How much people are willing to pay is likely more a function of their disposable income than of their concern over the environment. However, peoples’ willingness to shell out any amount of money, whether $1 or $100, decreased about 10 percent between 2021 and the present. This is likely because of the financial squeeze put on so many people by the pandemic and rising inflation; when you’re worried about making rent or buying groceries, helping the planet isn’t going to be high on your list.
“It’s striking that Americans’ willingness to pay even a $1 monthly fee to combat climate change fell to below half of respondents—the lowest level since we began tracking this data,” said Michael Greenstone, director of EPIC and an economics professor at the University of Chicago. “Americans’ willingness to pay for climate policy is far below what research projects climate change will cost society per ton of CO2 emissions.”
Similarly, 41 percent of people said they would buy an electric vehicle—if the long-term savings on gas and maintenance added up to more than the higher up-front cost of the car (cost was the biggest barrier to buying an EV). Those most likely to buy one are under 45 years old, live on the west coast in urban areas, and have high incomes. Unsurprisingly, people don’t want to be pushed into buying electric cars; just 35 percent support stricter fuel efficiency standards to encourage EV sales, and 27 percent are in favor of requiring new car sales to be electric or hybrid by 2035.
Help From Uncle SamBased on these responses, it seems we’re likely to find ourselves in a bit of a pickle in coming years. Despite believing in climate change, most Americans aren’t up for throwing much money at it. This must be partly due to the tough economic times we’re in; inflation and interest rates have soared, and whispers of an impending recession have been circulating for months.
But it’s also a sign that even once the economy improves and people feel more secure in their finances, real progress likely won’t be made without significant government intervention—that is, subsidies, regulation, and incentives. These need to be carefully balanced with practical concerns and realism, which can be a tall order.
Image Credit: Wikimedia Commons
The heyday of generative AI is upon us, with millions of images and billions of words per day being produced by models like DALL-E, Stable Diffusion, and GPT-3. By 2025, experts predict up to 90 percent of new online content could be AI-generated. But what does this mean for human creativity? How will our view of art and originality change if the most common way to create something starts with prompting an algorithm?
A German artist felt these questions deserved more attention than they’re getting, so he brought them to light—in quite a memorable way. Boris Eldagsen submitted an entry to the Sony world photography competition, and when he won he revealed that his “photo” wasn’t really a photo, but rather an image generated by an AI. Eldagsen turned down the prize money, suggesting it be donated to a photo festival in Ukraine.
His image, titled “Pseudomnesia: The Electrician,” depicts two women, one hanging on to the other from behind. It won the contest’s creative open category.
Image Credit: Boris Eldagsen“We, the photo world, need an open discussion…about what we want to consider photography and what not,” he said. “Is the umbrella of photography large enough to invite AI images to enter—or would this be a mistake? With my refusal of the award I hope to speed up this debate.”
Eldagsen’s statement is both timely and prescient. Though the debate around AI’s role in art has been going on for years, as the technology advances—and more importantly, becomes accessible to multitudes of people—it’s a conversation that will only get more relevant.
On the cynical side of the spectrum, some would argue that generative AI could dampen, ruin, or overtake human creativity; on the other side of the coin, AI could help creativity flourish, letting anyone who wants to become a creative generalist.
Though Eldagsen misled the photo contest organizers to some degree by submitting an AI-generated photo in the first place, he did confirm that his entry was “co-created” with AI before it was selected as the winner.
“AI images and photography should not compete with each other in an award like this,” he said. “They are different entities. AI is not photography. Therefore I will not accept the award.”
This isn’t the first time an AI art piece has won an award and caused controversy. Last year an image created with Midjourney—an AI program that converts lines of text into realistic graphics—took first place in the digital category at the Colorado State Fair. Though the artist was fully transparent about how his piece had been created, his win provoked criticism from fellow artists who accused him of cheating.
Unfortunately, the number of people who submit AI-generated content to competitions or exhibits and don’t disclose that they had help could far outnumber those who fess up—and herein lies one of the biggest problems we’re grappling with as the technology continues to advance. Being truly creative is hard, and AI is making it easy, and that’s not necessarily a great thing across the board.
ChatGPT, for example, has been used by students to write college essays or do homework, by fraudsters for phishing purposes, and by bad actors to spread disinformation and commit cybercrime. Even if AI-generated content isn’t used for nefarious purposes, maybe it gives creators a bit too easy of a leg up; there’s something to be said for days or months or years of learning a craft or laboring over a work of art.
If a human is involved in generating an image or text with AI, even just by inputting a prompt, how much credit should they get for their work—and who owns it? Is it fair to judge art that was made with AI alongside works conceived of purely by a human brain? Does extensive editing of an AI-generated piece count as mostly human work, or mostly algorithmic work?
These questions don’t yet have widely agreed-upon answers, but they’re going to have to in the not-too-distant future. Use of generative AI tools is only going to grow, so we’ll have no choice but to figure out how their products fit into the larger landscape of art and creativity.
The Sony photography contest’s organizers seem aware that Eldagsen’s stunt can’t be simply be dismissed as silly or irrelevant. “We recognize the importance of this subject and its impact on image-making today,” they stated. “While elements of AI practices are relevant in artistic contexts of image-making, the awards always have been and will continue to be a platform for championing the excellence and skill of photographers and artists working in the medium.”
Image Credit: Boris Eldagsen
Our bodies’ molecular machinery breaks down with age.
DNA accumulates mutations. The protective ends of chromosomes erode away. Mitochondria, the cell’s energy factory, falter and break down. The immune system goes haywire. The reserve pool of stem cells dwindles, while some mature cells enter a zombie-like state, spewing toxic chemicals into their environment.
The picture sounds dire, but it’s not all bad news. Aging is a complicated puzzle. By finding individual pieces, scientists can assemble a full picture of how and why we age—and engineer new ways to stave off age-related symptoms.
There’s already been some success. Senolytics—drugs that kill off zombie cells—are already in clinical trials. Partial reprogramming, which erases a cell’s identity and reverts it back to a stem-cell-like state, is gaining steam as a promising alternative treatment, and it’s one of the hottest longevity investments in Silicon Valley.
A new study in Nature hunted down another piece to the aging puzzle. In five species across the evolutionary scale—worms, flies, mice, rats, and humans—the team honed in on a critical molecular process that powers every single cell inside the body and degrades with age.
The process, called transcription, is the first step in turning our genetic material into proteins. Here, DNA letters are reworked into a “messenger” called RNA, which then shuttles the information to other parts of the cell to make proteins.
Scientists have long suspected that transcription may go awry with aging, but the new study offers proof that it doesn’t—with a twist. In all five of the species tested, as the organism grew older the process surprisingly sped up. But like trying to type faster when blindfolded, error rates also shot up.
There’s a fix. Using two interventions known to extend lifespan, the team was able to slow down transcription in multiple species, including mice. Genetic mutations that reversed the sloppy transcription also extended lifespan in worms and fruit flies, and boosted human cells’ ability to divide and grow.
The new hallmark of aging is hardly ready for human testing. But “it opens up a really fundamental new area of understanding how and why we age,” said Dr. Lindsay Wu at UNSW Sydney, who was not involved in the study.
The Genetic EditorTurning our genetic blueprint into proteins is a two-step process.
First, DNA’s four letters—A, T, C, and G—are transcribed into RNA. Also made up of four letters, RNA strands are basically molecular notes that can slip past DNA’s confined space to deliver messages to the cell’s protein-making factory. There, RNA is translated into the language of proteins.
The first step—turning DNA into RNA—is harder than it sounds. To conserve space, DNA is tightly wrapped around a group of proteins called histones, like bacon around eight stalks of asparagus. This effectively “hides” the genetic information, making it impossible for the cell to read.
It takes a whole village of protein helpers to unwind DNA and prepare it for transcription. But the star is Pol II (RNA polymerase II), a giant multicomplex that moves along a DNA strand helping it transform into an early version of RNA, aptly called pre-RNA.
Like a wordy sentence, pre-RNA strands are then copyedited into pithier sequences for building proteins, a process called splicing. Pol II oversees the entire process, making sure that hundreds of thousands of RNAs are perfectly made.
Yet as we age, the process degrades. No one has figured out why.
The new study asked: why not hone in on the star of the transcription show?
Spanning SpeciesDeciphering aging hallmarks comes with a stumbling block: a potential lead may only be relevant for one species.
The new study tackled the problem head-on by examining five species. Using a technique called RNA sequencing, they captured Pol II’s speed as it rolled down the DNA of worm, fruit fly, mouse, rat, and human cells at different ages. Human samples ranged from 21 to 70 years of age, along with two “immortal” cultured cell lines.
For an even more comprehensive view the team tested samples from multiple organs, including the brain, liver, kidneys, and blood.
The results came back as a surprise. Although every species had their own Pol II “speed signature,” the trend was the same: Poll II sped up across species with age in every tissue examined. The exact gene or tissue didn’t matter. The age-related change covered roughly 200 different genes in multiple species. Rather than a local change, the Pol II speed-up seemed to be a universal aging marker.
With speed, however, came errors. Splicing—which edits pre-RNAs—requires Pol II speed to be in a Goldilocks zone. Increasing the speed boosts the risk of bad translations, which in previous studies “has been associated with advanced age and shortened lifespan,” the authors explained.
“Increased speeds of Pol II can lead to more transcriptional errors because the proofreading capacity of Pol II is challenged,” they said.
Turning Back the ClockIf Pol II in overdrive contributes to aging, can we slow it down—and in turn combat aging?
In one test, the team tapped into two well-known treatments for delaying aging: inhibiting insulin signaling and caloric restriction. In worms, flies, and mice, genetically disrupting the insulin-sensing pathway slowed down the pace of Pol II. Putting mice on a diet in early adulthood and middle age—but not old age—also tapped the brakes on Pol II.
Another test honed in on the ultimate question: does Pol II acceleration drive aging? Here, the team tracked a horde of genetically engineered worms and fruit flies harboring mutations that reduce their Pol II speed. Compared to non-mutants, both engineered strains extended their lifespans by 10 to 20 percent.
When the team used CRISPR-Cas9 to reverse the Pol II mutations in worms, however, their lifespan shortened and matched the wild-type peers. It seems like Pol II is a cause for aging, explained the authors.
Why?Digging deeper into the transcription machinery, the team found one answer. Remember: DNA is wrapped in bacon-asparagus bundles, known scientifically as nucleosomes. By comparing human umbilical vein cells and lung cells, the team found that as cells age, the bundles slowly unwind and fall apart. This makes it far easier for Pol II to slide across a DNA strand, in turn triggering a transcription speed boost.
Further testing their theory, the team genetically inserted two types of histone proteins—the asparagus part of the nucleosome bundle—to form more nucleosomes in human cells in Petri dishes. This in turn created additional speed bumps for Pol II and slowed it down.
It worked. Cells with additional histone proteins had less chance of becoming zombie senescent cells. In fruit flies, a popular model for longevity research, the genetic tweak gave them a notable lifespan bump.
Although it’s still very early, the results are great news for potentially pursuing a novel class of anti-aging drugs. Pol II has been extensively researched in cancer therapy, with multiple medications already tested and approved, providing the chance of repurposing the medications for longevity research.
“Together, the data presented here reveal a molecular mechanism contributing to aging and serve as a means for assessing the fidelity of the cellular machinery during aging and disease,” said the team.
Image Credit: David Bushnell, Ken Westover and Roger Kornberg, Stanford University/NIH Image Gallery
Finding life on other planets might well be the holy grail of astronomy, but the hunt for suitable host planets that can sustain life is a resource-intensive task.
The search for exoplanets (planets outside our solar system) involves competing for time on Earth’s biggest telescopes—yet the hit rate of this search can be disappointingly low.
In a new study recently published in Science, my colleagues and I combined different search techniques to discover a new giant planet. It could change the way we try to image planets in the future.
Imaging Planets Is No Small FeatTo satisfy our curiosity about our place in the universe, astronomers have developed many techniques to search for planets orbiting other stars. Perhaps the simplest of these is called direct imaging. But it’s not easy.
Direct imaging involves attaching a powerful camera to a large telescope and trying to detect light emitted, or reflected, from a planet. Stars are bright, and planets are dim, so it’s akin to searching for fireflies dancing around a spotlight.
It’s no surprise only about 20 planets have been found with this technique to date.
Yet direct imaging is of great value. It helps shed light on a planet’s atmospheric properties, such as its temperature and composition, in a way other detection techniques can’t.
HIP99770b: A New Gas GiantOur direct imaging of a new planet, named HIP99770b, reveals a hot, giant and moderately cloudy planet. It orbits its star at a distance that falls somewhere between the orbital distances of Saturn and Uranus around our sun.
The HIP99770 star is almost 14 times brighter than the sun. But since its planet has an orbit larger than Saturn’s, the planet receives a similar amount of energy as Jupiter does from the sun. Author providedWith about 15 times the mass of Jupiter, HIP99770b is a real giant. However, it’s also more than 1,000℃, so it’s not a good prospect for a habitable world.
What the HIP99770 system does offer is an analogy to our own solar system. It has a cold “debris disk” of ice and rock far out from the star, akin to a scaled-up version of the Kuiper Belt in our solar system.
The main difference is that the HIP99770 system is dominated by one high-mass planet, rather than several smaller ones.
Images of the HIP99770 system, taken with exoplanet imager SCExAO (Subaru Coronagraphic Extreme Adaptive Optics Project) coupled with data from the CHARIS instrument (Coronagraphic High-Resolution Imager and Spectrograph). Author providedSearching With the Light OnWe reached our findings by first detecting hints of a planet via indirect detection methods. We noticed the star was wobbling in space, which hinted at the presence of a planet in the vicinity with a large gravitational pull.
This motivated our direct imaging efforts; we were no longer searching in the dark.
The extra data came from the European Space Agency’s Gaia spacecraft, which has been measuring the positions of nearly one billion stars since 2014. Gaia is sensitive enough to detect tiny variations of a star’s motion through space, such as those caused by planets.
We also supplemented these data with measurements from Gaia’s predecessor, Hipparcos. In total, we had 25 years’ worth of “astrometric” (positional) data to work with.
Previously, researchers have used indirect methods to guide imaging that has discovered companion stars, but not planets.
It’s not their fault: massive stars such as HIP99770—which is almost twice the mass of our sun—are reluctant to give up their secrets. Otherwise-successful search techniques can rarely reach the levels of precision required to detect planets around such massive stars.
Our detection, which used both direct imaging and astrometry, demonstrates a more efficient way to search for planets. It’s the first time the direct detection of an exoplanet has been guided through initial indirect detection methods.
Gaia is expected to continue observing until at least 2025, and its archive will remain useful for decades to come.
Mysteries remainAstrometry of HIP99770 suggests it belongs to the Argus association of stars—a group of stars that moves together through space. This would suggest the system is rather young, about 40 million years old. That would make it roughly one-hundredth of the age of our solar system.
However, our analysis of the star’s pulsations, as well as models of the planet’s brightness, suggest an older age of between 120 million and 200 million years. If this is the case, HIP99770 might just be an interloper in the Argus group.
Now that it’s known to host a planet, astronomers will aim to further unravel the mysteries of HIP99770 and its immediate environment.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Subaru Telescope image of HIP99770. T. Currie/Subaru Telescope, UTSA
“This is the true story of 25 video game characters picked to live in a town and have their lives taped…to find out what happens when computers stop being polite…and start getting real.”
Researchers at Google and Stanford recently created a new reality show of sorts—with AI agents instead of people.
Using OpenAI’s viral chatbot ChatGPT and some custom code, they generated 25 AI characters with back stories, personalities, memories, and motivations. Then the researchers dropped these characters into a 16-bit video game town—and let them get on with their lives. So, what does happen when computers start getting real?
“Generative agents wake up, cook breakfast, and head to work,” the researchers wrote in a preprint paper posted to the arXiv outlining the project. “Artists paint, while authors write; they form opinions, and notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day.”
Not exactly riveting television, but surprisingly lifelike for what boils down to an enormous machine learning algorithm…talking to itself.
The AI town, Smallville, is just the latest development in a fascinating moment for AI. While the basic version of ChatGPT takes interactions one at a time—write a prompt, get a reply—a number of offshoot projects are combining ChatGPT with other programs to automatically complete a cascade of tasks. These might include making a to-do list and checking off items on the list one by one, Googling information and summarizing the results, writing and debugging code, even critiquing and correcting ChatGPT’s own output.
It’s these kinds of cascading interactions that make Smallville work too. The researchers have crafted a series of companion algorithms that, together, power simple AI agents that can store memories and then reflect, plan, and act based on those memories.
The first step is to create a character. To do this, the researchers write a foundational memory in the form of a detailed prompt describing that character’s personality, motivations, and situation. Here’s an abbreviated example from the paper: “John Lin is a pharmacy shopkeeper at the Willow Market and Pharmacy who loves to help people. He is always looking for ways to make the process of getting medication easier for his customers; John Lin is living with his wife, Mei Lin, who is a college professor, and son, Eddy Lin, who is a student studying music theory.”
But characterization isn’t enough. Each character also needs a memory. So, the team created a database called the “memory stream” that logs an agent’s experiences in everyday language.
When accessing the memory stream, an agent surfaces the most recent, important, and relevant memories. Events of the highest “importance” are recorded as separate memories the researchers call “reflections.” Finally, the agent creates plans using a nest of increasingly detailed prompts that break the day into smaller and smaller increments of time—each high level plan is thus broken down into smaller steps. These plans are also added to the memory stream for retrieval.
As the agent goes about its day—translating text prompts into actions and conversations with other characters in the game—it taps its memory stream of experiences, reflections, and plans to inform each action and conversation. Meanwhile, new experiences feed back into the stream. The process is fairly simple, but when combined with OpenAI’s large language models by way of the ChatGPT interface, the output is surprisingly complex, even emergent.
In a test, the team prompted a character, Isabella, to plan a Valentine’s Day party and another, Maria, to have a crush on a third, Klaus. Isabella went on to invite friends and customers to the party, decorate the cafe, and recruit Maria, her friend, to help. Maria mentions the party to Klaus and invites him to go with her. Five agents attend the party—but equally human—several flake or simply fail to show up.
Beyond the initial seeds—the party plan and the crush—the rest emerged of its own accord. “The social behaviors of spreading the word, decorating, asking each other out, arriving at the party, and interacting with each other at the party, were initiated by the agent architecture,” the authors wrote.
It’s remarkable this can be accomplished, for the most part, by simply splitting ChatGPT into a number of functional parts and personalities and playing them off one another.
Video games are the most obvious application of this kind of believable, open-ended interaction, especially when combined with high-fidelity avatars. Non-player characters could evolve from scripted interactions to conversations with convincing personalities.
The researchers warn people may be tempted to form relationships with realistic characters—a trend that’s already here—and designers should take care to add content guardrails and always disclaim when a character is an agent. Other risks include those applicable to generative AI at large, such as the spread of misinformation and over-reliance on agents.
This approach may not be practical enough to work in mainstream video games just yet, but it does suggest such a future is likely coming soon.
The same is true of the larger trend in agents. Current implementations are still limited, despite the hype. But connecting multiple algorithms—complete with plugins and internet access—may allow for the creation of capable, assistant-like agents that can carry out multistep tasks at a prompt. Longer term, such automated AI could be quite useful, but also pose the risk of misaligned algorithms causing unanticipated problems at scale.
For now, what’s most obvious is how the dance between generative AI and a community of developers and researchers continues to surface surprising new directions and capabilities—a feedback loop that’s showing no signs of slowing just yet.
Image Credit: “Generative Agents: Interactive Simulacra of Human Behavior,” Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein
ARTIFICIAL INTELLIGENCEA New Approach to Computation Reimagines Artificial Intelligence
Anil Ananthaswamy | Quanta“By imbuing enormous vectors with semantic meaning, we can get machines to reason more abstractly—and efficiently—than before. …This is the starting point for a radically different approach to computation known as hyperdimensional computing. The key is that each piece of information, such as the notion of a car, or its make, model or color, or all of it together, is represented as a single entity: a hyperdimensional vector.”
BIOTECHBacteria Can Be Engineered to Fight Cancer in Mice. Human Trials Are Coming.
Jessica Hamzelou | MIT Technology Review“There are trillions of microbes living in and on our bodies—and we might be able to modify them to help us treat diseases. Scientists have altered the genomes of some of these bacteria that live on skin, essentially engineering microbes that can prevent or treat cancer. It appears to work in mice, and human trials are in the cards.”
SPACERelativity Space Is Moving on From the Terran 1 Rocket to Something Much Bigger
Eric Berger | Ars Technica“Foremost among these changes is the plan to move directly into development of the Terran R rocket. In response to customer demand, Ellis said, this rocket is getting even bigger than before. A fully expendable version will now be able to lift a staggering 33.5 metric tons. This sets up Relativity to compete directly with the largest players in the global launch industry. ‘It’s a big, bold bet,’ Relativity Space Chief Executive Tim Ellis said in an interview. ‘But it’s actually a really obvious decision.’i”
ENERGYNo, Fusion Energy Won’t Be ‘Limitless’
Gregory Barber | Wired“…as the physics progresses, some are now beginning to explore the likely practical and economic limits on fusion. The early conclusion is that fusion energy ain’t going to be cheap—certainly not the cheapest source of electricity over the coming decades as more solar and wind come online. But fusion may still find its place, because the grid needs energy in different forms and at different times.”
SCIENCEThat Famous Black Hole Just Got Bigger and Darker
Dennis Overbye | The New York Times“i‘We used machine learning to fill in the gaps,’ Dr. Medeiros said in an interview. Her team trained the neural network to recognize the black hole by feeding the AI simulations of all kinds of black holes consistent with Einstein’s equations. In the improved version, Dr. Medeiros said, the doughnut of doom—the visible radiation from matter falling into the hole—is thinner than in the original. And the empty spot in the doughnut’s center appears blacker and bigger, bolstering the idea that there really is a black hole there.”
ARTIFICIAL INTELLIGENCEOpenAI’s CEO Confirms the Company Isn’t Training GPT-5 and ‘Won’t for Some Time’
James Vincent | The Verge“However, just because OpenAI is not working on GPT-5 doesn’t mean it’s not expanding the capabilities of GPT-4—or, as Altman was keen to stress, considering the safety implications of such work. ‘We are doing other things on top of GPT-4 that I think have all sorts of safety issues that are important to address and were totally left out of the letter,’ he said. Altman’s comments are interesting—though not necessarily because of what they reveal about OpenAI’s future plans. Instead, they highlight a significant challenge in the debate about AI safety: the difficulty of measuring and tracking progress.”
CRYPTOCURRENCYEthereum’s Shanghai Update Opens a Rift in Crypto
Joel Khalili | Wired“At 19:27 Eastern time on April 12, the Ethereum blockchain, home to the world’s second-most-popular cryptocurrency, ether, will finally sever its links to crypto mining. …By demonstrating that a large-scale blockchain can shift from one system to another, Shanghai will reignite a debate over whether the practice of mining that still supports bitcoin, the most widely traded cryptocurrency, is viable and sustainable.”
SECURITYThe Hacking of ChatGPT Is Just Getting Started
Matt Burgess | Wired“The attacks are essentially a form of hacking—albeit unconventionally—using carefully crafted and refined sentences, rather than code, to exploit system weaknesses. While the attack types are largely being used to get around content filters, security researchers warn that the rush to roll out generative AI systems opens up the possibility of data being stolen and cybercriminals causing havoc across the web.”
Image Credit: Ambrose Chua / Unsplash
Finding ways to integrate electronics into living tissue could be crucial for everything from brain implants to new medical technologies. A new approach has shown that it’s possible to 3D print circuits into living worms.
There has been growing interest in finding ways to more closely integrate technology with the human body, in particular when it comes to interfacing electronics with the nervous system. This will be crucial for future brain-machine interfaces and could also be used to treat a host of neurological conditions.
But for the most part, it’s proven difficult to make these kinds of connections in ways that are non-invasive, long-lasting, and effective. The rigid nature of standard electronics means they don’t mix well with the squishy world of biology, and getting them inside the body in the first place can require risky surgical procedures.
A new approach relies instead on laser-based 3D printing to grow flexible, conductive wires inside the body. In a recent paper in Advanced Materials Technologies, researchers showed they could use the approach to produce star- and square-shaped structures inside the bodies of microscopic worms.
“Hypothetically, it will be possible to print quite deep inside the tissue,” John Hardy at Lancaster University, who led the study, told New Scientist. “So, in principle, with a human or other larger organism, you could print around 10 centimeters in.”
The researchers’ approach involves a high-resolution Nanoscribe 3D printer, which fires out an infrared laser that can cure a variety of light-sensitive materials with very high precision. They also created a bespoke ink that includes the conducting polymer polypyrrole, which previous research had shown could be used to electrically stimulate cells in living animals.
To prove the scheme could achieve the primary goal of interfacing with living cells, the researchers first printed circuits into a polymer scaffold and then placed the scaffold on top of a slice of mouse brain tissue being kept alive in a petri dish. They then passed a current through the flexible electronic circuit and showed that it produced the expected response in the mouse brain cells.
The team then decided to demonstrate the approach could be used to print conductive circuits inside a living creature, something that had so far not been achieved. The researchers decided to use the roundworm C. elegans due to its sensitivity to heat, injury, and drying out, which they said would make for a stringent test of how safe the approach is.
First, the team had to adjust their ink to make sure it wasn’t toxic to the animals. They then had to get it inside the worms by mixing it with the bacterial paste they’re fed on.
Once the animals had ingested the ink, they were placed under the Nanoscribe printer, which was used to create square and star shapes a few micrometers across on the worms’ skin and within their guts. The shapes didn’t come out properly in the moving gut though, the researchers admit, due to the fact it was constantly moving.
The shapes printed inside the worms’ bodies had no functionality. But Ivan Minev from the University of Sheffield told New Scientist the approach could one day make it possible to build electronics intertwined with living tissue, though it would still take considerable work before it was applicable in humans.
The authors also admit that adapting the approach for biomedical applications would require significant further research. But in the long run, they believe their work could enable tailor-made brain-machine interfaces for medical purposes, future neuromodulation implants, and virtual reality systems. It could also make it possible to easily repair bioelectronic implants within the body.
All that’s likely still a long way from being realized, but the approach shows the potential of combining 3D printing with flexible, biocompatible electronics to help interface the worlds of biology and technology.
Image Credit: Kbradnam/Wikimedia Commons
Over the short span of just 300 years, since the invention of modern physics, we have gained a deeper understanding of how our universe works on both small and large scales. Yet, physics is still very young and when it comes to using it to explain life, physicists struggle.
Even today, we can’t really explain what the difference is between a living lump of matter and a dead one. But my colleagues and I are creating a new physics of life that might soon provide answers.
More than 150 years ago, Darwin poignantly noted the dichotomy between what we understand in physics and what we observe in life—noting at the end of The Origin of Species “…whilst this planet has gone cycling on according to the fixed law of gravity, from so simple a beginning endless forms most beautiful and most wonderful have been and are being evolved.”
The Importance of TimeIsaac Newton described a universe where the laws never change, and time is an immutable and absolute backdrop against which everything moves. Darwin, however, observed a universe where endless forms are generated, each changing features of what came before, suggesting that time should not only have a direction, but that it in some ways folds back on itself. New evolutionary forms can only arise via selection on the past.
Presumably these two areas of science are describing the same universe, but how can two such diametrically opposite views be unified? The key to understanding why life is not explainable in current physics may be to reconsider our notions of time as the key difference between the universe as described by Newton and that of Darwin. Time has, in fact, been reinvented many times through the history of physics.
Although Newton’s time was fixed and absolute, Einstein’s time became a dimension—just like space. And just as all points in space exist all at once, so do all points in time. This philosophy of time is sometimes referred to as the “block universe” where the past, present, and future are equally real and exist in a static structure—with no special “now.” In quantum mechanics, the passage of time emerges from how quantum states change from one to the next.
The invention of thermodynamics gave time its arrow, explaining why it’s moving forward rather than backwards. That’s because there are clear examples of systems in our universe, such as a working engine, that are irreversible—only working in one direction. Each new area of fundamental physics, whether describing space and time (Newton/Einstein), matter and light (quantum mechanics), or heat and work (thermodynamics) has introduced a new concept of time.
But what about evolution and life? To build novel things, evolution requires time. Endless novelty can only come to be in a universe where time exists and has a clear direction. Evolution is the only physical process in our universe that can generate the succession of novel objects we associate to life—things like microbes, mammals, trees, and even cellphones.
Information and MemorySuch objects cannot fluctuate into existence spontaneously. They require a memory, based on what existed in the past, to construct things in the present. It is such “selection” that determines the dividing line between the universe described by current physics and what Darwin saw: it is the mechanism that turns a universe where memory does not matter in determining what exists, to one where it does.
Life is information. Image Credit: ANIRUDH / UnsplashThink about it, everything in the living world requires some kind of memory and information flow. The DNA in our cells is our blueprint. And to invent new things, such as rockets or medication, living beings also need information—knowledge of the laws of physics and chemistry.
To explain life, we therefore need to understand how the complex objects life creates exist in time. With my collaborators, we have been doing just that in a newly proposed theory of physics called assembly theory.
A key conjecture of assembly theory is that, as objects become more complex, the number of unique parts that make it up increases, and so does the need for local memory to store how to assemble the object from its unique parts. We quantify this in assembly theory as the shortest number of physical steps to build an object from its elementary building blocks, called the assembly index.
Importantly, assembly theory treats this shortest path as an intrinsic property of the object, and indeed we have shown how an assembly index can be measured for molecules using several different measuring techniques including mass spectrometry (an analytical method to measure the mass-to-charge ratio of molecules).
With this approach, we have shown in the lab, with measurements on both biological and non-biological samples, how molecules with an assembly index above 15 steps are only found in living samples.
This suggests that assembly theory is indeed capable of testing our hypothesis that life is the only physics that generates complex objects. And we can do so by identifying those objects that are so complex the only physical mechanism to form them is evolution.
We are aiming to use our theory to estimate when the origin if life happens by measuring the point at which molecules in a chemical soup become so complex that they start using information to make copies of themselves—the threshold at which life arises from non-life. We may then apply the theory to experiments aiming to generate a new origin of life event in the lab.
And when we know this, we can use the theory to look for life on worlds that are radically different to Earth, and may therefore look so alien that we wouldn’t recognize life there.
If the theory holds, it will force a radical rethink on time in physics. According to our theory, assembly can be measured as an intrinsic property for molecules, which corresponds to their size in time—meaning time is a physical attribute.
Ultimately, time is intrinsic to our experiences of the world, and it is necessary for evolution to happen. If we want physics to be capable of explaining life—and us—it may be that we need to treat time as a material property for the first time in physics.
This is perhaps the most radical departure for physics of life from standard physics, but it may be the critical insight needed to explain what life is.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Zdeněk Macháček / Unsplash
From Iceland to Wyoming and Switzerland to Texas, carbon capture plants are under construction or working away to suck CO2 out of the air and store it underground. A pilot project in Kenya will soon join their ranks, except it will be focused more on the storage part than the capture part.
Cella Mineral Storage is aiming to store CO2 by injecting it into water then pumping that water underground, where it will make contact with volcanic rock. A chemical reaction will occur between the minerals in the rock and the CO2, causing the CO2 to turn to stone in a process called mineralization.
Mineralization occurs naturally, but Cella wants to speed it up; it’s an attempt to start righting a balance that’s been thrown out of whack over the last 200 or so years.
Speeding Up the Carbon CycleWe know there’s been more and more CO2 dumped into Earth’s atmosphere over the last few decades, and that these carbon emissions are contributing to climate change. But the part we may not remember is that the total amount of carbon in Earth’s ecosystem doesn’t change much. Carbon isn’t escaping to nor coming from outer space; Earth and its atmosphere are a closed environment. It’s where on Earth the carbon is located that’s constantly changing.
The process of carbon atoms moving from Earth into the atmosphere and eventually back to Earth, where they get stored in rocks, sediments, the ocean, and living things, is called the carbon cycle. Carbon gets released into the atmosphere when plants or animals die, volcanoes erupt, or most notably for our purposes, when humans burn fossil fuels for energy.
Burning coal, oil, and natural gas inserted a fire hose (or rather, many of them) into the carbon cycle that had been mostly humming along smoothly for millennia; we sped up one end of the cycle without any way of counterbalancing it on the other end. That’s the problem direct air capture plants aim to solve, and what Cella wants to do in Kenya.
Chemical ReactionsMost carbon capture projects inject CO2 into underground rock reservoirs, where it eventually dissolves into groundwater and solidifies through chemical reactions with certain minerals. Dissolving CO2 in water before injecting it underground accelerates the process and makes it harder for small amounts of the carbon to escape before it’s mineralized. Carbon storage experiments in Iceland found that 90 percent of the CO2 that was captured this way transformed into minerals in just two years.
Basalt rock, specifically, is rich in calcium and magnesium, which react with CO2 to make calcite, dolomite, and magnesite. Basalt is a volcanic rock formed from cooled lava. It can be found all over the Earth and continues to form today at active rift sites—one of which is in Iceland, and another the East African Rift Valley. The latter runs from Ethiopia to Mozambique, with Kenya right at its center.
New Territory for Carbon StorageKenya is the eighth-largest producer of geothermal energy in the world, and gets almost 40 percent of its power from geothermal sources. The country’s geology means there’s a massive amount of heat relatively close to the Earth’s surface. The movement of tectonic plates creates fissures in the Earth’s crust that bring underground water into contact with ultra-hot rock, creating steam. This phenomenon exists all over the world, but in most places you’d have to drill much deeper into the ground to find temperatures like those in Kenya.
Besides this bottomless source of energy, the country’s geology also means there’s a huge amount of volcanic rock—like basalt—underground (including the Mega Basalt Field along the Kenya-Ethiopia border). Combine abundant geothermal power with easy access to volcanic rock, and you’ve got an ideal site for direct air capture and carbon sequestration.
The scope of Cella’s work in Kenya will initially be limited to carbon sequestration via mineralization; the company will buy CO2 from outside sources to use as a proof of concept. Eventually, the startup will partner with another company that will build a carbon capture plant, bringing the capture-and-store process full circle. Kenya’s biggest geothermal power plant, Olkaria, is set to open its newest unit later this year. Cella will start its pilot around the same time.
In Kenya, proponents of carbon storage—from the country’s president, William Ruto, to the cofounder of Climate Action Platform-Africa, James Mwangi—are endorsing the idea of their country as an ideal location for the technology, speaking of building a “Great Carbon Valley” and exporting carbon credits.
Reality CheckBut how realistic is this? There are major gaps to bridge from an infrastructure and business standpoint before carbon capture could truly become a viable industry in Kenya—not to mention the viability of carbon capture in general. Even if we build thousands of direct air capture plants, we can’t hope to capture more than a fraction of the CO2 that’s already in the atmosphere. Using geothermal energy helps by ensuring the capture process itself doesn’t emit CO2, but it remains debatable whether these plants will make enough of a difference to be worth the cost of building and running them.
Finally, we should pause to consider the ethics of using a large amount of power to store carbon in a country that hasn’t created much carbon, and could use said power to grow its economy. Kenya’s grid has the capacity to produce more electricity from renewable sources than the country uses, yet more than a quarter of its population doesn’t have reliable access to electricity. It’s something of a chicken-before-the-egg problem; there’s not enough industry to build out the grid, but as long as the grid remains limited, there’s no power for industry.
On the one hand, then, you may as well put the surplus of renewable power to good use, like by capturing and storing CO2. But how might these projects impact the likelihood of the country’s economy growing and diversifying, and the cost of and access to energy for the average Kenyan?
Cella believes its activities will have a net positive impact on Kenya and its people, and this could certainly end up being the case. But it will be a while before we find out, since the pilot is just the first step of many.
In our continued quest to repair the damage done to Earth, carbon capture in Kenya and elsewhere will continue to be explored as options; time and economics will tell whether the technology persists.
Image Credit: Roma Neus/Wikimedia Commons
A decade ago, the Human Brain Project launched with a blue-sky goal: digitizing a human brain.
The goal wasn’t to construct an average brain from groups of people. Rather, it was to replicate parts of a person’s unique neural connections in a personalized virtual brain twin.
The implications were huge: simulated brains could provide crucial clues to help crack some of the most troubling neurological diseases. Rather than using animal models, they might better represent an Alzheimer’s brain, or one from people with autism or epilepsy.
The billion-euro project was initially met with much skepticism. Yet as the project wrapped up last month, it achieved a milestone. In a study published this January, the teams showed that virtual brain models of people with epilepsy can help neurosurgeons better hunt down the brain regions responsible for their seizures.
Each virtual brain tapped into a computational model dubbed the Virtual Epileptic Patient (VEP), which uses a person’s brain scans to create their digital twin. With a dose of AI, the team simulated how seizure activity spreads across the brain, making it easier to spot hotspots and better target surgical intervention. The method is now being tested in an ongoing clinical trial called EPINOV. If successful, it’ll be the first personalized brain modeling method used for epilepsy surgery and could pave the road for tackling other neurological disorders.
The results will be part of the legacy of the Virtual Brain (TVB), a computational platform to digitize personalized neural connections. Hunting seizures is just the beginning. To Dr. Viktor Jirsa at the Aix-Marseille University in France, who led the effort, these simulations may transform how we diagnose and treat neurological disorders.
To be clear: the models aren’t exact replicas of a human brain. There’s no evidence they are “thinking” or conscious in any way. Rather, they simulate personalized brain networks—that is, how one brain region “talks” to another—based on images of their wiring.
“As evidence accumulates in support of the predictive power of personalized virtual brain models, and as methods are tested in clinical trials, virtual brains might inform clinical practice in the near future,” Jirsa and colleagues wrote.
Biological to Digital BrainsLarge-scale brain mapping projects now seem trivial. From those that map connections across a mammalian brain to those that distill the brain’s algorithms from neural wiring, brain maps have grown into multiple atlases and 3D models for anyone to explore.
Flashback to 2013. AI for deciphering the brain was just a dream—but one already pursued by a scrappy startup now known as DeepMind. Neuroscientists were hunting down the neural code—the brain’s algorithms—with success, but in independent labs.
What if we combined those efforts?
Enter the Human Brain Project (HBP). With more than 500 scientists across 140 universities and other research institutions, the European Union project became one of the first large-scale programs—along with the US’s BRAIN Initiative and Japan’s Brain/MINDS—to attempt to solve the brain’s mysteries by digitally mapping its intricate connections.
At the HBP’s core is a digital platform dubbed EBRAINS. Think of it as a public square, where neuroscientists gather and openly share their data to collaborate with a broader community. In turn, it’s hoped, the global effort can generate better models of the brain’s inner workings.
Why care? Our thoughts, memories, and emotions are all encoded in the brain’s neural networks. Like how Google Maps for local roads gives insight into traffic patterns, brain maps can spark ideas on how neural networks normally communicate—and when they go awry.
One example: Epilepsy.
The Virtual Epilepsy TwinEpilepsy affects roughly 50 million people worldwide and is triggered by abnormal brain activity. There are medical treatments. Unfortunately, around one-third of patients don’t respond to anti-seizure medications and need surgery.
It’s a tough procedure. Patients are implanted with multiple electrodes to hunt down the source of the seizures (called the epileptogenic zone). A surgeon then snips away those parts of the brain, hoping to silence unwanted neural lightning storms and minimize side effects.
The surgery is a “huge game changer” for people with untreatable epilepsy, said Dr. Aswin Chari at University College London, who was not involved in the study. But the procedure has only a roughly 60 percent success rate, largely because the epileptogenic zone is hard to pinpoint.
“Before surgery can take place, the patient must have a presurgical evaluation to establish whether and how surgical treatment might stop their seizures without causing neurological deficits,” said Jirsa and colleagues.
The current method relies on a myriad of brain scans. MRI (magnetic resonance imaging), for example, can map detailed structures of the brain. EEG (electroencephalography) captures the brain’s electrical patterns with strategically placed electrodes over the scalp.
SEEG (stereoelectroencephalography) is the next seizure hunter. Here, up to 16 electrodes are placed directly into the skull to monitor suspicious areas for up to two weeks. The method, while powerful, is far from perfect. The brain’s electrical activity “hums” at different frequencies. Like a pair of basic headphones, SEEG captures high-frequency brain activity but misses the “bass”—low-frequency aberrations sometimes seen in seizures.
In the new study, the team integrated all these test results into the Virtual Epileptic Patient model built on the Virtual Brain platform. It starts with images of each patient’s brain from MRI and CT scans—the latter track down the white matter highways connecting brain regions. The data, when combined with SEEG recordings, are rolled up into personalized maps with “nodes”—parts of the brain that are highly connected with each other.
These personalized maps become part of the presurgical screening routine, with no extra effort or stress on the patient.
Using machine-learning-based simulations, the team can build a “digital twin” that roughly mimics a person’s brain structure, activity, and dynamics. In a retrospective test of 53 people with epilepsy, they used these virtual brains to hunt down the brain region responsible for each person’s seizures by triggering seizure-like activity in the digital brains. Testing multiple virtual surgeries, the team found regions to remove for the best outcome.
In one example, the team generated a virtual brain for a patient who had 19 parts of his brain removed to rid him of his seizures. Using simulated surgery, the virtual results matched the outcome of the actual ones.
Overall, the simulations encompass the whole brain. They’re personalized atlases of 162 brain regions with a resolution of around one square millimeter—roughly the size of a small grain of sand. The team is already working to increase the resolution by a thousand times.
A Personalized FutureThe ongoing epilepsy trial EPINOV has recruited over 350 people. Scientists will follow up on their outcomes for a year to see if a digital surrogate brain helps keep them free of seizures.
Despite a decade of work, it’s still early days for using virtual brain models to treat disorders. For one, neural connections change over time. A model of an epilepsy patient is just a snapshot in time and may not capture their health status following treatment or other life events.
But the Virtual Brain is a powerful tool. Beyond epilepsy, it’s set to help scientists explore other neurological disorders, such as Parkinson’s disease or multiple sclerosis. In the end, said Jirsa, it’s all about collaboration.
“Computational neuromedicine needs to integrate high-resolution brain data and patient specificity,” he said. “Our approach heavily relies on the research technologies in EBRAINS and could only have been possible in a large-scale, collaborative project such as the Human Brain Project.”
Image Credit: KOMMERS / Unsplash
Picking out separate objects in a visual scene seems intuitive to us, but machines struggle with this task. Now a new AI model from Meta has developed a broad idea of what an object is, allowing it to separate out objects even if it’s never seen them before.
It might seem like a fairly prosaic computer vision task, but being able to parse an image and work out where one object ends and another begins is a pretty fundamental skill, without which a host of more complicated tasks would be unsolvable.
“Object segmentation” is nothing new; AI researchers have worked on it for years. But typically, building these models has been a time-consuming process requiring lots of human annotation of images and considerable computing resources. And typically the resulting models were highly specialized to particular use cases.
Now though, researchers at Meta have unveiled the Segment Anything Model (SAM), which is able to cut out any object in any scene, regardless of whether it’s seen anything like it before. The model can also do this in response to a variety of different prompts, from text description to mouse clicks or even eye-tracking data.
“SAM has learned a general notion of what objects are, and it can generate masks for any object in any image or any video,” the researchers wrote in a blog post. “We believe the possibilities are broad, and we are excited by the many potential use cases we haven’t even imagined yet.”
Key to the development of the model was a massive new dataset of 1.1 billion segmentation masks, which refers to regions of an image that have been isolated and annotated to denote that they contain a particular object. It was created through a combination of manual human annotation of images and automated processes, and is by far the largest collection of this type assembled to date.
By training on such a massive dataset, Meta’s researchers say it has developed a general concept of what an object is, which allows it to segment things it hasn’t even seen before. This ability to generalize led the researchers to dub SAM a “foundation model,” a controversial term used to describe other massive pre-trained models such as OpenAI’s GPT series, whose capabilities are supposedly so general they can be used as the foundations for a host of applications.
Image segmentation is definitely a key ingredient in a wide range of computer vision tasks. If you can’t separate out the different components of a scene, it’s hard to do anything more complicated with it. In their blog, the researchers say it could prove invaluable in video and image editing, or help with the analysis of scientific imagery.
Perhaps more pertinently for the company’s metaverse ambitions, they provide a demo of how it could be used in conjunction with a virtual reality headset to select specific objects based on the user’s gaze. They also say it could potentially be paired with a large language model to create a multi-modal system able to understand both the visual and textual content of a web page.
The ability to deal with a wide range of prompts makes the system particularly flexible. In a web page demoing the new model, the company shows that after analyzing an image it can be prompted to separate out specific objects by simply clicking on them with a mouse cursor, typing in what it is you want to segment, or just breaking up the entire image into separate objects.
And most importantly, the company is open-sourcing both the model and the dataset for research purposes so that others can build on their work. This is the same approach the company took with its LLaMA large-language model, which led to it rapidly being leaked online and spurring a wave of experimentation by hobbyists and hackers.
Whether the same will happen with SAM remains to be seen, but either way it’s a gift to the AI research community that could accelerate progress on a host of important computer vision problems.
Image Credit: Meta AI
It’s easy to envisage other universes, governed by slightly different laws of physics, in which no intelligent life, nor indeed any kind of organized complex systems, could arise. Should we therefore be surprised that a universe exists in which we were able to emerge?
That’s a question physicists including me have tried to answer for decades. But it is proving difficult. Although we can confidently trace cosmic history back to one second after the Big Bang, what happened before is harder to gauge. Our accelerators simply can’t produce enough energy to replicate the extreme conditions that prevailed in the first nanosecond.
But we expect that it’s in that first tiny fraction of a second that the key features of our universe were imprinted.
The conditions of the universe can be described through its “fundamental constants”—fixed quantities in nature, such as the gravitational constant (called G) or the speed of light (called C). There are about 30 of these representing the sizes and strengths of parameters such as particle masses, forces, or the universe’s expansion. But our theories don’t explain what values these constants should have. Instead, we have to measure them and plug their values into our equations to accurately describe nature.
The values of the constants are in the range that allows complex systems such as stars, planets, carbon, and ultimately humans to evolve. Physicists have discovered that if we tweaked some of these parameters by just a few percent, it would render our universe lifeless. The fact that life exists, therefore, takes some explaining.
Some argue it is just a lucky coincidence. An alternative explanation, however, is that we live in a multiverse, containing domains with different physical laws and values of fundamental constants. Most might be wholly unsuitable for life. But a few should, statistically speaking, be life-friendly.
Impending Revolution?What is the extent of physical reality? We’re confident that it’s more extensive than the domain that astronomers can ever observe, even in principle. That domain is definitely finite. That’s essentially because, like on the ocean, there’s a horizon that we can’t see beyond. And just as we don’t think the ocean stops just beyond our horizon, we expect galaxies beyond the limit of our observable universe. In our accelerating universe, our remote descendants will also never be able to observe them.
Most physicists would agree there are galaxies that we can’t ever see, and that these outnumber the ones we can observe. If they stretched far enough, then everything we could ever imagine happening may be repeated over and over. Far beyond the horizon, we could all have avatars.
This vast (and mainly unobservable) domain would be the aftermath of “our” Big Bang—and would probably be governed by the same physical laws that prevail in the parts of the universe we can observe. But was our Big Bang the only one?
The theory of inflation, which suggests that the early universe underwent a period when it doubled in size every trillionth of a trillionth of a trillionth of a second has genuine observational support. It accounts for why the universe is so large and smooth, except for fluctuations and ripples that are the “seeds” for galaxy formation.
But physicists including Andrei Linde have shown that, under some specific but plausible assumptions about the uncertain physics at this ancient era, there would be an “eternal” production of Big Bangs—each giving rise to a new universe.
String theory, which is an attempt to unify gravity with the laws of microphysics, conjectures everything in the universe is made up of tiny, vibrating strings. But it makes the assumption that there are more dimensions than the ones we experience. These extra dimensions, it suggests, are compacted so tightly together that we don’t notice them all. And each type of compactification could create a universe with different microphysics—so other Big Bangs, when they cool down, could be governed by different laws.
The “laws of nature” may therefore, in this still grander perspective, be local by-laws governing our own cosmic patch.
We can only see a fraction of the universe. Image Credit: NASA/James Webb Space TelescopeIf physical reality is like this, then there’s a real motivation to explore “counterfactual” universes—places with different gravity, different physics and so forth—to explore what range of parameters would allow complexity to emerge, and which would lead to sterile or “stillborn” cosmos. Excitingly, this is ongoing, with recent research suggesting you could imagine universes that are even more friendly to life than our own. Most “tweakings” of the physical constants, however, would render a universe stillborn.
That said, some don’t like the concept of the multiverse. They worry it would render the hope for a fundamental theory to explain the constants as vain as Kepler’s numerological quest to relate planetary orbits to nested platonic solids.
But our preferences are irrelevant to the way physical reality actually is—so we should surely be open minded to the possibility of an imminent grand cosmological revolution. First we had the Copernican realization that the Earth wasn’t the center of the solar system—it revolves around the sun. Then we realized that there are zillions of planetary systems in our galaxy, and that there are zillions of galaxies in our observable universe.
So could it be that our observable domain—indeed our Big Bang—is a tiny part of a far larger and possibly diverse ensemble?
Physics or Metaphysics?How do we know just how atypical our universe is? To answer that we need to work out the probabilities of each combination of constants. And that’s a can of worms that we can’t yet open—it will have to await huge theoretical advances.
We don’t ultimately know if there are other Big Bangs. But they’re not just metaphysics. We might one day have reasons to believe that they exist.
Specifically, if we had a theory that described physics under the extreme conditions of the ultra-early Big Bang—and if that theory had been corroborated in other ways, for instance by deriving some unexplained parameters in the standard model of particle physics—then if it predicted multiple Big Bangs, we should take it seriously.
Critics sometimes argue that the multiverse is unscientific because we can’t ever observe other universes. But I disagree. We can’t observe the interior of black holes, but we believe what physicist Roger Penrose says about what happens there—his theory has gained credibility by agreeing with many things we can observe.
About 15 years ago, I was on a panel at Stanford where we were asked how seriously we took the multiverse concept—on the scale “would you bet your goldfish, your dog, or your life” on it. I said I was nearly at the dog level. Linde said he’d almost bet his life. Later, on being told this, physicist Steven Weinberg said he’d “happily bet Martin Rees’ dog and Andrei Linde’s life.”
Sadly, I suspect Linde, my dog, and I will all be dead before we have an answer.
Indeed, we can’t even be sure we’d understand the answer—just as quantum theory is too difficult for monkeys. It’s conceivable that machine intelligence could explore the geometrical intricacies of some string theories and spew out, for instance, some generic features of the standard model. We’d then have confidence in the theory and take its other predictions seriously.
But we’d never have the “aha” insight moment that’s the greatest satisfaction for a theorist. Physical reality at its deepest level could be so profound that its elucidation would have to await posthuman species—depressing or exhilarating as that may be, according to taste. But it’s no reason to dismiss the multiverse as unscientific.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Lanju Fotografie / Unsplash
ARTIFICIAL INTELLIGENCEDevelopers Are Connecting Multiple AI Agents to Make More ‘Autonomous’ AI
Chloe Xiang | Motherboard“Multiple developers are trying to create an ‘autonomous’ system by stringing together multiple instances of OpenAI’s large language model (LLM) GPT that can do a number of things on its own, such as execute a series of tasks without intervention, write, debug, and develop its own code, and critique and fix its own mistakes in written outputs.”
ROBOTICSAI Is Running Circles Around Robotics
Jacob Stern | The Atlantic“AI’s recent progress has been fueled to a significant extent by training larger models with greater computation power on larger data sets. Roboticists inclined toward this approach—hoping to apply the same machine-learning techniques that have proved so fruitful for large language models—run into problems.”
NANOTECHTiny Hybrid Robot Can Identify, Capture a Single Cell
Paul McClure | New Atlas“Once the hybrid propulsion system was assembled, researchers were able to demonstrate the micro-robot’s capabilities. They used it to capture a single red blood cell, cancer cells, and a single bacterium, demonstrating that the micro-robot could distinguish between a healthy cell and one that had been damaged by a drug or a dying cell and one that was undergoing a natural ‘suicide’ process (apoptosis). Once captured, the cell can be moved to an external instrument for further analysis.”
FUTURE OF FOODLab-Grown Burgers Have a Secret Ingredient: Plants
Matt Reynolds | Wired“Two companies in the US have the Food and Drug Administration’s nod that their cultivated meat is safe for human consumption, and are awaiting further sign-off from the Department of Agriculture before they can sell their meat in restaurants and stores. But the economics of growing animal cells in bioreactors are still eye-watering. The easiest way to get meat out there that people can afford is to blend expensive bioreactor-brewed animal cells with much cheaper plant-based proteins. The immediate future of cultivated meat is hybrid.”
TECHThis AI Clock Uses ChatGPT to Generate Tiny Poems That Tell the Time
James Vincent | The Verge“ChatGPT has been one of the internet’s favorite toys for months now, but people are still finding novel and fun ways to use the AI chatbot. Case in point is this rhyming E Ink clock created by designer and blogger Matt Webb. It uses ChatGPT to create a short two-line rhyme that also tells the time for every minute of the day. It’s incredible and we want one.”
SECURITYThree Ways AI Chatbots Are a Security Disaster
Melissa Heikkiläarchive page | MIT Technology Review“Tech companies are racing to embed these models into tons of products to help people do everything from book trips to organize their calendars to take notes in meetings. But the way these products work—receiving instructions from users and then scouring the internet for answers—creates a ton of new risks. With AI, they could be used for all sorts of malicious tasks, including leaking people’s private information and helping criminals phish, spam, and scam people. Experts warn we are heading toward a security and privacy ‘disaster.’i”
LAWStable Diffusion Copyright Lawsuits Could Be a Legal Earthquake for AI
Timothy B. Lee | Ars Technica“There are some strong arguments that copyright’s fair use doctrine allows Stability AI to use the images. But there are also strong arguments on the other side. There’s a real possibility that the courts could decide that Stability AI violated copyright law on a massive scale. That would be a legal earthquake for this still-nascent industry.”
ETHICSThe Call to Halt ‘Dangerous’ AI Research Ignores a Simple Truth
Sasha Luccioni | Wired“Tech leaders’ Open Letter proposed a pause on ChatGPT. But researchers already know how to make artificial intelligence safer. …Instead of focusing on ways that AI may fail in the future, we should focus on clearly defining what constitutes an AI success in the present.“
SPACEIcy Moons With Vast Oceans Are the Latest Candidates for Alien Life
Editorial Staff | The Economist“All four of the solar system’s gas giants are either known or suspected to have watery moons of their own. There is even some evidence that the same may be true for Pluto, a dwarf planet that orbits in the frigid darkness beyond the orbit of Uranus. Assuming that gas giants in other star systems also have moons—and there is no reason to assume they do not—that drastically raises the number of places in the galaxy in which life could have arisen.”
Image Credit: Maxim Berg / Unsplash
More than 200 years ago, the English scientist Thomas Young carried out a famous test known as the “double-slit experiment.” He shone a beam of light at a screen with two slits in it, and observed that the light that passed through the apertures formed a pattern of dark and bright bands.
At the time, the experiment was understood to demonstrate that light was a wave. The “interference pattern” is caused by light waves passing through both slits and interfering with each other on the other side, producing bright bands where the peaks of the two waves line up and dark bands where a peak meets a trough and the two cancel out.
In the 20th century, physicists realized the experiment could be adapted to demonstrate that light not only behaves like a wave, but also like a particle (called a photon). In quantum mechanical theory, this particle still has wave properties—so the wave associated with even a single photon passes through both slits, and creates interference.
In a new twist on the classic experiment, we replaced the slits in the screen with “slits” in time—and discovered a new kind of interference pattern. Our results were published this week in Nature Physics.
Slits in TimeOur team, led by Riccardo Sapienza at Imperial College London, fired light through a material that changes its properties in femtoseconds (quadrillionths of a second), only allowing light to pass through at specific times in quick succession.
We still saw interference patterns, but instead of showing up as bands of bright and dark, they showed up as changes in the frequency or color of the beams of light.
To carry out our experiment, we devised a way to switch on and off the reflectivity of a screen incredibly quickly. We had a transparent screen that became a mirror for two brief instants, creating the equivalent of two slits in time.
Color InterferenceSo what do these slits in time do to light? If we think of light as a particle, a photon sent at this screen might be reflected by the first increase of reflectivity or by the second, and reach a detector.
However, the wave nature of the process means the photon is in a sense reflected by both temporal slits. This creates interference, and a varying pattern of color in the light that reaches the detector.
The amount of change in color is related to how fast the mirror changes its reflectivity. These changes must be on timescales comparable with the length of a single cycle of a light-wave, which is measured in femtoseconds.
Electronic devices cannot function quickly enough for this. So we had to use light to switch on and off the reflectivity of our screen.
We took a screen of indium tin oxide, a transparent material used in mobile phone screens, and made it reflective with a brief pulse of laser light.
From Space to TimeOur experiment is a beautiful demonstration of wave physics, and also shows how we can transfer concepts such as interference from the domain of space to the domain of time.
The experiment has also helped us in understanding materials that can minutely control the behavior of light in space and time. This will have applications in signal processing and perhaps even light-powered computers.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Tobias Carlsson on Unsplash
The human brain remains the most mysterious organ in our bodies. From memory and consciousness to mental illness and neurological disorders, there remain volumes of research and study to be done before we understand the intricacies of our own minds. But to some degree, researchers have succeeded in tapping into our thoughts and feelings, whether roughly grasping the content of our dreams, observing the impact of psilocybin on brain networks disrupted by depression, or being able to predict what sorts of faces we’ll find attractive.
A study published earlier this year described a similar feat of decoding brain activity. Ian Daly, a researcher from the University of Sussex in England, used brain scans to predict what piece of music people were listening to with 72 percent accuracy. Daly described his work, which used two different forms of “neural decoders,” in a paper in Nature.
While participants in his study listened to music, Daly recorded their brain activity using both electroencephalography (EEG)—which uses a network of electrodes and wires to pick up the electrical signals of neurons firing in the brain—and functional magnetic resonance imaging (fMRI), which shows changes in blood oxygenation and flow that occur in response to neural activity.
EEG and fMRI have opposite strengths: the former is able to record brain activity over short periods of time, but only from the surface of the brain, since the electrodes sit on the scalp. The latter can capture activity deeper in the brain, but only over longer periods of time. Using both gave Daly the best of both worlds.
He monitored the brain regions that had high activity during music trials versus no-music trials, pinpointing the left and right auditory cortex, the cerebellum, and the hippocampus as the critical regions for listening to music and having an emotional response to it—though he noted that there was a lot of variation between different participants in terms of the activity in each region. This makes sense, as one person may have an emotional response to a given piece of music while another finds the same piece boring.
Using both EEG and fMRI, Daly recorded brain activity from 18 people while they listened to 36 different songs. He fed the brain activity data into a bi-directional long term short term (biLSTM) deep neural network, creating a model that could reconstruct the music heard by participants using their EEG.
A biLSTM is a type of recurrent neural network that’s commonly used for natural language processing applications. It adds an extra layer onto a regular long-short term memory network, and that extra layer reverses its information flow and allows the input sequence to flow backward. The network’s input thus flows both forwards and backwards (hence the “bi-directional” piece), and it’s capable of utilizing information from both sides. This makes it a good tool for modeling the dependencies between words and phrases—or, in this case, between musical notes and sequences.
Daly used the data from the biLSTM network to roughly reconstruct songs based on peoples’ EEG activity, and he was able to figure out which piece of music they’d been listening to with 72 percent accuracy.
He then recorded data from 20 new participants just using EEG, with his initial dataset providing insight into the sources of these signals. Based on that data, his accuracy for pinpointing songs went down to 59 percent.
However, Daly believes his method can be used to help develop brain-computer interfaces (BCIs) to assist people who’ve had a stroke or who suffer from other neurological conditions that can cause paralysis, such as ALS. BCIs that can translate brain activity into words would allow these people to communicate with their loved ones and care providers in a way that may otherwise be impossible. While solutions already exist in the form of brain implants, if technology like Daly’s could accomplish similar outcomes, it would be much less invasive to patients.
“Music is a form of emotional communication and is also a complex acoustic signal that shares many temporal, spectral, and grammatical similarities with human speech,” Daly wrote in the paper. “Thus, a neural decoding model that is able to reconstruct heard music from brain activity can form a reasonable step towards other forms of neural decoding models that have applications for aiding communication.”
Image Credit: Alina Grubnyak on Unsplash
Self-driving cars get all the hype (or, they did before people realized they weren’t going to be ready by 2020…or 2022…or this year), but self-driving trucks are likely to hit the road first. Not only is the bulk of their driving done on highways, which is far simpler than navigating urban roads with all their obstacles; there’s been a shortage of truck drivers for years, and it doesn’t seem to be getting better, so there’s a market need for trucks that can drive themselves.
If one company’s plan plays out, trucks without drivers will be cruising down highways by next year. On Monday driverless hardware and software specialist Aurora Innovation announced that its Aurora Driver—a system of sensors, software, and a computer designed to give any vehicle self-driving capabilities—is “feature complete.” This means all the product’s technical capabilities are in place and it’s entering its final phase of development. The company is planning to launch the Driver commercially next year.
The current version of the product is Beta 6.0, and it’s specifically built for service on Aurora’s Dallas to Houston route, which is one of the most highly trafficked shipping corridors in the country. At 240 miles long, the route is mostly a straight shot on Interstate 45, and made up the initial run of an autonomous freight pilot Aurora did with FedEx (they subsequently added a 600-mile route between El Paso and Fort Worth).
As a company press release explains, the main difference between Beta 6.0 and its predecessor is the system’s improved ability to handle uncommon road scenarios that impact safety, like high winds, collisions, or sudden heavy rain, snow, or fog. Beta 6.0 can detect the severity of these conditions and either slow down or look for a safe place to pull over. If a vehicle does get in an accident, the system is trained to pull over and alert one of the company’s command center specialists. In addition, Aurora worked with Waymo to design a flashing beacon that alerts oncoming traffic when one of its trucks is pulled over on the side of the road.
Aurora was founded by industry veterans Chris Urmson, who previously led Waymo; Sterling Anderson, who oversaw Tesla’s autopilot program; and Drew Bagnell, who worked on Uber’s self-driving program. The company’s 2019 funding round raised more than $530 million, with Amazon being one of the main investors. The company has partnered with major automakers like Volvo, Volkswagen, Toyota, and Hyundai, among others. In 2020, Aurora acquired Uber’s self-driving unit, and in 2021 went public via a $13 billion special-purpose acquisition company (SPAC) deal with Reinvent Technology Partners. Subsequently, though, its stock fell by more than 85 percent.
Urmson’s optimism seems unshakable, though. “We look at trucking, and we see a landscape where we feel like [we’re] the only viable player,” he recently told Fast Company. “It’s an $800 billion business in the US, and we’re a company that is well capitalized, that’s got incredible talent, amazing partnerships, and awesome technology. We’re like, ‘Let’s just go execute.’i”
Now that Aurora Driver’s architecture is complete, the company will shift its focus to closing its Driver Safety Case, outlining its approach to safety and demonstrating that vehicles equipped with its self-driving system are safe to be on public roads. The US Department of Transportation requires this documentation before the company can commercially launch its product.
If all goes to plan, trucks outfitted with Aurora Driver will be cruising the Houston-Dallas corridor a year from now. Ultimately the goal is to lighten the burden on truckers and give the freight industry a needed boost, improving efficiency and economic feasibility across the board.
Image Credit: Aurora Innovation
A tiny molecular syringe with bizarre origins could overhaul one of the thorniest problems in medicine: getting drugs to their target destinations inside the body. The source? Bacteria living in the gut of insects.
The brain child of Dr. Feng Zhang at the Howard Hughes Medical Institute and Broad Institute, the spring-loaded nanomachine looks a bit like a rocket ship. Once docked, an injector shoots down to penetrate the cells and deliver precious payloads.
When further developed, the molecular injectors could shuttle cancer immunotherapies only to tumor cells, sparing healthy ones and limiting side effects. The system can also safely tunnel into the brain—a notoriously difficult organ for drugs to access—potentially shuttling in proteins that could help diagnose strokes, Alzheimer’s, and other neurological disorders.
Published in Nature, the injector was inspired by the bacterial kingdom. Zhang is no stranger to exploring the dark matter of the bioverse. Best known for his seminal work on CRISPR-Cas9 gene editing, which originated as a bacterial defense system against viruses, Zhang has long taken hints from evolution to craft next-generation biotechnological wonders.
This time, however, his team brought another collaborator into the mix: AlphaFold.
Developed by DeepMind, the AI made headlines for its uncanny ability to predict protein structures. Putting the tool to use, the team optimized a core part of the bacterial injector, making it switch from their preferred target—insect cells—to a variety of mice and human ones.
Several proof-of-concept studies in both cultured cells and mice showcased the new syringe’s prowess. One experiment delivered a toxin to cancer cells without harming others. Another injected Cas9—the protein “scissors” in the gene editing tool CRISPR—into cultured human cells and edited the target genes with high efficiency.
This ability to plug and play makes the system a delivery powerhouse. “We show that just by putting a tag onto the protein, we can load different types of proteins into these needles,” said Zhang.
“Having the ability to deliver particular proteins into specific cell types would offer tremendous potential for research in the life sciences, as well as for the treatment of disease,” said Charles Ericson and Dr. Martin Pilhofer at ETH Zürich, who were not involved in the work.
The system, when combined with others, sets the foundation for a powerful mix-and-match toolbox for both research and medicine. Although currently only capable of shuttling proteins, co-opting other natural molecular syringes could expand the system to DNA and other biomolecules.
“It’s still early days for this as a technology,” said Zhang.
Delivery NightmaresImagine drug delivery as DoorDash. You want your order to come only to you, not your neighbors, and with the food intact.
It sounds trivial, but it’s a task that’s hard to achieve with drugs and gene therapy. Medication in the form of pills, patches, or intravenous needles—think saline bags or chemotherapies—enter the bloodstream. The result is that they flood different organs and tissues and often cause side effects.
In stark contrast, another problem is that some drugs can’t burrow into their targets. Cells are fortresses surrounded by a double-layered fatty membrane, with mechanisms that sometimes actively spit out unwanted intruders. When those intruders are gene therapy elements or therapeutic proteins, the cells’ defense system becomes a massive headache.
Scientists have devised ways to bypass these defenses. One is using harmless viruses to smuggle in vaccine materials. Another is lipid nanospheres, which are made of little fatty bubbles. Once merged with the cell, the bubbles “burst” and release the payload. While foundational for genetic engineering, these systems aren’t as precise as we’d like. Going back to the DoorDash analogy, the dasher will give you some of your order—while bringing the rest to your unsuspecting and unwilling neighbors.
Bacterial InspirationIn the new study, Zhang threw away the playbook and went completely outside the box. He and his colleague Joseph Kreitz tapped into a molecular syringe crafted by evolution.
The unexpected resource is a bioluminescent bacteria called Photorhabdus asymbiotica, which lives in the gut of insects. They come heftily armed: each is equipped with tiny molecular syringes—roughly 100 nanometers long—with “feet” that grasp host cells. Once docked, a plunger drives through the cell’s membrane, shooting in a toxin that kills the host—and in turn allows the bacteria to escape and colonize other cells.
The dangerous-sounding mechanism—dubbed a contractile injection system, or CIS—hardly seems fit for a safe delivery system. But one quirk caught the team’s eye: bacteria injectors usually only work with other bacteria, not animal cells. So why not rejigger the Photorhabdus syringe to also inject human cells?
The team first honed in on a part of the injector called tail fibers. These “tentacle-like things” help the nanomachine latch onto cells, explained Zhang. The key is matching the receptors, or docking stations, on the surface of cells. Each cell type has a myriad of docks unique to their biological character—a neuron, for example, has several that are massively different from those of heart cells. Those from different living creatures are even more divergent.
So it’s no surprise that the syringes, designed to work in insect cells, failed in human ones. Knowing that tail fibers are the crux, the team brought in a new collaborator: AlphaFold. Using the AI, the team generated a 3D model of the tricky protein found in a region that guides the injector towards insect cells.
They then genetically modified this region, chopping off the tail fiber’s end and adding different protein chunks to guide the injector towards specific mouse and human cells.
“[AlphaFold] gave us the information we needed to make a new delivery strategy that can be changed to target different cells,” said Kreitz.
Mix and MatchThe team tested their programmable injectors with several experiments.
In one, they loaded the syringe with a protein that, once injected, caused human cells in culture to glow a vibrant green in the dark. A similar syringe was reworked to track down cancer cells dotted with epidermal growth factor receptor (EGFR) on its surface. Loaded with toxins, the treatment killed nearly all the cells with the receptor but spared others. Similarly, the team easily delivered Cas9 into a variety of human cells, which when supplied with a guide RNA edited the genome at predicted points.
Finally, in the ultimate test, the team injected the system into the hippocampus of mice. Infused with a fluorescent protein, the cells glowed a bright green. Importantly, although derived from bacteria, the injectors didn’t trigger an immune response.
The system isn’t perfect. Although efficient in tested tissues, the team is hoping to expand its range to different types of tissues and disease models. Another goal is to hunt for other natural injectors and potentially grow them into a whole family of delivery tools—in a vein similar to CRISPR’s growth. For now, the system only carries proteins. But further engineering could allow specific delivery of DNA, RNA, and other biomolecules, and perhaps even control their dosage.
“It’s still early for this approach, but I think it’s really important to explore [the system’s ability] to be able to treat many different types of diseases that affect human health,” said Zhang.
Image Credit: Joseph Kreitz, Broad Institute of MIT and Harvard, McGovern Institute for Brain Research at MIT
In architecture, new materials rarely emerge.
For centuries, wood, masonry, and concrete formed the basis for most structures on Earth.
In the 1880s, adoption of the steel frame changed architecture forever. Steel allowed architects to design taller buildings with larger windows, giving rise to the skyscrapers that define city skylines today.
Since the industrial revolution, construction materials have been largely confined to a range of mass-produced elements. From steel beams to plywood panels, this standardized kit of parts has informed the design and construction of buildings for over 150 years.
That may soon change with advances in what’s called “large-scale additive manufacturing.” Not since the adoption of the steel frame has there been a development with as much potential to transform the way buildings are conceived and constructed.
Large-scale additive manufacturing, like desktop 3D printing, involves building objects one layer at a time. Whether it’s clay, concrete, or plastic, the print material is extruded in a fluid state and hardens into its final form.
As director of the Institute for Smart Structures at the University of Tennessee, I’ve been fortunate to work on a series of projects that deploy this new technology.
While some roadblocks to the widespread adoption still exist, I can foresee a future in which buildings are built entirely from recycled materials or materials sourced on-site, with forms inspired by the geometries of nature.
Promising PrototypesAmong these is the Trillium Pavilion, an open-air structure printed from recycled ABS polymer, a common plastic used in a wide range of consumer products.
The structure’s thin, double-curved surfaces were inspired by the petals of its namesake flower. The project was designed by students, printed by Loci Robotics and constructed on the University of Tennessee Research Park at Cherokee Farm in Knoxville.
Other recent examples of large-scale additive manufacturing include Tecla, a 450-square-foot (41.8-square-meter) prototype dwelling designed by Mario Cucinella Architects and printed in Massa Lombarda, a small town in Italy.
Tecla was built from locally sourced clay. Image Credit: WASPThe architects printed Tecla out of clay sourced from a local river. The unique combination of this inexpensive material and radial geometry created an energy-efficient form of alternative housing.
Back in the US, the architecture firm Lake Flato partnered with the construction technology firm ICON to print concrete exterior walls for a home dubbed “House Zero” in Austin, Texas.
The 2,000-square-foot (185.8-square-meter) home demonstrates the speed and efficiency of 3D-printed concrete, and the structure displays a pleasing contrast between its curvilinear walls and its exposed timber frame.
The Planning ProcessLarge-scale additive manufacturing involves three knowledge areas: digital design, digital fabrication, and material science.
To begin, architects create computer models of all the components that will be printed. These designers can then use software to test how the components will respond to structural forces and tweak the components accordingly. These tools can also help the designer figure out how to reduce the weight of components and automate certain design processes, such as smoothing complex geometric intersections, prior to printing.
A piece of software known as a slicer then translates the computer model into a set of instructions for the 3D printer.
You might assume 3D printers work at a relatively small scale—think cellphone cases and toothbrush holders.
But advances in 3D printing technology have allowed the hardware to scale up in a serious way. Sometimes the printing is done via what’s called a gantry-based system—a rectangular framework of sliding rails similar to a desktop 3D printer. Increasingly, robotic arms are used due to their ability to print in any orientation.
The printing site can also vary. Furnishings and smaller components can be printed in factories, while entire houses must be printed on-site.
A range of materials can be used for large-scale additive manufacturing. Concrete is a popular choice due to its familiarity and durability. Clay is an intriguing alternative because it can be harvested on-site—which is what the designers of Tecla did.
But plastics and polymers could have the broadest application. These materials are incredibly versatile, and they can be formulated in ways that meet a wide range of specific structural and aesthetic requirements. They can also be produced from recycled and organically derived materials.
Inspiration From NatureBecause additive manufacturing builds layer by layer, using only the material and energy required to make a particular component, it’s a far more efficient building process than “subtractive methods,” which involve cutting away excess material—think milling a wood beam out of a tree.
Even common materials like concrete and plastics benefit from being 3D-printed, since there’s no need for additional formwork or molds.
Most construction materials today are mass-produced on assembly lines that are designed to produce the same components. While reducing cost, this process leaves little room for customization.
Since there is no need for tooling, forms or dies, large-scale additive manufacturing allows each part to be unique, with no time penalty for added complexity or customization.
Another interesting feature of large-scale additive manufacturing is the capability to produce complex components with internal voids. This may one day allow for walls to be printed with conduit or ductwork already in place.
In addition, research is taking place to explore the possibilities of multi-material 3D printing, a technique that could allow windows, insulation, structural reinforcement—even wiring—to be fully integrated into a single printed component.
One of the aspects of additive manufacturing that excites me most is the way in which building layer by layer, with a slowly hardening material, mirrors natural processes, like shell formation.
This opens up windows of opportunity, allowing designers to implement geometries that are difficult to produce using other construction methods, but are common in nature.
Structural frames inspired by the fine structure of bird bones could create lightweight lattices of tubes, with varying sizes reflecting the forces acting upon them. Façades that evoke the shapes of plant leaves might be designed to simultaneously shade the building and produce solar power.
Overcoming the Learning CurveDespite the many positive aspects of large-scale additive manufacturing, there are a number of impediments to its wider adoption.
Perhaps the biggest to overcome is its novelty. There is an entire infrastructure built around traditional forms of construction like steel, concrete and wood, which include supply chains and building codes. In addition, the cost of digital fabrication hardware is relatively high, and the specific design skills needed to work with these new materials are not yet widely taught.
In order for 3D printing in architecture to become more widely adopted, it will need to find its niche. Similar to how word processing helped popularize desktop computers, I think it will be a specific application of large-scale additive manufacturing that will lead to its common use.
Perhaps it will be its ability to print highly efficient structural frames. I also already see its promise for creating unique sculptural façades that can be recycled and reprinted at the end of their useful life.
Either way, it seems likely that some combination of factors will ensure that future buildings will, in some part, be 3D-printed.
A 3D-printed façade in Foshan, China. Image Credit: The Association for Computer Aided Design in ArchitectureThis article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: House Zero in Austin, Texas, is a 2,000-square-foot home that was built with 3D-printed concrete. Casey Dunn/ICON
Our ability to extend human lifespans is improving dramatically, but whether there is any natural limit to how far we can push is an outstanding question. New research contradicts claims that we’re approaching a maximum human lifespan.
The question of whether or not there is a limit to how long humans can live has fascinated scientists for decades. While answering this question is likely to require a better understanding of the physiological process of aging, researchers have long tried to divine trends in demographic data that could give clues as to what the upper limit might be.
One study predicted that the human lifespan is unlikely to go past around 150 years no matter what medical innovations we come up with. Another came to the even more conservative conclusion of 115 years. But a new study that uses novel statistical techniques appears to show that people born between 1900 and 1950 could live much longer than previous analyses suggest, opening up the prospect that no natural limit is currently on the horizon.
“In most of the countries we examined, we project that the maximum age will rise dramatically in the future,” David McCarthy from the University of Georgia told LiveScience. “This will lead to longevity records being broken in the next 40 years or so.”
While previous studies of this kind have often grouped people based on their year of death, the researchers instead lumped together people born in the same year. They used this approach to analyze data from the Human Mortality Database, which contains records of hundreds of millions of people from 19 countries as far back as 1700.
What they found was that those born between 1910 and 1950 saw their risk of dying increase more slowly with each extra year compared to older generations. Because people in these groups have yet to reach extreme old age, it’s impossible to tell how long the oldest will live, but the trend suggests it could be considerably longer than previous generations.
In their paper in PLOS One, the researchers explained that if an upper limit on lifespan did exist, you would expect to see a compression in the distribution of age at death. If fewer people are dying at younger ages, the rate of mortality at older ages would have to increase to compensate.
But that was not what the team found in the data they analyzed, suggesting that mortality was instead being postponed. The authors suggest this sudden step change in lifespans could be due to the rapid improvements made in medicine and public health at the start of the 20th century.
Not everyone is convinced, though. Jan Vijg from the Albert Einstein College of Medicine in New York, who was behind the prediction of a 115-year lifespan, told New Scientist that the researchers’ analysis relies on an assumption that mortality risk increases exponentially up until around 105, after which it plateaus. They aren’t the first to rely on this assumption, but not everyone agrees with it, he says.
It’s also important to remember that no matter what the demographic data shows, human lifespans will ultimately be governed by both their physiology and medical innovation. “The duration of life is at its heart a biological phenomenon, not a mathematical one,” Stuart Jay Olshanky from the University of Illinois Chicago told LiveScience.
However, that’s likely to lead to an upward revision on these predictions, if anything. There’s a growing revolution in the science of aging underway, and research is starting to show that there are a host of medical interventions that could slow or even reverse aging. If the field lives up to its promises, we could be on the cusp of another step change in lifespans similar to the one the researchers predict for those born in the early 20th century.
Image Credit: Matt Bennett / Unsplash
INNOVATIONChatGPT Is About to Revolutionize the Economy. We Need to Decide What That Looks Like.
David Rotman | MIT Technology Review“The optimistic view: it will prove to be a powerful tool for many workers, improving their capabilities and expertise, while providing a boost to the overall economy. The pessimistic one: companies will simply use it to destroy what once looked like automation-proof jobs, well-paying ones that require creative skills and logical reasoning; a few high-tech companies and tech elites will get even richer, but it will do little for overall economic growth.”
ARTIFICIAL INTELLIGENCEWhy Exams Intended for Humans Might Not Be Good Benchmarks for LLMs Like GPT-4
Ben Dickson | VentureBeat“According to a technical report released by OpenAI, GPT-4 performs impressively on bar exams, SAT math tests, and reading and writing exams. However, tests designed for humans may not be good benchmarks for measuring LLMs’ capabilities. Language models encompass knowledge in intricate ways, sometimes producing results that match or exceed average human performance. However, the way they obtain the knowledge and use it is often incompatible with that of humans. That can lead us to draw wrong conclusions from test results.”
COMPUTINGThe Unbelievable Zombie Comeback of Analog Computing
Charles Platt | Wired“When old tech dies, it usually stays dead. No one expects rotary phones or adding machines to come crawling back from oblivion. Floppy diskettes, VHS tapes, cathode-ray tubes—they shall rest in peace. Likewise, we won’t see old analog computers in data centers anytime soon. They were monstrous beasts: difficult to program, expensive to maintain, and limited in accuracy. Or so I thought. Then I came across this confounding statement: Bringing back analog computers in much more advanced forms than their historic ancestors will change the world of computing drastically and forever. Seriously?”
BIOTECHThe Woolly-Mammoth Meatball Is an All-Time Great Food Stunt
Yasmin Tayag | The Atlantic“i‘Typically unexpected, funny, or edgy, stunt foods are ‘pure marketing,’ Mark Lang, a marketing professor at the University of Tampa, told me. They work because they’re bonkers enough to break through the noise of social media and get people talking, he said. But so far, they have caught our attention by twisting familiar items. Lab-grown meat, and all the permutations of protein it makes possible, is pushing us into a new era of stunt marketing, one involving foods people may have never tried.”
SPACEA Big Rover Aims to Be Like ‘UPS for the Moon’
Kenneth Chang | The New York Times“[Founder and CEO Jaret Matthews] said Astrolab would make money by lifting and deploying cargo for customers on the lunar surface. That could include scientific instruments. In the future, the rover could help build lunar infrastructure. ‘Essentially providing what I like to call last-mile mobility on the moon,’ Mr. Matthews said. ‘You can kind of think of it like being UPS for the moon. And in this analogy, Starship is the container ship crossing the ocean, and we’re the local distribution solution.’i”
FUTUREWould Building a Dyson Sphere Be Worth It? We Ran the Numbers.
Paul Sutter | Ars Technica“What if we decided to build a Dyson sphere around our sun? Could we do it? How much energy would it cost us to rearrange our solar system, and how long would it take to get our investment back? Before we put too much thought into whether humanity is capable of this amazing feat, even theoretically, we should decide if it’s worth the effort. Can we actually achieve a net gain in energy by building a Dyson sphere?”
ETHICSCops Used Creepy Clearview AI a Million Times, CEO Says
Mack DeGeurin | Gizmodo“Clearview AI, the shady US facial recognition firm whose surveillance tech is used by at least 2,400 law enforcement agencies, says police have run nearly a million searches using its service. The company’s database of images scraped from social media sites now reportedly numbers around 30 billion, a staggering 50% increase from figures reported just last year. Despite repeated fines and years of pushback from civil liberties organizations, the figures suggest business is still booming for Clearview.”
GOVERNANCEWhy the AI Industry Could Stand to Slow Down a Little
Casey Newton | The Verge“Tech coverage tends to focus on innovation and the immediate disruptions that stem from it. It’s typically less adept at thinking through how new technologies might cause society-level change. And yet the potential for AI to dramatically affect the job market, the information environment, cybersecurity, and geopolitics—to name just four concerns—should gives us all reason to think bigger.”
TECHThat Was Fast! Microsoft Slips Ads Into AI-Powered Bing Chat
Devin Coldewey | TechCrunch“While no one expects Microsoft, or Google, Amazon, Meta and all the others to operate these expensive and computation-hungry language models out of the goodness of their hearts (assuming they have hearts and there is good in them), it would be nice to see a little more thought put into how advertising can better be integrated. When the whole model is changing, the obvious solution—that happens to be a lot like one you used in the old times—is unlikely to be the best.”
AUTOMATIONChatGPT Can Replace the Underpaid Workers Who Train AI, Researchers Say
Chloe Xiang | Motherboard“In a new paper, political science researchers from the University of Zurich found that ChatGPT could outperform crowd-workers who perform text annotation tasks—that is, labeling text to be used in training an AI system. They found that ChatGPT could label text with more accuracy and consistency than human annotators that they found on Mechanical Turk, an Amazon-owned crowdsourcing platform, as well as trained annotators such as research assistants.”
Image Credit: Danist Soh / Unsplash
Self-driving cars are taking longer to arrive on our roads than we thought they would. Auto industry experts and tech companies predicted they’d be here by 2020 and go mainstream by 2021. But it turns out that putting cars on the road without drivers is a far more complicated endeavor than initially envisioned, and we’re still inching very slowly towards a vision of autonomous individual transport.
But the extended timeline hasn’t discouraged researchers and engineers, who are hard at work figuring out how to make self-driving cars efficient, affordable, and most importantly, safe. To that end, a research team from the University of Michigan recently had a novel idea: expose driverless cars to terrible drivers. They described their approach in a paper published last week in Nature.
It may not be too hard for self-driving algorithms to get down the basics of operating a vehicle, but what throws them (and humans) is egregious road behavior from other drivers, and random hazardous scenarios (a cyclist suddenly veers into the middle of the road; a child runs in front of a car to retrieve a toy; an animal trots right into your headlights out of nowhere).
Luckily these aren’t too common, which is why they’re considered edge cases—rare occurrences that pop up when you’re not expecting them. Edge cases account for a lot of the risk on the road, but they’re hard to categorize or plan for since they’re not highly likely for drivers to encounter. Human drivers are often able to react to these scenarios in time to avoid fatalities, but teaching algorithms to do the same is a bit of a tall order.
As Henry Liu, the paper’s lead author, put it, “For human drivers, we might have…one fatality per 100 million miles. So if you want to validate an autonomous vehicle to safety performances better than human drivers, then statistically you really need billions of miles.”
Rather than driving billions of miles to build up an adequate sample of edge cases, why not cut straight to the chase and build a virtual environment that’s full of them?
That’s exactly what Liu’s team did. They built a virtual environment filled with cars, trucks, deer, cyclists, and pedestrians. Their test tracks—both highway and urban—used augmented reality to combine simulated background vehicles with physical road infrastructure and a real autonomous test car, with the augmented reality obstacles being fed into the car’s sensors so the car would react as if they were real.
The team skewed the training data to focus on dangerous driving, calling the approach “dense deep-reinforcement-learning.” The situations the car encountered weren’t pre-programmed, but were generated by the AI, so as it goes along the AI learns how to better test the vehicle.
The system learned to identify hazards (and filter out non-hazards) far faster than conventionally-trained self-driving algorithms. The team wrote that their AI agents were able to “accelerate the evaluation process by multiple orders of magnitude, 10³ to 10⁵ times faster.”
Training self-driving algorithms in a fully virtual environment isn’t a new concept, but the Michigan team’s focus on complex scenarios provides a safe way to expose autonomous cars to dangerous situations. The team also built up a training data set of edge cases for other “safety-critical autonomous systems” to use.
With a few more tools like this, perhaps self-driving cars will be here sooner than we’re now predicting.
Image Credit: Nature/Henry Liu et. al.
A team of scientists has just published evidence in Nature Astronomy for what might be producing mysterious bursts of radio waves coming from distant galaxies, known as fast radio bursts or FRBs.
Two colliding neutron stars—each the super-dense core of an exploded star—produced a burst of gravitational waves when they merged into a “supramassive” neutron star. The team found that two and a half hours later they produced an FRB when the neutron star collapsed into a black hole.
Or so they think. The key piece of evidence that would confirm or refute their theory—an optical or gamma-ray flash coming from the direction of the fast radio burst—vanished almost four years ago. In a few months, they might get another chance to find out if they are correct.
Brief and PowerfulFRBs are incredibly powerful pulses of radio waves from space lasting about a thousandth of a second. Using data from a radio telescope in Australia, the Australian Square Kilometre Array Pathfinder (ASKAP), astronomers have found that most FRBs come from galaxies so distant, light takes billions of years to reach us. But what produces these radio wave bursts has been puzzling astronomers since an initial detection in 2007.
The best clue comes from an object in our galaxy known as SGR 1935+2154. It’s a magnetar, which is a neutron star with magnetic fields about a trillion times stronger than a fridge magnet. On April 28 2020, it produced a violent burst of radio waves—similar to an FRB, although less powerful.
Astronomers have long predicted that two neutron stars—a binary—merging to produce a black hole should also produce a burst of radio waves. The two neutron stars will be highly magnetic, and black holes cannot have magnetic fields. The idea is the sudden vanishing of magnetic fields when the neutron stars merge and collapse to a black hole produces a fast radio burst. Changing magnetic fields produce electric fields—it’s how most power stations produce electricity. And the huge change in magnetic fields at the time of collapse could produce the intense electromagnetic fields of an FRB.
Artist’s impression of a fast radio burst traveling through space and reaching Earth. Image Credit: ESO/M. Kornmesser, CC BYThe Search for the Smoking GunTo test this idea, Alexandra Moroianu, a masters student at the University of Western Australia, looked for merging neutron stars detected by the Laser Interferometer Gravitational-Wave Observatory (LIGO) in the US. The gravitational waves LIGO searches for are ripples in spacetime, produced by the collisions of two massive objects, such as neutron stars.
LIGO has found two binary neutron star mergers. Crucially, the second, known as GW190425, occurred when a new FRB-hunting telescope called CHIME was also operational. However, being new, it took CHIME two years to release its first batch of data. When it did so, Moroianu quickly identified a fast radio burst called FRB 20190425A which occurred only two and a half hours after GW190425.
Exciting as this was, there was a problem—only one of LIGO’s two detectors was working at the time, making it very uncertain where exactly GW190425 had come from. In fact, there was a five percent chance this could just be a coincidence.
Worse, the Fermi satellite, which could have detected gamma rays from the merger—the “smoking gun” confirming the origin of GW190425—was blocked by Earth at the time.
CHIME, the Canadian Hydrogen Intensity Mapping Experiment, has turned out to be uniquely suited to detecting FRBs. Image Credit: Andre Renard/Dunlap Institute/CHIME CollaborationUnlikely to Be a CoincidenceHowever, the critical clue was that FRBs trace the total amount of gas they have passed through. We know this because high-frequency radio waves travel faster through the gas than low-frequency waves, so the time difference between them tells us the amount of gas.
Because we know the average gas density of the universe, we can relate this gas content to distance, which is known as the Macquart relation. And the distance travelled by FRB 20190425A was a near-perfect match for the distance to GW190425. Bingo!
So have we discovered the source of all FRBs? No. There are not enough merging neutron stars in the Universe to explain the number of FRBs—some must still come from magnetars, like SGR 1935+2154 did.
And even with all the evidence, there’s still a 1 in 200 chance this could all be a giant coincidence. However, LIGO and two other gravitational wave detectors, Virgo and KAGRA, will turn back on in May this year, and be more sensitive than ever, while CHIME and other radio telescopes are ready to immediately detect any FRBs from neutron star mergers.
In a few months, we may find out if we’ve made a key breakthrough—or if it was just a flash in the pan.
Clancy W. James would like to acknowledge Alexandra Moroianu, the lead author of the study; his co-authors, Linqing Wen, Fiona Panther, Manoj Kovalem (University of Western Australia), Bing Zhang and Shunke Ai (University of Nevada); and his late mentor, Jean-Pierre Macquart, who experimentally verified the gas-distance relation, which is now named after him.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: CSIRO/Alex Cherney
Universal basic income (UBI) schemes are often dismissed as being too expensive to implement on a large scale, but several cities are trying them out among small subsets of their populations. Giving people even a small financial leg up can go a long way towards bridging the gap between surviving and thriving. The biggest guaranteed income pilot in the US is currently underway in Chicago, where 500 families are receiving $500 per month, with no strings attached, for 12 months.
A far bigger UBI trial will be launching in India later this year. Announced last week by finance minister Palanivel Thiaga Rajan, the trial will take place in Tamil Nadu, the country’s southernmost state and its seventh-most populous with 81.5 million people.
Called Magalir Urimai Thogai, which in Tamil means “Women’s Right to Assistance,” the trial will give the female heads of eligible households 1,000 rupees per month. That’s somewhere between $12 to $13. It doesn’t sound like much, but the average annual per capita income in Tamil Nadu is around 225,000 rupees ($2,733). That breaks down to $52 a week, and it’s an average; the lowest-income families earn far less.
Specific eligibility guidelines for the program haven’t been finalized yet, but it’s geared towards families living below the poverty line. Recipients will be selected from the state’s TIPPS system (Tamil Nadu Integrated Poverty Portal Service), where data from income and population surveys is stored.
The state’s minister for social welfare and female empowerment, P Geetha Jeevan, said, “The income benefit aimed at supporting impoverished families will not cover the rich, government employees, and a few others. Approximately 80 to 90 lakh women are expected to avail themselves of this benefit.” (A lakh is a unit in the Indian numbering system equal to 100,000—so the pilot could benefit up to 9 million women).
While guaranteed income pilots in the US and other countries tend not to be gender-specific, the decision to give these payments exclusively to women in India was very intentional. In short, it’s part of an ongoing effort by the government to reduce long-standing, pronounced gender inequality in the country.
While gender norms won’t change overnight simply because female heads of households receive payments instead of males, this will help break down old stereotypes, as well as giving women a sense of agency and an incentive to understand more about finance. Similarly, a program called Ujjwala was launched in 2016 to provide gas stoves and subsidized cooking gas to poor families—but only women could receive the payments.
Since then, the price of gas has shot up, making refills too expensive for many families (even with the subsidy) and causing some to return to traditional wood-fired stoves, which emit toxic fumes that are harmful to human health. Speaking about the guaranteed income program, Rajan said, “This will be of great help for women heads of families who have been affected adversely by the steep increase in cooking gas prices by the union government and the overall inflation.”
Details about how the results of the payment program will be monitored haven’t been released yet, but as with all such trials, the hope is that by giving families a small extra cushion of financial security, more of their basic needs will be covered, freeing up time and resources to devote to additional pursuits.
Tamil Nadu’s guaranteed income pilot is slated to launch in September of this year.
Image Credit: Joshuva Daniel on Unsplash
Seven mice just joined the pantheon of offspring created from same-sex parents—and opened the door to offspring born from a single parent.
In a study published in Nature, researchers described how they scraped skin cells from the tails of male mice and used them to create functional egg cells. When fertilized with sperm and transplanted into a surrogate, the embryos gave rise to healthy pups, which grew up and had babies of their own.
The study is the latest in a decade-long attempt to rewrite reproduction. Egg meets sperm remains the dogma. What’s at play is how the two halves are generated. Thanks to iPSC (induced pluripotent stem cell) technology, scientists have been able to bypass nature to engineer functional eggs, reconstruct artificial ovaries, and give rise to healthy mice from two mothers. Yet no one has been able to crack the recipe of healthy offspring born from two dads.
Enter Dr. Katsuhiko Hayashi at Kyushu University, who has led the ambitious goal to engineer gametes—sperm and egg—outside the body. His solution came from a clever hack. When grown inside petri dishes, iPSC cells tend to lose bundles of their DNA, called chromosomes. Normally, this is a massive headache because it disrupts the cell’s genetic integrity.
Hayashi realized he could hijack the mechanism. Selecting for cells that shed the Y chromosome, the team nurtured the cells until they fully developed into mature egg cells. The cells—which started as male skin cells—eventually developed into normal mice after fertilization with normal sperm.
“Murakami and co-workers’ protocol opens up new avenues in reproductive biology and fertility research,” said Drs. Jonathan Bayerl and Diana Laird at the University of California, San Francisco (UCSF), who were not involved in the study.
Whether the strategy will work in humans remains to be seen. The success rate in mice was very low at just a snippet over one percent. Yet the study is a proof of concept that further pushes the boundaries of the reproductive realm of possibilities. And perhaps more immediately, the underlying technology can help tackle some of our most prevalent chromosomal disorders, such as Down syndrome.
“This is a very important breakthrough for the generation of eggs and sperm from stem cells,” said Dr. Rod Mitchell at the MRC Centre for Reproductive Health, University of Edinburgh, who was not involved in the study.
A Reproductive RevolutionHayashi is a long-time veteran at transforming reproductive technologies. In 2020, his team described genetic alterations that help cells mature into egg cells inside a dish. A year later, they reconstructed ovary cells that nurtured fertilized eggs into healthy mouse pups.
At the core of these technologies are iPSCs. Using a chemical bath, scientists can transform mature cells, such as skin cells, back into a stem-cell-like state. iPSCs are basically biological playdough: with a soup of chemical “kneading,” they can be coaxed and fashioned into nearly any type of cell.
Because of their flexibility, iPSCs are also hard to control. Like most cells, they divide. But when kept inside a petri dish for too long, they rebel and either shed—or duplicate—some of their chromosomes. This teenage anarchy, called aneuploidy, is the bane of scientists’ work when trying to keep a uniform population of cells.
But as the new study shows, that molecular rebellion is a gift for generating eggs from male cells.
X Meets Y and…Meets O?Let’s talk sex chromosomes.
Most people have either XX or XY. Both X and Y are chromosomes, which are large bundles of DNA—picture threads wrapped around a spool. Biologically, XX usually generates eggs, whereas XY normally produces sperm.
But here’s the thing: scientists have long known that both type of cells start from the same stock. Dubbed primordial germ cells, or PGCs, these cells don’t rely on either X or Y chromosomes, but rather on their surrounding chemical environment for their initial development, explained Bayerl and Laird.
In 2017, for example, Hayashi’s team transformed embryonic stem cells into PGCs, which when mixed with fetal ovary or testes cells matured into either artificial eggs or sperm.
Here, the team took on the harder task of transforming an XY cell into an XX one. They started with a group of embryonic stem cells from mice that shed their Y chromosomes—a rare and controversial resource. Using a glow-in-the-dark tag that grabs only onto X chromosomes, they could monitor how many copies there were inside a cell based on light intensity (remember, XX will shine brighter than XY).
After growing the cells for eight rounds inside petri dishes, the team found that roughly six percent of the cells sporadically lost their Y chromosome. Rather than XY, they now only harbored one X—like missing half of a chopstick pair. The team then selectively coaxed these cells, dubbed XO, to divide.
The reason? Cells duplicate their chromosomes before splitting into two new ones. Because the cells only have one X chromosome, after duplication some of the daughter cells will end up with XX—in other words, biologically female. Adding a drug called reversine helped the process along, increasing the number of XX cells.
The team then tapped into their previous work. They converted XX cells into PGC-like cells—the ones that can develop into egg or sperm—and then added fetal ovary cells to push the transformed male skin cells into mature eggs.
As the ultimate test, they injected sperm from a normal mouse into the lab-made eggs. With the help of a female surrogate, the blue-sky experiment produced over a half-dozen pups. Their weights were similar to mice born the traditional way, and their surrogate mom developed a healthy placenta. All of the pups grew into adulthood and had babies of their own.
Pushing BoundariesThe tech is still in its early days. For one, its success rate is extremely low: only 7 out of 630 transferred embryos lived to be full-grown adults. With a mere 1.1 percent chance at succeeding—especially in mice—it’s a tough sell for bringing the technology to male human couples. Although the baby mice seemed relatively normal in terms of weight and could reproduce, they could also harbor genetic or other deficiencies—something that the team wants to further investigate.
“There are big differences between a mouse and the human,” said Hayashi at an earlier conference.
That said, reproduction aside, the study may immediately help to understand chromosomal disorders. Down syndrome, for example, is caused by an extra copy of chromosome 21. In the study, the team found that treating mouse embryonic stem cells harboring a similar defect with reversine—the drug that helps convert XY to XX cells—rid the mice of the extra copy without affecting other chromosomes. It’s far from being ready for human use. However, the technology could help other scientists hunt down preventative or screening measures for similar chromosomal disorders.
But perhaps what’s most intriguing is where the technology can take reproductive biology. In an audacious experiment, the team showed that cells from a single male iPSC line can birth offspring—pups that grew into adulthood.
With the help of surrogate mothers, “it also suggests that a single man could have a biological child…in the far future,” said Dr. Tetsuya Ishii, a bioethicist at Hokkaido University. The work could also propel bioconservation, propagating endangered mammals from just a single male.
Hayashi is well aware of the ethics and social implications of his work. But for now, his focus is on helping people and deciphering—and rewriting—the rules of reproduction.
The study marks “a milestone in reproductive biology,” said Bayerl and Laird.
Image Credit: Katsuhiko Hayashi, Osaka University
At the end of last year, Israeli cultured meat company Believer Meats broke ground on a 200,000-square-foot factory outside Raleigh, North Carolina. The facility will be the biggest cultured meat factory in the world (well, unless a bigger one goes up before it’s done, which is unlikely).
However, the sale of cultured meat isn’t fully legal in the US yet (in fact, the only countries where the meat can be sold right now are Singapore and Israel), so regulations are going to need to keep pace with production capacity to make such facilities worth building. Last week California-based Good Meat took a step in this direction, receiving a crucial FDA approval for sale of its cultured chicken in the US.
Cultured meat is made by taking muscle cells from a live animal (without harming it) and feeding those cells a mixture of nutrients and growth factors to make them multiply, differentiate, and grow to form muscle tissue. The harvested tissue then needs to be refined and shaped into a final product, which can involve extrusion cooking, molding, or 3D printing.
Good Meat was the first company in the world to start selling cultured meat, with its chicken hitting the Singaporean market in 2020. This past January the company hit another milestone when the Singapore Food Agency granted them approval to sell serum-free meat in Singapore (“serum-free” means they can use synthetic ingredients in their production process, specifically eliminating fetal bovine serum, which makes animal cells duplicate).
Now Good Meat has made headway in what it hopes will be its biggest market, the US. They received an FDA approval called a No Questions letter, which states that after conducting a thorough evaluation of the company’s meat, the agency concluded it’s safe for consumers to eat. Besides meeting microbiological and purity standards (the press release notes that cultured chicken’s microbiological levels are “significantly cleaner” than conventional chicken), the evaluation found that Good Meat’s chicken contains “high protein content, a well-balanced amino acid profile, and is a rich source of minerals.”
Good Meat isn’t the first company to receive this approval in the US. Its competitor Upside Foods got a No Questions letter for its cultured chicken last November. Their 53,000-square-foot production center in the Bay Area will eventually be able to produce more than 400,000 pounds of meat, poultry, and seafood per year. Before becoming available in grocery stores, Upside’s chicken will be introduced to consumers in restaurants, starting with an upscale restaurant in San Francisco whose chef is Michelin-starred.
Similarly, Good Meat plans to launch its cultured chicken at a Washington DC restaurant owned by celebrity chef José Andrés. Before that can happen, though, the company has to work with the US Department of Agriculture to receive additional approvals for its production facilities and its product.
The company is building a demonstration plant in Singapore, and announced plans last year to build a large-scale facility in the US with an annual production capacity of 30 million pounds of meat (which means it will be bigger than the Believer Meats plant in North Carolina).
Good Meat will have its work cut out for it, as there are more than 80 other companies vying for a slice of the lab-grown meat market, which is projected to reach a value of $12.7 billion by 2030. Given that all of its competitors will have to go through the FDA and USDA approvals process, though, Good Meat has a leg up.
Image Credit: Good Meat
If computer chips make the modern world go around, then Nvidia and TSMC are flywheels keeping it spinning. It’s worth paying attention when the former says they’ve made a chipmaking breakthrough, and the latter confirms they’re about to put it into practice.
At Nvidia’s GTC developer conference this week, CEO Jensen Huang said Nvidia has developed software to make a chipmaking step, called inverse lithography, over 40 times faster. A process that usually takes weeks can now be completed overnight, and instead of requiring some 40,000 CPU servers and 35 megawatts of power, it should only need 500 Nvidia DGX H100 GPU-based systems and 5 megawatts.
“With cuLitho, TSMC can reduce prototype cycle time, increase throughput and reduce the carbon footprint of their manufacturing, and prepare for 2nm and beyond,” he said.
Nvidia partnered with some of the biggest names in the industry on the work. TSMC, the largest chip foundry in the world, plans to qualify the approach in production this summer. Meanwhile, chip designer, Synopsis, and equipment maker, ASML, said in a press release they will integrate cuLitho into their chip design and lithography software.
What Is Inverse Lithography?To fabricate a modern computer chip, makers shine ultraviolet light through intricate “stencils” to etch billions of patterns—like wires and transistors—onto smooth silicon wafers at near-atomic resolutions. This step, called photolithography, is how every new chip design, from Nvidia to Apple to Intel, is manifested physically in silicon.
The machines that make it happen, built by ASML, cost hundreds of millions of dollars and can produce near-flawless works of nanoscale art on chips. The end product, an example of which is humming away near your fingertips as you read this, is probably the most complex commodity in history. (TSMC churns out a quintillion transistors every six months—for Apple alone.)
To make more powerful chips, with ever-more, ever-smaller transistors, engineers have had to get creative.
Remember that stencil mentioned above? It’s the weirdest stencil you’ve ever seen. Today’s transistors are smaller than the wavelength of light used to etch them. Chipmakers have to use some extremely clever tricks to design stencils—or technically, photomasks—that can bend light into interference patterns whose features are smaller than the light’s wavelength and perfectly match the chip’s design.
Whereas photomasks once had a more one-to-one shape—a rectangle projected a rectangle—they’ve necessarily become more and more complicated over the years. The most advanced masks these days are more like mandalas than simple polygons.
“Stencils” or photomasks have become more and more complicated as the patterns they etch have shrunk into the atomic realm. Image Credit: NvidiaTo design these advanced photomask patterns, engineers reverse the process.
They start with the design they want, then stuff it through a wicked mess of equations describing the physics involved to design a suitable pattern. This step is called inverse lithography, and as the gap between light wavelength and feature size has increased, it’s become increasingly crucial to the whole process. But as the complexity of photomasks increases, so too does the computing power, time, and cost required to design them.
“Computational lithography is the largest computation workload in chip design and manufacturing, consuming tens of billions of CPU hours annually,” Huang said. “Massive data centers run 24/7 to create reticles used in lithography systems.”
In the broader category of computational lithography—the methods used to design photomasks—inverse lithography is one of the newer, more advanced approaches. Its advantages include greater depth of field and resolution and should benefit the entire chip, but due its heavy computational lift, it’s currently only used sparingly.
A Library in ParallelNvidia aims to reduce that lift by making the computation more amenable to graphics processing units, or GPUs. These powerful chips are used for tasks with lots of simple computations that can be completed in parallel, like video games and machine learning. So it isn’t just about running existing processes on GPUs, which only yields a modest improvement, but modifying those processes specifically for GPUs.
That’s what the new software, cuLitho, is designed to do. The product, developed over the last four years, is a library of algorithms for the basic operations used in inverse lithography. By breaking inverse lithography down into these smaller, more repetitive computations, the whole process can now be split and parallelized on GPUs. And that, according to Nvidia, significantly speeds everything up.
A new library of inverse lithography algorithms can speed up the process by breaking it down into smaller tasks and running them in parallel on GPUs. Image Credit: Nvidia“If [inverse lithography] was sped up 40x, would many more people and companies use full-chip ILT on many more layers? I am sure of it,” said Vivek Singh, VP of Nvidia’s Advanced Technology Group, in a talk at GTC.
With a speedier, less computationally hungry process, makers can more rapidly iterate on experimental designs, tweak existing designs, make more photomasks per day, and generally, expand the use of inverse lithography to more of the chip, he said.
This last detail is critical. Wider use of inverse lithography should reduce print errors by sharpening the projected image—meaning chipmakers can churn out more working chips per silicon wafer—and be precise enough to make features at 2 nanometers and beyond.
It turns out making better chips isn’t all about the hardware. Software improvements, like cuLitho or the increased use of machine learning in design, can have a big impact too.
Image Credit: Nvidia
ARTIFICIAL INTELLIGENCEOpenAI Connects ChatGPT to the Internet
Kyle Wiggers | TechCrunch“[This week, OpenAI] launched plugins for ChatGPT, which extend the bot’s functionality by granting it access to third-party knowledge sources and databases, including the web. Easily the most intriguing plugin is OpenAI’s first-party web-browsing plugin, which allows ChatGPT to draw data from around the web to answer the various questions posed to it.”
COMPUTINGNvidia Speeds Key Chipmaking Computation by 40x
Samuel K. Moore | IEEE Spectrum“Called inverse lithography, it’s a key tool that allows chipmakers to print nanometer-scale features using light with a longer wavelength than the size of those features. Inverse lithography’s use has been limited by the massive size of the needed computation. Nvidia’s answer, cuLitho, is a set of algorithms designed for use with GPUs, turns what has been two weeks of work into an overnight job.”
DIGITAL MEDIAEpic’s New Motion-Capture Animation Tech Has to Be Seen to Be Believed
Kyle Orland | Ars Technica“Epic’s upcoming MetaHuman facial animation tool looks set to revolutionize [the]…labor- and time-intensive workflow [of motion-capture]. In an impressive demonstration at Wednesday’s State of Unreal stage presentation, Epic showed off the new machine-learning-powered system, which needed just a few minutes to generate impressively real, uncanny-valley-leaping facial animation from a simple head-on video taken on an iPhone.”
TRANSPORTATIONUnited to Fly Electric Air Taxis to O’Hare Beginning in 2025
Stefano Esposito | Chicago Sun Times“The trip between O’Hare and the Illinois Medical District is expected to take about 10 minutes, according to California-based Archer Aviation, which is partnering with United Airlines. …An Archer spokesman said they hope to make the fare competitive with Uber Black, a ride-hailing service that provides luxury vehicles and top-rated drivers to customers. On Thursday afternoon, an Uber Black ride from for Vertiport to O’Hare was $101.”
ETHICSThese New Tools Let You See for Yourself How Biased AI Image Models Are
Melissa Heikkilä | MIT Technology Review“Popular AI image-generating systems notoriously tend to amplify harmful biases and stereotypes. But just how big a problem is it? You can now see for yourself using interactive new online tools. (Spoiler alert: it’s big.) The tools, built by researchers at AI startup Hugging Face and Leipzig University and detailed in a non-peer-reviewed paper, allow people to examine biases in three popular AI image-generating models: DALL-E 2 and the two recent versions of Stable Diffusion.”
VIRTUAL REALITYBMW’s New Factory Doesn’t Exist in Real Life, but It Will Still Change the Car Industry
Jesus Diaz | Fast Company“Before construction on [a new car] factory begins, thousands of engineers draw millions of CAD drawings and meet for thousands of hours. Worse yet, they know that no amount of planning will prevent a long list of bugs once the factory finally opens, which can result in millions of dollars lost every day until the bugs are resolved. At least, that’s how it used to work. This is all about to change thanks to the world’s first virtual factory, a perfect digital twin of BMW’s future 400-hectare plant in Debrecen, Hungary, which will reportedly produce around 150,000 vehicles every year when it opens in 2025.”
ENERGYFusion Power Is Coming Back Into Fashion
Editorial Staff | The Economist“[Forty two companies] think they can succeed, where others failed, in taking fusion from the lab to the grid—and do so with machines far smaller and cheaper than the latest intergovernmental behemoth, ITER, now being built in the south of France at a cost estimated by America’s energy department to be $65bn. In some cases that optimism is based on the use of technologies and materials not available in the past; in others, on simpler designs.”
ENVIRONMENTPlastic Paving: Egyptian Startup Turns Millions of Bags Into Tiles
Editorial Staff | Reuters“An Egyptian startup is aiming to turn more than 5 billion plastic bags into tiles tougher than cement as it tackles the twin problems of tons of waste entering the Mediterranean Sea and high levels of building sector emissions. ‘So far, we have recycled more than 5 million plastic bags, but this is just the beginning,’ TileGreen co-founder Khaled Raafat told Reuters. ‘We aim that by 2025, we will have recycled more than 5 billion plastic bags.’ ”
Image Credit: BoliviaInteligente / Unsplash
There’s a common perception that artificial intelligence (AI) will help streamline our work. There are even fears that it could wipe out the need for some jobs altogether.
But in a study of science laboratories I carried out with three colleagues at the University of Manchester, the introduction of automated processes that aim to simplify work—and free people’s time—can also make that work more complex, generating new tasks that many workers might perceive as mundane.
In the study, published in Research Policy, we looked at the work of scientists in a field called synthetic biology, or synbio for short. Synbio is concerned with redesigning organisms to have new abilities. It is involved in growing meat in the lab, in new ways of producing fertilizers, and in the discovery of new drugs.
Synbio experiments rely on advanced robotic platforms to repetitively move a large number of samples. They also use machine learning to analyze the results of large-scale experiments.
These, in turn, generate large amounts of digital data. This process is known as “digitalization,” where digital technologies are used to transform traditional methods and ways of working.
Some of the key objectives of automating and digitalizing scientific processes are to scale up the science that can be done while saving researchers time to focus on what they would consider more “valuable” work.
Paradoxical ResultHowever, in our study, scientists were not released from repetitive, manual, or boring tasks as one might expect. Instead, the use of robotic platforms amplified and diversified the kinds of tasks researchers had to perform. There are several reasons for this.
Among them is the fact that the number of hypotheses (the scientific term for a testable explanation for some observed phenomenon) and experiments that needed to be performed increased. With automated methods, the possibilities are amplified.
Scientists said it allowed them to evaluate a greater number of hypotheses, along with the number of ways that scientists could make subtle changes to the experimental set-up. This had the effect of boosting the volume of data that needed checking, standardizing, and sharing.
Also, robots needed to be “trained” in performing experiments previously carried out manually. Humans, too, needed to develop new skills for preparing, repairing, and supervising robots. This was done to ensure there were no errors in the scientific process.
Scientific work is often judged on output such as peer-reviewed publications and grants. However, the time taken to clean, troubleshoot, and supervise automated systems competes with the tasks traditionally rewarded in science. These less valued tasks may also be largely invisible—particularly because managers are the ones who would be unaware of mundane work due to not spending as much time in the lab.
The synbio scientists carrying out these responsibilities were not better paid or more autonomous than their managers. They also assessed their own workload as being higher than those above them in the job hierarchy.
Wider LessonsIt’s possible these lessons might apply to other areas of work too. ChatGPT is an AI-powered chatbot that “learns” from information available on the web. When prompted by questions from online users, the chatbot offers answers that appear well-crafted and convincing.
According to Time magazine, in order for ChatGPT to avoid returning answers that were racist, sexist, or offensive in other ways, workers in Kenya were hired to filter toxic content delivered by the bot.
There are many often invisible work practices needed for the development and maintenance of digital infrastructure. This phenomenon could be described as a “digitalization paradox.” It challenges the assumption that everyone involved or affected by digitalization becomes more productive or has more free time when parts of their workflow are automated.
Concerns over a decline in productivity are a key motivation behind organizational and political efforts to automate and digitalize everyday work. But we should not take promises of gains in productivity at face value.
Instead, we should challenge the ways we measure productivity by considering the invisible types of tasks humans can accomplish, beyond the more visible work that is usually rewarded.
We also need to consider how to design and manage these processes so that technology can more positively add to human capabilities.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Gerd Altmann from Pixabay
Reducing the cost of space launches will be critical if we want humanity to have a more permanent presence beyond orbit. The partially successful launch of the first 3D-printed rocket could be a significant step in that direction.
Getting stuff into space is dramatically cheaper than it used to be thanks to a wave of innovation in the private space industry led by SpaceX. More affordable launches have brought on a rapid expansion in access to space and made a host of new space-based applications feasible. But costs are still a major barrier.
That’s largely because rockets are incredibly expensive and difficult to build. A promising way round this is to use 3D printing to simplify the design and manufacturing process. SpaceX has experimented with the idea for years, and the engines on Rocket Lab’s Electron launch vehicle are almost entirely 3D-printed.
But one company wants to take things even further. Relativity Space has built one of the largest metal 3D printers in the world and uses it to fabricate almost all of its Terran 1 rocket. The rocket blasted off for the first time yesterday, and while the launch vehicle didn’t quite make orbit, it survived max-q, or the part of flight when the rocket is subjected to maximum mechanical stress.
“Today is a huge win, with many historic firsts,” the company said in a tweet following the launch. “We successfully made it through max-q, the highest stress state on our printed structures. This is the biggest proof point for our novel additive manufacturing approach.”
This was the company’s third bite at the cherry after two previous launches were called off earlier in the month. The rocket lifted off from a launchpad at the US Space Force’s launch facility in Cape Canaveral, Florida at 8:25 pm (EST) and flew for about three minutes.
Shortly after making it through max-q and the successful separation of the second stage from the booster, the rocket’s engine cut out due to what the company cryptically referred to as “an anomaly,” though it promised to provide updates once flight data has been analyzed.
While that meant Terran 1 didn’t make it into orbit, the launch is nonetheless likely to be seen as a success. It’s fairly common for the first launch of a new rocket to go awry—Space X’s first three launches failed—so getting off the launch pad and passing key milestones like max-q and first stage separation are significant achievements.
This is particularly important for Relativity Space, which is taking a radically different approach to manufacturing its rockets compared to competitors. Prior to the launch, cofounder Tim Ellis said the company’s main goal was to prove the structural integrity of their 3D-printed design.
“We have already proven on the ground what we hope to prove in-flight—that when dynamic pressures and stresses on the vehicle are highest, 3D printed structures can withstand these forces,” he said in a tweet. “This will essentially prove the viability of using additive manufacturing tech to produce products that fly.”
There is a lot that is novel about Relativity’s design. At present, roughly 85 percent of the structure by mass is 3D-printed, but the company hopes to push that to 95 percent in future iterations. This has allowed Relativity to use 100 times fewer parts than traditional rockets and go from raw materials to a finished product in just 60 days.
The engines also run on a mixture of liquid methane and liquid oxygen, which is the same technology SpaceX is pursuing for its massive Starship rocket. This fuel mix is seen as the most promising for Mars exploration as it can be produced on the red planet itself, eliminating the need to carry fuel for the return journey.
But while the 110-foot-tall Terran 1 can carry up to 2,756 pounds to low-Earth orbit, and Relativity is selling rides on the rocket for around $12 million, it is really a test bed for a more advanced rocket. That rocket, the Terran R, will be 216 feet tall and able to carry 44,000 pounds when it makes it onto the launchpad as early as 2024.
Relativity isn’t the only company working hard to bring 3D printing to the space industry.
California startup, Launcher, has created a satellite platform called Orbiter that’s powered by 3D-printed rocket engines, and Colorado-based Ursa Major is 3D printing rocket engines it hopes others will use in their vehicles. At the same time, UK-based Orbex is using metal 3D printers from German manufacturer EOS to manufacture entire rockets.
Now that 3D-printed rockets have passed their first true test and made it into space, don’t be surprised to see more companies following in the footsteps of these early pioneers.
Image Credit: Relativity Space
The hype around artificial intelligence has been building for years, and you could say it reached a crescendo with OpenAI’s recent release of ChatGPT (and now GPT-4). It only took two months for ChatGPT to reach 100 million users, making it the fastest-growing consumer application in history (it took Instagram two and a half years to gain the same user base, and TikTok nine months).
But in Ian Beacraft’s opinion, we’re in an AI hype bubble, way above the top of the peak of inflated expectations on the Gartner Hype Cycle. But it may be justified, because the AI tools we’re seeing really do have the power to overhaul the way we work, learn, and create value.
Beacraft is the founder of the strategic foresight agency Signal & Cipher and co-owner of a production studio that designs virtual worlds. In a talk at South by Southwest last week, he shared his predictions of how AI will shape society in the years and decades to come.
A Revolution in Knowledge WorkBeacraft pointed out that with the Industrial Revolution we were able to take skills of human labor and amplify them far beyond what the human body is capable of. “Now we’re doing the same thing with knowledge work,” he said. “We’re able to do so much more, put so much more power behind it.” The Industrial Revolution mechanized skills, and today we’re digitizing skills. Digitized skills are programmable, composable, and upgradeable—and AI is taking it all to another level.
Say you want to write a novel in the style of a specific writer. You could prompt ChatGPT to do so, be it by the sentence, paragraph, or chapter, then tweak the language to your liking (whether that’s cheating or some form of plagiarism is another issue, and a pretty significant one); you’re programming the algorithm to extract years worth of study and knowledge—years that you don’t have to put in. Composable means you can stack skills on top of each other, and upgradeable means anytime anytime an AI gets an upgrade, so do you. “You didn’t have to go back to school for it, but all of a sudden you have new skills that came from the upgrade,” Beacraft said.
The Era of the GeneralistDue to these features, he believes AI is going to turn us all into creative generalists. Right now we’re told to specialize from an early age and build expertise in one area—but what happens once AI can quickly outpace us in any domain? Will it still make sense to become an expert in a single field?
“Those who have expertise and depth in several domains, and interest and passion and curiosity across a broad swathe—those are the people who are going to dominate the next era,” Beacraft said. “When you have an understanding of how something works, you can now produce for it. You don’t have to have expertise in all the different layers to make that happen. You can know how the general territory or field operate, then have machines abstract the rest of the skills.”
For example, a graphic designer who draws a comic book could use AI-powered design tools to turn that comic book into a 3D production, and he doesn’t have to know 3D modeling, camera movement, blending, or motion capture; AI now enables just one person to perform all of the virtual production elements. “This wouldn’t have been possible a couple years ago, and now—with some effort—it is,” Beacraft said. The video below was created entirely by one person using generative AI, including the imagery, sound, motion, and talk track.
Generative AI tools are also starting to learn how to use other tools themselves, and they’re only going to get better at it. ChatGPT, for example, isn’t very good at hard science, but it could pass those kinds of questions off to something like WolframAlpha and include the tool’s answer in its reply.
This is not only going to change our work, Beacraft said, it’s going to change our relationship with work. Right now, organizations expect incremental employee improvement in narrowly-defined roles. Job titles like designer, accountant, or project manager have key performance indicators that typically improve two to three percent per year. “But if employees only grow incrementally, how can organizations expect exponential growth?” Beacraft asked.
AI will take our traditional job roles and make them horizontal, giving us the ability to flex in any direction. As a result, we’ll have just-in-time skills and expertise on demand. “We will not lose our jobs, we will lose our job descriptions,” Beacraft said. “When organizations have teams of people working horizontally, all that new capability is net new, not incremental—and all of a sudden you have exponential growth.”
More Work, Not LessThat growth could do the opposite of what the predominant narrative tells us: that AI, robotics, and automation will take over various kinds of work and do away with our jobs. But AI could very well end up creating more work for us.
For example, team of scientists using AI to help them run experiments more efficiently could increase the number of experiments they perform—but then they have more results, more data to analyze, and more work sifting through all this information to ultimately draw a conclusion or find what they’re looking for. But hey—AI is getting good at handling extra administrative work, too.
We may be in an AI hype bubble, but this technology is reaching more people than it ever has before. While there are certainly nefarious uses for generative AI—just look at all the students trying to turn in essays written by ChatGPT, or how deepfakes are becoming harder to pinpoint—there are as many or more productive uses that will impact society, the economy, and our lives in positive ways.
“It’s not just about data and information, it’s about how these AIs can help us shape the world,” Beacraft said. “It’s about how we project what we want to create onto the world around us.”
Image Credit: DeepMind / Unsplash
The race to solve every protein structure just welcomed another tech giant: Meta AI.
A research offshoot of Meta, known for Facebook and Instagram, the team came onto the protein shape prediction scene with an ambitious goal: to decipher the “dark matter” of the protein universe. Often found in bacteria, viruses, and other microorganisms, these proteins lounge in our everyday environments but are complete mysteries to science.
“These are the structures we know the least about. These are incredibly mysterious proteins. I think they offer the potential for great insight into biology,” said senior author Dr. Alexander Rives to Nature.
In other words, they’re a treasure trove of inspiration for biotechnology. Hidden in their secretive shapes are keys for designing efficient biofuels, antibiotics, enzymes, or even entirely new organisms. In turn, the data from protein predictions could further train AI models.
At the heart of Meta’s new AI, dubbed ESMFold, is a large language model. It might sound familiar. These machine learning algorithms have taken the world by storm with the rockstar chatbot ChatGPT. Known for its ability to generate beautiful essays, poems, and lyrics with simple prompts, ChatGPT—and the recently-launched GPT-4—are trained with millions of publicly-available texts. Eventually the AI learns to predict letters, words, and even write entire paragraphs and, in the case of Bing’s similar chatbot, hold conversations that sometimes turn slightly unnerving.
The new study, published in Science, bridges the AI model with biology. Proteins are made of 20 “letters.” Thanks to evolution, the sequence of letters help generate their ultimate shapes. If large language models can easily construe the 26 letters of the English alphabet into coherent messages, why can’t they also work for proteins?
Spoiler: they do. ESM-2 blasted through roughly 600 million protein structure predictions in just two weeks using 2,000 graphic processing units (GPUs). Compared to previous attempts, the AI made the process up to 60 times faster. The authors put every structure into the ESM Metagenomic Atlas, which you can explore here.
To Dr. Alfonso Valencia at the Barcelona National Supercomputing Center (BCS), who was not involved in the work, the beauty of using large language systems is a “conceptual simplicity.” With further development, the AI can predict “the structure of non-natural proteins, expanding the known universe beyond what evolutionary processes have explored.”
Let’s Talk EvolutionESMFold follows a simple guideline: sequence predicts structure.
Let’s backtrack. Proteins are made from 20 amino acids—each one a “letter”—and strung up like spiky beads on a string. Our cells then shape them up into delicate features: some look like rumpled bed sheets, others like a swirly candy cane or loose ribbons. The proteins can then grab onto each other to form a multiplex—for example, a tunnel that crosses the brain cell membrane that controls its actions, and in turn controls how we think and remember.
Scientists have long known that amino acid letters help shape the final structure of a protein. Similar to letters or characters in a language, only certain ones when strung together make sense. In the case of proteins, these sequences make them functional.
“The biological properties of a protein constrain the mutations to its sequence that are selected through evolution,” the authors said.
Similar to how different letters in the alphabet converge to create words, sentences, and paragraphs without sounding like complete gibberish, the protein letters do the same. There is an “evolutionary dictionary” of sorts that helps string up amino acids into structures the body can comprehend.
“The logic of the succession of amino acids in known proteins is the result of an evolutionary process that has led them to have the specific structure with which they perform a particular function,” said Valencia.
Mr. AI, Make Me a ProteinLife’s relatively limited dictionary is great news for large language models.
These AI models scour readily available texts to learn and build up predictions of the next word. The end result, as seen in GPT-3 and ChatGPT, are strikingly natural conversations and fantastical artistic images.
Meta AI used the same concept, but rewrote the playbook for protein structure predictions. Rather than feeding the algorithm with texts, they gave the program sequences of known proteins.
The AI model—called a transformer protein language model—learned the general architecture of proteins using up to 15 billion “settings.” It saw roughly 65 million different protein sequences overall.
In their next step the team hid certain letters from the AI, prompting it to fill in the blanks. In what amounts to autocomplete, the program eventually learned how different amino acids connect to (or repel) each other. In the end, the AI formed an intuitive understanding of evolutionary protein sequences—and how they work together to make functional proteins.
Into the UnknownAs a proof of concept, the team tested ESMFold using two well-known test sets. One, CAMEO, involved nearly 200 structures; the other, CASP14, has 51 publicly-released protein shapes.
Overall, the AI “provides state-of-the-art structure prediction accuracy,” the team said, “matching AlphaFold2 performance on more than half the proteins.” It also reliably tackled large protein complexes—for example, the channels on neurons that control their actions.
The team then took their AI a step further, venturing into the world of metagenomics.
Metagenomes are what they sound like: a hodgepodge of DNA material. Normally these come from environmental sources such as the dirt under your feet, seawater, or even normally inhospitable thermal vents. Most of the microbes can’t be artificially grown in labs, yet some have superpowers such as resisting volcanic-level heat, making them a biological dark matter yet to be explored.
At the time the paper was published, the AI had predicted over 600 million of these proteins. The count is now up to over 700 million with the latest release. The predictions came fast and furious in roughly two weeks. In contrast, previous modeling attempts took up to 10 minutes for just a single protein.
Roughly a third of the protein predictions were of high confidence, with enough detail to zoom into the atomic-level scale. Because the protein predictions were based solely on their sequences, millions of “aliens” popped up—structures unlike anything in established databases or those previously tested.
“It’s interesting that more than 10 percent of the predictions are for proteins that bear no resemblance to other known proteins,” said Valencia. It might be due to the magic of language models, which are far more flexible at exploring—and potentially generating—previously unheard of sequences that make up functional proteins. “This is a new space for the design of proteins with new sequences and biochemical properties with applications in biotechnology and biomedicine,” he said.
As an example, ESMFold could potentially help suss out the consequences of single-letter changes in a protein. Called point mutations, these seemingly benign edits wreak havoc in the body, causing devastating metabolic syndromes, sickle cell anemia, and cancer. A lean, mean, and relatively simple AI brings results to the average biomedical research lab, while scaling up protein shape predictions thanks to the AI’s speed.
Biomedicine aside, another fascinating idea is that proteins may help train large language models in a way texts can’t. As Valencia explained, “On the one hand, protein sequences are more abundant than texts, have more defined sizes, and a higher degree of variability. On the other hand, proteins have a strong internal ‘meaning’—that is, a strong relationship between sequence and structure, a meaning or coherence that is much more diffuse in texts,” bridging the two fields into a virtuous feedback loop.
Image Credit: Meta AI
Solar power is going to play a major role in combating climate change, but it requires huge amounts of land. Floating solar panels on top of reservoirs could provide up to a third of the world’s electricity without taking up extra space, and also save trillions of gallons of water from evaporating.
So called “floating photovoltaic” systems have a lot going for them. The surface of reservoirs can’t be used for much else, so it’s comparatively cheap real estate, and it also frees up land for other important purposes. And because these bodies of water are designed to service major urban centers, they’re normally close to where the power will be needed, making electricity distribution simpler.
By shielding the water from the sun, floating solar panels can also significantly reduce evaporation, which can be a major concern in the hot dry climates where solar works best. And what evaporation does occur can actually help to cool the panels, which operate more efficiently at lower temperatures and therefore squeeze out extra power.
Just how promising the approach could be had remained unclear, as so far analyses had been limited to individual countries or regions. A new study in Nature Sustainability has now provided a comprehensive assessment of the global potential of floating solar power, finding that it could provide between a fifth and half of the world’s electricity needs while saving 26 trillion gallons of water from evaporating.
The new research was made possible by combining several databases mapping reservoirs around the world. This allowed the researchers to identify a total of 114,555 water bodies with a total area of 556,111 square kilometers (214,716 square miles).
They then used a model developed at the US Department of Energy’s Sandia National Laboratory that can simulate solar panel performance in different climatic conditions. Finally, they used regional hydrology simulations to predict how much the solar panels would reduce evaporation based on local climate data.
In their baseline study, the researchers assumed that solar panels would only cover 30 percent of a reservoir’s surface, or 30 square kilometers (11.6 square miles), depending on which is lower. This was done to take into account the practical difficulties of building larger arrays and also the potential ecological impact of completely covering up the body of water.
Given these limitations, the researchers calculated that the global generating potential for floating solar panels was a massive 9,434 terawatt-hours a year, which is roughly 40 percent of the 22,848 terawatt-hours the world consumes yearly, according to the International Energy Agency’s latest figures.
If the total coverage was limited to a much more reasonable 10 percent, the researchers found floating solar power could still generate as much as 4,356 terawatt-hours a year. And if the largest reservoirs were allowed to have up to 50 square kilometers (19 square miles) of panels then the total capacity rose to 11,012 terawatt-hours, almost half of global electricity needs.
The authors note that this capacity isn’t evenly distributed, and some countries stand to gain more than others. With more than 25,000 reservoirs, the US has the most to gain and could generate 1,911 terawatt-hours a year, almost half its total consumption. China, India, and Brazil could also source a significant amount of their power this way.
But most interestingly, the analysis showed that as many as 6,256 cities could theoretically meet all of their electricity demands with floating solar power. Most have a population below 50,000, but as many as 150 are cities with more than a million people.
It’s important to note that this study was simply assessing the potential of the idea. Floating solar panels have been around for some time, but they are more expensive to deploy than land-based panels, and there are significant concerns about what kind of impact blocking out sunlight could have on reservoir ecosystems.
But given the need to rapidly scale up renewable energy generation, and the scarcity of land for large solar installations, turning our reservoirs into power stations could prove to be a smart idea.
Image Credit: Juan00 / Pixabay
In 2020, artificial intelligence company OpenAI stunned the tech world with its GPT-3 machine learning algorithm. After ingesting a broad slice of the internet, GPT-3 could generate writing that was hard to distinguish from text authored by a person, do basic math, write code, and even whip up simple web pages.
OpenAI followed up GPT-3 with more specialized algorithms that could seed new products, like an AI called Codex to help developers write code and the wildly popular (and controversial) image-generator DALL-E 2. Then late last year, the company upgraded GPT-3 and dropped a viral chatbot called ChatGPT—by far, its biggest hit yet.
Now, a rush of competitors is battling it out in the nascent generative AI space, from new startups flush with cash to venerable tech giants like Google. Billions of dollars are flowing into the industry, including a $10-billion follow-up investment by Microsoft into OpenAI.
This week, after months of rather over-the-top speculation, OpenAI’s GPT-3 sequel, GPT-4, officially launched. In a blog post, interviews, and two reports (here and here), OpenAI said GPT-4 is better than GPT-3 in nearly every way.
More Than a Passing GradeGPT-4 is multimodal, which is a fancy way of saying it was trained on both images and text and can identify, describe, and riff on what’s in an image using natural language. OpenAI said the algorithm’s output is higher quality, more accurate, and less prone to bizarre or toxic outbursts than prior versions. It also outperformed the upgraded GPT-3 (called GPT 3.5) on a slew of standardized tests, placing among the top 10 percent of human test-takers on the bar licensing exam for lawyers and scoring either a 4 or a 5 on 13 out of 15 college-level advanced placement (AP) exams for high school students.
To show off its multimodal abilities—which have yet to be offered more widely as the company evaluates them for misuse—OpenAI president Greg Brockman sketched a schematic of a website on a pad of paper during a developer demo. He took a photo and asked GPT-4 to create a webpage from the image. In seconds, the algorithm generated and implemented code for a working website. In another example, described by The New York Times, the algorithm suggested meals based on an image of food in a refrigerator.
The company also outlined its work to reduce risk inherent in models like GPT-4. Notably, the raw algorithm was complete last August. OpenAI spent eight months working to improve the model and rein in its excesses.
Much of this work was accomplished by teams of experts poking and prodding the algorithm and giving feedback, which was then used to refine the model with reinforcement learning. The version launched this week is an improvement on the raw version from last August, but OpenAI admits it still exhibits known weaknesses of large language models, including algorithmic bias and an unreliable grasp of the facts.
By this account, GPT-4 is a big improvement technically and makes progress mitigating, but not solving, familiar risks. In contrast to prior releases, however, we’ll largely have to take OpenAI’s word for it. Citing an increasingly “competitive landscape and the safety implications of large-scale models like GPT-4,” the company opted to withhold specifics about how GPT-4 was made, including model size and architecture, computing resources used in training, what was included in its training dataset, and how it was trained.
Ilya Sutskever, chief technology officer and cofounder at OpenAI, told The Verge “it took pretty much all of OpenAI working together for a very long time to produce this thing” and lots of other companies “would like to do the same thing.” He went on to suggest that as the models grow more powerful, the potential for abuse and harm makes open-sourcing them a dangerous proposition. But this is hotly debated among experts in the field, and some pointed out the decision to withhold so much runs counter to OpenAI’s stated values when it was founded as a nonprofit. (OpenAI reorganized as a capped-profit company in 2019.)
The algorithm’s full capabilities and drawbacks may not become apparent until access widens further and more people test (and stress) it out. Before reining it in, Microsoft’s Bing chatbot caused an uproar as users pushed it into bizarre, unsettling exchanges.
Overall, the technology is quite impressive—like its predecessors—but also, despite the hype, more iterative than GPT-3. With the exception of its new image-analyzing skills, most abilities highlighted by OpenAI are improvements and refinements of older algorithms. Not even access to GPT-4 is novel. Microsoft revealed this week that it secretly used GPT-4 to power its Bing chatbot, which had recorded some 45 million chats as of March 8.
AI for the MassesWhile GPT-4 may not to be the step change some predicted, the scale of its deployment almost certainly will be.
GPT-3 was a stunning research algorithm that wowed tech geeks and made headlines; GPT-4 is a far more polished algorithm that’s about to be rolled out to millions of people in familiar settings like search bars, Word docs, and LinkedIn profiles.
In addition to its Bing chatbot, Microsoft announced plans to offer services powered by GPT-4 in LinkedIn Premium and Office 365. These will be limited rollouts at first, but as each iteration is refined in response to feedback, Microsoft could offer them to the hundreds of millions of people using their products. (Earlier this year, the free version of ChatGPT hit 100 million users faster than any app in history.)
It’s not only Microsoft layering generative AI into widely used software.
Google said this week it plans to weave generative algorithms into its own productivity software—like Gmail and Google Docs, Slides, and Sheets—and will offer developers API access to PaLM, a GPT-4 competitor, so they can build their own apps on top of it. Other models are coming too. Facebook recently gave researchers access to its open-source LLaMa model—it was later leaked online—while a Google-backed startup, Anthropic, and China’s tech giant Baidu rolled out their own chatbots, Claude and Ernie, this week.
As models like GPT-4 make their way into products, they can be updated behind the scenes at will. OpenAI and Microsoft continually tweaked ChatGPT and Bing as feedback rolled in. ChatGPT Plus users (a $20/month subscription) were granted access to GPT-4 at launch.
It’s easy to imagine GPT-5 and other future models slotting into the ecosystem being built now as simply, and invisibly, as a smartphone operating system that upgrades overnight.
Then What?If there’s anything we’ve learned in recent years, it’s that scale reveals all.
It’s hard to predict how new tech will succeed or fail until it makes contact with a broad slice of society. The next months may bring more examples of algorithms revealing new abilities and breaking or being broken, as their makers scramble to keep pace.
“Safety is not a binary thing; it is a process,” Sutskever told MIT Technology Review. “Things get complicated any time you reach a level of new capabilities. A lot of these capabilities are now quite well understood, but I’m sure that some will still be surprising.”
Longer term, when the novelty wears off, bigger questions may loom.
The industry is throwing spaghetti at the wall to see what sticks. But it’s not clear generative AI is useful—or appropriate—in every instance. Chatbots in search, for example, may not outperform older approaches until they’ve proven to be far more reliable than they are today. And the cost of running generative AI, particularly at scale, is daunting. Can companies keep expenses under control, and will users find products compelling enough to vindicate the cost?
Also, the fact that GPT-4 makes progress on but hasn’t solved the best-known weaknesses of these models should give us pause. Some prominent AI experts believe these shortcomings are inherent to the current deep learning approach and won’t be solved without fundamental breakthroughs.
Factual missteps and biased or toxic responses in a fraction of interactions are less impactful when numbers are small. But on a scale of hundreds of millions or more, even less than a percent equates to a big number.
“LLMs are best used when the errors and hallucinations are not high impact,” Matthew Lodge, the CEO of Diffblue, recently told IEEE Spectrum. Indeed, companies are appending disclaimers warning users not to rely on them too much—like keeping your hands on the steering wheel of that Tesla.
It’s clear the industry is eager to keep the experiment going though. And so, hands on the wheel (one hopes), millions of people may soon begin churning out presentation slides, emails, and websites in a jiffy, as the new crop of AI sidekicks arrives in force.
Image Credit: Luke Jones / Unsplash
ARTIFICIAL INTELLIGENCEYou Can Now Run a GPT-3-Level AI Model on Your Laptop, Phone, and Raspberry Pi
Benj Edwards | Ars Technica“On Friday, a software developer named Georgi Gerganov created a tool called “llama.cpp” that can run Meta’s new GPT-3-class AI large language model, LLaMA, locally on a Mac laptop. Soon thereafter, people worked out how to run LLaMA on Windows as well. Then someone showed it running on a Pixel 6 phone, and next came a Raspberry Pi (albeit running very slowly). If this keeps up, we may be looking at a pocket-sized ChatGPT competitor before we know it.”
BIOTECHA Gene Therapy Cure for Sickle Cell Is on the Horizon
Emily Mullin | Wired“[Evie] Junior…is one of dozens of sickle cell patients in the US and Europe who have received gene therapies in clinical trials—some led by universities, others by biotech companies. Two such therapies, one from Bluebird Bio and the other from CRISPR Therapeutics and Vertex Pharmaceuticals, are the closest to coming to market. The companies are now seeking regulatory approval in the US and Europe. If successful, more patients could soon benefit from these therapies, although access and affordability could limit who gets them.”
VIRTUAL REALITYThis Couple Just Got Married in the Taco Bell Metaverse
Tanya Basu | MIT Technology Review“The chapel at the company’s Taco Bell Cantina restaurant in Las Vegas has married 800 couples so far. There were copycat virtual weddings, too. ‘Taco Bell saw fans of the brand interact in the metaverse and decided to meet them quite literally where they were,’ a spokesperson said. That meant dancing hot sauce packets, a Taco Bell–themed dance floor, a turban for Mohnot, and the famous bell branding everywhere.”
ENERGYInside the Global Race to Turn Water Into Fuel
Max Bearak | The New York Times“A consortium of energy companies led by BP plans to cover an expanse of land eight times as large as New York City with as many as 1,743 wind turbines, each nearly as tall as the Empire State Building, along with 10 million or so solar panels and more than a thousand miles of access roads to connect them all. But none of the 26 gigawatts of energy the site expects to produce, equivalent to a third of what Australia’s grid currently requires, will go toward public use. Instead, it will be used to manufacture a novel kind of industrial fuel: green hydrogen.”
3D PRINTINGHas the 3D Printing Revolution Finally Arrived?
Tim Lewis | The Guardian“i‘What happened 10 years ago, when there was this massive hype, was there was so much nonsense being written: “You’ll print anything with these machines! It’ll take over the world!”‘ says Hague. ‘But it’s now becoming a really mature technology, it’s not an emerging technology really any more. It’s widely implemented by the likes of Rolls-Royce and General Electric, and we work with AstraZeneca, GSK, a whole bunch of different people. Printing things at home was never going to happen, but it’s developed into a multibillion-dollar industry.’i”
AUTOMATIONAI-Imager Midjourney v5 Stuns With Photorealistic Images—and 5-Fingered Hands
Benj Edwards | Ars Technica“Midjourney v5 is available now as an alpha test for customers who subscribe to the Midjourney service, which is available through Discord. ‘MJ v5 currently feels to me like finally getting glasses after ignoring bad eyesight for a little bit too long,’ said Julie Wieland, a graphic designer who often shares her Midjourney creations on Twitter. ‘Suddenly you see everything in 4k, it feels weirdly overwhelming but also amazing.’i”
GOVERNANCEAI-Generated Images From Text Can’t Be Copyrighted, US Government Rules
Kris Holt | Engadget“That’s according to the US Copyright Office (USCO), which has equated such prompts to a buyer giving directions to a commissioned artist. ‘They identify what the prompter wishes to have depicted, but the machine determines how those instructions are implemented in its output,’ the USCO wrote in new guidance it published to the Federal Register. ‘When an AI technology receives solely a prompt from a human and produces complex written, visual, or musical works in response, the “traditional elements of authorship” are determined and executed by the technology—not the human user,’ the office stated.”
ARTIFICIAL INTELLIGENCEGPT-4 Has the Memory of a Goldfish
Jacob Stern | The Atlantic“By this point, the many defects of AI-based language models have been analyzed to death—their incorrigible dishonesty, their capacity for bias and bigotry, their lack of common sense. …But large language models have another shortcoming that has so far gotten relatively little attention: their shoddy recall. These multibillion-dollar programs, which require several city blocks’ worth of energy to run, may now be able to code websites, plan vacations, and draft company-wide emails in the style of William Faulkner. But they have the memory of a goldfish.”
ETHICSMicrosoft Lays Off an Ethical AI Team as It Doubles Down on OpenAI
Rebecca Bellan | TechCrunch“The move calls into question Microsoft’s commitment to ensuring its product design and AI principles are closely intertwined at a time when the company is making its controversial AI tools available to the mainstream. Microsoft still maintains its Office of Responsible AI (ORA), which sets rules for responsible AI through governance and public policy work. But employees told Platformer that the ethics and society team was responsible for ensuring Microsoft’s responsible AI principles are actually reflected in the design of products that ship.”
GENE EDITINGIt’s Official: No More Crispr Babies—for Now
Grace Browne | Wired“After several days of experts chewing on the scientific, ethical, and governance issues associated with human genome editing, the [Third International Summit on Human Genome Editing’s] organizing committee put out its closing statement. Heritable human genome editing—editing embryos that are then implanted to establish a pregnancy, which can pass on their edited DNA—’remains unacceptable at this time,’ the committee concluded. ‘Public discussions and policy debates continue and are important for resolving whether this technology should be used.’i”
Image Credit: Kenan Alboshi / Unsplash
Without water, life on Earth could not exist as it does today. Understanding the history of water in the universe is critical to understanding how planets like Earth come to be.
Astronomers typically refer to the journey water takes from its formation as individual molecules in space to its resting place on the surfaces of planets as “the water trail.” The trail starts in the interstellar medium with hydrogen and oxygen gas and ends with oceans and ice caps on planets, with icy moons orbiting gas giants and icy comets and asteroids that orbit stars. The beginnings and ends of this trail are easy to see, but the middle has remained a mystery.
I am an astronomer who studies the formation of stars and planets using observations from radio and infrared telescopes. In a new paper, my colleagues and I describe the first measurements ever made of this previously hidden middle part of the water trail and what these findings mean for the water found on planets like Earth.
Star and planet formation is an intertwined process that starts with a cloud of molecules in space. Image Credit: Bill Saxton, NRAO/AUI/NSF, CC BYHow Planets Are FormedThe formation of stars and planets is intertwined. The so-called “emptiness of space”—or the interstellar medium—in fact contains large amounts of gaseous hydrogen, smaller amounts of other gases, and grains of dust. Due to gravity, some pockets of the interstellar medium will become more dense as particles attract each other and form clouds. As the density of these clouds increases, atoms begin to collide more frequently and form larger molecules, including water that forms on dust grains and coats the dust in ice.
Stars begin to form when parts of the collapsing cloud reach a certain density and heat up enough to start fusing hydrogen atoms together. Since only a small fraction of the gas initially collapses into the newborn protostar, the rest of the gas and dust forms a flattened disk of material circling around the spinning, newborn star. Astronomers call this a proto-planetary disk.
As icy dust particles collide with each other inside a proto-planetary disk, they begin to clump together. The process continues and eventually forms the familiar objects of space like asteroids, comets, rocky planets like Earth and gas giants like Jupiter or Saturn.
Two Theories for the Source of WaterThere are two potential pathways that water in our solar system could have taken. The first, called chemical inheritance, is when the water molecules originally formed in the interstellar medium are delivered to proto-planetary disks and all the bodies they create without going through any changes.
The second theory is called chemical reset. In this process, the heat from the formation of the proto-planetary disk and newborn star breaks apart water molecules, which then reform once the proto-planetary disk cools.
To test these theories, astronomers like me look at the ratio between normal water and a special kind of water called semi-heavy water. Water is normally made of two hydrogen atoms and one oxygen atom. Semi-heavy water is made of one oxygen atom, one hydrogen atom and one atom of deuterium—a heavier isotope of hydrogen with an extra neutron in its nucleus.
The ratio of semi-heavy to normal water is a guiding light on the water trail—measuring the ratio can tell astronomers a lot about the source of water. Chemical models and experiments have shown that about 1,000 times more semi-heavy water will be produced in the cold interstellar medium than in the conditions of a protoplanetary disk.
This difference means that by measuring the ratio of semi-heavy to normal water in a place, astronomers can tell whether that water went through the chemical inheritance or chemical reset pathway.
V883 Orionis is a young star system with a rare star at its center that makes measuring water in the proto-planetary cloud, shown in the cutaway, possible. Image Credit: ALMA (ESO/NAOJ/NRAO), B. Saxton (NRAO/AUI/NSF), CC BYMeasuring Water During the Formation of a PlanetComets have a ratio of semi-heavy to normal water almost perfectly in line with chemical inheritance, meaning the water hasn’t undergone a major chemical change since it was first created in space. Earth’s ratio sits somewhere in between the inheritance and reset ratio, making it unclear where the water came from.
To truly determine where the water on planets comes from, astronomers needed to find a goldilocks proto-planetary disk—one that is just the right temperature and size to allow observations of water. Doing so has proved to be incredibly difficult. It is possible to detect semi-heavy and normal water when water is a gas; unfortunately for astronomers, the vast majority of proto-plantary disks are very cold and contain mostly ice, and it is nearly impossible to measure water ratios from ice at interstellar distances.
A breakthrough came in 2016, when my colleagues and I were studying proto-planetary disks around a rare type of young star called FU Orionis stars. Most young stars consume matter from the proto-planetary disks around them. FU Orionis stars are unique because they consume matter about 100 times faster than typical young stars and, as a result, emit hundreds of times more energy. Due to this higher energy output, the proto-planetary disks around FU Orionis stars are heated to much higher temperatures, turning ice into water vapor out to large distances from the star.
Using the Atacama Large Millimeter/submillimeter Array, a powerful radio telescope in northern Chile, we discovered a large, warm proto-planetary disk around the sunlike young star V883 Ori, about 1,300 light years from Earth in the constellation Orion.
V883 Ori emits 200 times more energy than the sun, and my colleagues and I recognized that it was an ideal candidate to observe the semi-heavy to normal water ratio.
The proto-planetary disk around V883 Ori contains gaseous water, shown in the orange layer, allowing astronomers to measure the ratio of semi-heavy to normal water. Image Credit: ALMA (ESO/NAOJ/NRAO), J. Tobin, B. Saxton (NRAO/AUI/NSF), CC BYCompleting the Water TrailIn 2021, the Atacama Large Millimeter/submillimeter Array took measurements of V883 Ori for six hours. The data revealed a strong signature of semi-heavy and normal water coming from V883 Ori’s proto-planetary disk. We measured the ratio of semi-heavy to normal water and found that the ratio was very similar to ratios found in comets as well as the ratios found in younger protostar systems.
These results fill in the gap of the water trail forging a direct link between water in the interstellar medium, protostars, proto-planetary disks, and planets like Earth through the process of inheritance, not chemical reset.
The new results show definitively that a substantial portion of the water on Earth most likely formed billions of years ago, before the sun had even ignited. Confirming this missing piece of water’s path through the universe offers clues to origins of water on Earth. Scientists have previously suggested that most water on Earth came from comets impacting the planet. The fact that Earth has less semi-heavy water than comets and V883 Ori, but more than chemical reset theory would produce, means that water on Earth likely came from more than one source.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: A. Angelich (NRAO/AUI/NSF)/ALMA (ESO/NAOJ/NRAO), CC BY
Of all the advanced technologies currently under development, one of the most fascinating and frightening is brain-computer interfaces. They’re fascinating because we still have so much to learn about the human brain, yet scientists are already able to tap into certain parts of it. And they’re frightening because of the sinister possibilities that come with being able to influence, read, or hijack peoples’ thoughts.
But the worst-case scenarios that have been played out in science fiction are just one side of the coin, and brain-computer interfaces could also be a tremendous boon to humanity—if we create, manage, and regulate them correctly. In a panel discussion at South by Southwest this week, four experts in the neuroscience and computing field discussed how to do this.
Panelists included Ben Hersh, a staff interaction designer at Google; Anna Wexler, an assistant professor of medical ethics and health policy at the University of Pennsylvania; Afshin Mehin, the founder of a creative studio that helps companies give form to the future called Card79; and Jacob Robinson, an associate professor in electrical and computer engineering at Rice University and co-founder of Motif Neurotech, a company creating minimally invasive electronic therapies for mental health.
“This is a field that has a lot of potential for good, and there’s a lot that we don’t know yet,” Hersh said. “It’s also an area that has a lot of expectations that we’ve absorbed from science fiction.” In his opinion, “mind control for good” is not only a possibility, it’s an imperative.
The Mysterious BrainOf all the organs in our bodies, the brain is by far the most complex—and the one we know the least about. “Two people can perceive the same stimuli and have a very different subjective experience, and there are no real rules to help us understand what translates your experience of the world into your subjective reality,” Robinson said.
But, he added, if we zoom in on the fundamental aspect of what’s happening in our brains, it is governed by physical processes. Could it be possible to control aspects of the brain and our subjective experiences with the level of precision we have in fields like physics and engineering?
“Part of why we’ve struggled with treating mental health conditions is that we don’t have a fundamental understanding of what leads to these disorders,” Robinson said. “But we know that they are network-level problems…we’re beginning to interface with the networks that are underlying these types of conditions, and help to restore them.”
BCIs TodayElon Musk’s Neuralink has brought BCIs into the public eye more than they’re ever been before, but there’s been a consumer neurotechnoloy market since the mid-2000s. Electroencephalography (EEG) uses electrodes placed on the head to record basic measures of brain wave activity. Consumer brain stimulation devices are marketed for cognitive enhancement, such as improving focus, memory, or attention.
More advanced neural interfaces are being used as assistive technology for people with conditions like ALS or paralysis, helping them communicate or move in ways they otherwise wouldn’t be able to: translating thoughts into text, movements, speech, or written sentences. One brain implant succeeded in alleviating treatment-resistant depression via small, targeted doses of electrical stimulation.
“Some of the things that are coming up are actually kind of extraordinary,” Hersh said. “People are working on therapies where electronics are implanted in the brain and can help deal with illnesses beyond the reach of modern medicine.”
Dystopian PossibilitiesThis sounds pretty great, so what could go wrong? Well, unfortunately, lots. The idea of someone tapping into your brain and being able to control it is terrifying, and we’re not just talking dramatic scenarios like The Matrix; what if you had a brain implant for a medical purpose, but someone was able to subtly influence your choices around products or services you purchase? What if a record of your emotional state was released to someone you didn’t want to have it, or your private thoughts were made public? (I know what you’re thinking: ‘Wait—isn’t that what Twitter’s for?’)
Even tools with a positive intent could have unwanted impacts. Mehin’s company created a series of video vignettes imagining what BCI tech could do in day-to-day life. “The scenarios we imagined were spread between horrifying—imagine having an AI chatbot living inside your head—to actually useful, like being able to share how you’re feeling with a friend so they can help you sort through a difficult time.”
He shared that upon showing the videos at a design conference where there were students in the audience, a teacher spoke up and said, “This is horrible, kids will never be able to communicate with each other.” But then a student got up and said “We already can’t communicate with each other, this would actually be really useful.”
Would you want to live in a world where we need brain implants to communicate our emotions to one another? Where you wouldn’t sit and have coffee with a friend to talk about your career stress or marital strife, you’d just let them tap straight into your thoughts?
No thanks.
BCI UtopiaA brain-computer interface utopia sounds like an oxymoron; the real utopia would be one where we’re healthy, productive, and happy without the need for invasive technology tapping into the networks that dictate our every thought, feeling, and action.
But reality is that the state of mental health in the US is far from ideal. Millions of people suffer from conditions like PTSD, ADHD, anxiety, and depression, and pharmaceuticals haven’t been able to come up with a great cure for any of these. Pills like Adderall, Xanax, or Prozac come with unwanted side effects, and for some people they don’t work at all.
“One in ten people in the US suffer from a mental health disorder that’s not effectively treated by their drugs,” said Robinson. “Our hope is that BCIs could offer a 20-minute outpatient procedure that would provide therapeutic benefit for conditions like treatment-resistant depression, PTSD, or ADHD, and could last the rest of your life.”
He envisions a future where everyone has the ability to communicate rapidly and seamlessly, regardless of any disability, and where BCIs actually let us get back some of the humanity that has been stolen by social media and smartphones. “Maybe BCIs could help us rebalance the neural circuits we need to have control over our focus and our mood,” he said. “We would feel better, do better, and everyone could communicate.”
In the near term, the technology will continue to advance most in medical applications. Robinson believes we should keep moving BCIs forward despite the risks, because they can help people.
“There’s a risk that people see that vision of the dystopian future and decide to stop building these things because something bad could happen,” he said. “My hope is that we don’t do that. We should figure out how to go forward responsibly, because there’s a moral obligation to the people who need these things.”
Image Credit: Gerd Altmann from Pixabay
From 1850 to 2019, human activity released 2.4 trillion tons of CO2 into the atmosphere. In 2022 alone, we released 37 more tons. While renewable energy is making a difference, it’s small: last year it offset a mere 230 million tons of emissions—less than one percent of the global total.
Energy demand is expected to triple by 2050. Amid calls for emissions reductions and net-zero targets, we need a reality check: how are we going to reverse climate change if energy is in everything we do, and energy itself contributes to the problem?
We need solutions that will help us pull trillions of tons of carbon from the air without adding more in the process—a tool far more powerful than solar panels or wind turbines. This tool already exists, and it’s nuclear power.
In a talk at South By Southwest this week, Bret Kugelmass, founder and CEO of Last Energy, explained how nuclear power has been misunderstood and devalued for decades, and the price we’ve paid as a result. “Infinitely abundant, carbon-free, always on, and incredibly energy-dense, nuclear energy could meet and exceed our energy needs,” he said.
Instead, this powerful technology has stagnated for decades, leaving us scrambling for other forms of energy that won’t keep pumping CO2 into the atmosphere. Kugelmass left a career in Silicon Valley with the sole purpose of finding a keystone technology to combat climate change. He visited 15 countries and all kinds of facilities to learn about nuclear power and compare it to other forms of energy. His conclusion was that if it’s done right, nuclear can enable continued growth—and a cleaner planet—in a way that no other power source can.
How Did We Get Here?So why did a power source with so much potential stagnate? In 1963, then-President John F. Kennedy said nuclear power would account for half of all US energy production by end of that decade. His administration put together a perspective for rapid development of nuclear power production, and he had the Atomic Energy Commission conduct a study on the role civilian nuclear power could play in the US economy.
According to Kugelmass, the effort stalled in its tracks not because of public perception or safety fears, but due to economic malfeasance. Rather than focusing on standardization, “We pursued ever-larger, ever more complex construction projects…from 1968 to 1970, we saw a 10-fold increase in the cost to build gigawatt-scale plants,” he said. Most of the cost of nuclear energy, he added, is in the interest accrued during the construction process. “It accounts for 60 percent of the delivered cost of energy,” he said.
The result, unsurprisingly, was that nuclear simply became too expensive to compete with other power sources. The US is now close to completing its first new nuclear project in decades—and at 10 years late and $20 billion over budget, it’s still not done.
If we had built out nuclear in a viable way starting in the 1960s, we’d live in a very different world today: less pollution, less panic about carbon emissions, more energy security, cheaper end prices for consumers. Is it too late to turn things around? “There is nothing broken with the nuclear technology we have today,” Kugelmass said. “What’s broken is the business model, and the delivery model. What nuclear needs to scale isn’t novel: productize, modularize, and mass-manufacture.”
Bringing Nuclear BackKugelmass founded a non-profit research organization called the Energy Impact Center (EIC), which in 2020 launched the OPEN100 project to provide open-source blueprints for the design, construction, and financing of a 100-megawatt nuclear reactor. EIC’s for-profit spinoff is Last Energy, which aims to connect private investors with opportunities to develop new nuclear projects around the world.
Rather than experimenting with newer technology, Last Energy’s sticking with tried-and-true pressurized water reactors (the kind used over the last several decades), but bringing their costs down by making the technology modular and standardized. They’re taking a play from the oil and gas industry, which can build entire power plants in a factory then deploy them to their final location.
“There’s a whole avenue of innovation related to constructability, rather than your underlying technology,” Kugelmass said. “If you deviate too much from the standard supply chain you’re going to see hidden costs everywhere.” He estimated, for example, that building a pump to move the salt for molten salt reactors, which use molten salt as a coolant instead of pressurized water, requires a billion dollars in research and development costs.
Building standardized small modular reactors, though, can be done for less than $1,000 per kilowatt. Making nuclear power affordable would mean it could be used for energy-intensive industrial applications that will become increasingly necessary in coming years, like water desalination and carbon removal.
Time for a Revival?Energy underlies everything we do, and it’s essential for modern societies to grow and thrive. It enables human well-being, entrepreneurship, geopolitical independence, security, and opportunity. Given our current geopolitical situation and the unsustainable energy costs in Europe, could now be the time for a nuclear revival?
Kugelmass is hopeful. “Every 10 to 15 years the industry thinks it’s gong to have a renaissance, but then it falls flat,” he said. “Now global macro issues have granted nuclear the opportunity to have another shot.”
In fact, Last Energy is looking to launch in Europe, where the need for affordable energy is dire. The company has signed deals in Romania, Poland, and the UK, and its first set of reactors is slated to come online in the next two years. Kugelmass noted that negotiating with utilities and governments in these countries is far more straightforward than in the US. “Maybe we’ll come to US someday, but we could be selling hundreds of gigawatts in Europe before that happens,” he said.
There may be hope for the US yet: in 2020 the Department of Energy launched its Advanced Reactor Demonstration Program, investing $230 million in research and development for small modular reactors.
Kugelmass is focused on making a solid product, no matter where it ends up being used. “We are an American company and we build the reactors here in Texas,” he said. “What previously took decades to build and cost billions is now a scalable product that can be pre-fabricated and deployed in under two years.”
Image Credit: Albrecht Fietz from Pixabay
Breakthroughs don’t often happen in neuroscience, but we just had one. In a tour-de-force, an international team released the full brain connectivity map of the young fruit fly, described in a paper published last week in Science. Containing 3,016 neurons and 548,000 synapses, the map—called a connectome—is the most complex whole brain-wiring diagram to date.
“It’s a ‘wow,’” said Dr. Shinya Yamamoto at Baylor College of Medicine, who was not involved in the work.
Why care about a fruit fly? Far from uninvited guests at the dinner table, Drosophila melanogaster is a neuroscience darling. Although its brain is smaller than a poppy seed—a far cry from the 100 billion neurons that power human brains—the fly’s neural system shares similar principles to those that underlie our own brains.
This makes them excellent models to hone in on ideas of how our neural circuits wire to encode memories, make difficult decisions, or navigate social situations like flirting with a potential partner or hanging with a swarm of new friends.
To lead author Dr. Marta Zlatic at the University of Cambridge, MRC Laboratory of Molecular Biology and Janelia Research Campus,“All brains are similar—they are all networks of interconnected neurons—and all brains of all species have to perform many complex behaviors: they all need to process sensory information, learn, select actions, navigate their environments, choose food, escape from predators, etc.”
With the new connectome map, “we now have a reference brain,” she said.
A Behemoth AtlasConnectomes are precious resources. Popularized by Sebastian Seung, the maps draw out neural connections within and across brain regions. Similar to tracing computer wires to reverse-engineer how different chips and processors fit together, the connectome is a valuable resource to crack the brain’s “neural code”—that is, the algorithms underlying its computations.
In other words, the connectome is essential to understanding the brain’s functions. It’s why similar work is underway in mice and humans, though at a much smaller scale or with far less detail.
Until now, scientists have only mapped three full-brain connectomes, all in worms—including the first animal to gain the honor, the nematode C. elegans. With just over 300 neurons, the project took over a decade, with an update released for both sexes in 2019.
Drosophila represents a far larger challenge with roughly ten times the number of neurons as C. elegans. But it’s also an ideal next candidate. For one, scientists have already sequenced its entire genome, making it possible to match genetic information to the fly’s neural wiring. This could especially come in handy for, say, deciphering how genes contributing to Alzheimer’s disease alters neural circuits. For another, fruit fly larvae have transparent bodies, making them far easier to image under a microscope.
Not all brain-wiring maps are created equal. Here, the team went for the highest resolution: mapping the whole brain at the synapse level. Synapses are junctions between neurons where they connect: picture two mushroom-shaped structures hovering near each other with a gap. Although neurons are often touted as the basic component of computing, synapses are where the magic happens—their connectivity helps functionally wire up neural circuits.
Neuron connectivity in the brain. Each dot represents a neuron, and those with more similar connectivity are closer. The lines show how neurons connect. Image Credit: Benjamin PedigoSlice and Dice and…Robots?To map out synapses, the team turned to the big guns of microscopy: the electron microscope. Compared to microscopes in high-school biology, this hardware can capture images at the nanoscale—roughly a tenth the width of a human hair.
The whole process sounds a bit like a wild dinner recipe. The team first soaked a single six-hour-old larvae brain inside a solution packed with heavy metals, which marinated into the neurons’ membranes and proteins inside synapses. The brains are then painstakingly sliced into ultra-thin sections with a diamond blade—imagine a deli-meat slicer—and put under a microscope.
The resulting images—all 21 million of them—were stitched together using software. The whole process took over a year and a half, with many hours spent on manually checking the reconstructed neurons and synapses.
The final brain map didn’t just contain the location of neurons and their synapses—it also highlighted wiring quirks that could support highly efficient neural computations.
Winding RoadsThe beauty of the new map is that it provides bird’s-eye information on brain connectivity, supercharged with the power of zoom-and-enhance.
“The most challenging aspect of this work was understanding and interpreting what we saw,” said Zlatic.
In one analysis, the team found that neurons can be grouped into 93 different types based on their connectivity, even if they share the same physical structure. It’s a drastic departure from the most common way of categorizing neurons. Rather than clustering them based on appearance or function, it may be more useful to focus on their connectivity “social network” instead.
Digging down to synapses, the team ran into another surprise. Let me explain: neurons have two main branches. One is the larger input cable—the axon—and the other is a tree-shaped output—the dendrite. Neurons usually “wire up” when synapses connect those two cables.
More recent studies, however, show that synapses on axons can connect with other synapses on axons; the same goes for dendrites. Analyzing the reconstructed brain, the team found evidence of these non-traditional connections.
“Now we need to reconsider them: we probably need to think about creating a new computational model of the nervous system,” said Dr. Chung-Chuang Lo at the National Tsing Hua University in Taiwan.
On a broader scale, the map showed that neurons are eager to chat with others a half-world away. Almost 93 percent of neurons connected with a partner neuron in the other brain hemisphere, suggesting that long-range connections are incredibly common. Even more surprising was a peculiar population that didn’t reach out: dubbed Kenyon cells, these neurons mostly populate the fly’s learning and memory center. Why this happens is still unclear, but it illustrates the brain map’s ability to generate new insights and hypotheses.
Although the neurons and synapses are wired in a nicely compact “nested” multilayered structure, the connectome showed that some loved developed connections that jumped through layers—a shortcut that hooks up otherwise separate circuits.
Even more fascinating was how much the brain “talks” to itself. Nearly 41 percent of neurons received recurrent input—that is, feedback from other parts of the brain. Each region had its own feedback program. For example, information generally flows from sensory areas of the brain to motor regions, although the reverse also happens and creates a feedback loop.
But perhaps the most socially adept neurons are those that pump out dopamine. Well known for encoding reward and driving learning, these neurons also had some of the most complex recurrent wirings compared to other types.
From shortcuts to recurrent wirings, these biological hardware structures could increase the brain’s computational capacity and compensate for the limited number of neurons and their biological restraints.
“None of us expected this at all,” said study author Dr. Michael Winding.
From Fly to AIThe study isn’t the first to map the Drosophila brain. Previously, a team led by Dr. Davi Bock at the Janella Research Campus targeted a small nub of the adult fruit fly brain responsible for learning and remembering smells with synapse-level detail. Zlatic’s team has also tracked a sensory circuit in the fruit fly larvae for making decisions by mapping only 138 neurons.
The full-brain connectome is a game-changer. For one, scientists now have a sophisticated reference brain to test out theories for neural computation. For another, the connectome map and its inferred computation resembles state-of-the-art machine learning.
“That’s really quite nice because we know that recurrent neural networks are pretty powerful in artificial intelligence,” said Zlatic. “By comparing this biological system, we can potentially also inspire better artificial networks.”
Image Credit: Michael Winding
Many stories about the future are formed by imagining worst-case scenarios, then extracting lessons from them about what we should try to avoid. Much of the best science fiction takes this angle, and it makes for good reading (or watching or listening). But there can be as much value—if not more—in the opposite approach; what if we imagine a world where our efforts to fix today’s biggest problems have paid off, and both humanity and the planet are flourishing? Then we can take steps towards making that vision a reality.
In a discussion at South by Southwest this week titled Life on a Reforested Planet, the panelists took such a future retrospective point of view. What, they asked, will the world look like decades from now if we succeed in cleaning up the environment, bringing carbon emissions down, and restoring degraded forests? What opportunities are there around these scenarios? And how will we get there?
The discussion was led by Yee Lee, the VP of growth at a company called Terraformation whose mission is to accelerate natural carbon capture by resolving bottlenecks to forest restoration. Lee spoke with Jad Daley, president and CEO of American Forests, the oldest national nonprofit conservation organization in the US; Clara Rowe, CEO of a global network of restoration and conservation sites called Restor; and Josh Parrish, VP of carbon origination at Pachama, which uses remote sensing and AI to protect and restore natural carbon sinks.
There are about three trillion trees on Earth today. That’s more trees than there are stars in the Milky Way, but it’s only about half as many as there were at the dawn of human civilization. Scientists have estimated we can bring back one trillion trees on degraded lands we aren’t using for agriculture. If those trillion trees were to be planted all together, they’d cover the entire continental US—but every continent except Antarctica has reforestable lands. Furthermore, if we restore one trillion trees, they’d be able to sequester around 30 percent of the carbon we’ve put into the atmosphere since the industrial revolution.
Planting a trillion trees is obviously no small task. It requires the right kind of seeds, well-trained forestry professionals, collaboration with local and national governments, and multiple levels of in-depth research and planning—not to mention a lot of time, space, and hard work. In outlining what the world will look like if we make it happen, the panelists highlighted current challenges that would be resolved as well as opportunities we’d encounter along the way. Here are a few of the changes we’ll see in our lives and the environment if we can make this vision a reality.
Nature EquityWe think of nature and trees as having blanket benefits across society: they’re beautiful, they clean the air, they provide shade and habitats for wildlife. But the unfortunate reality we’re living in has an unequal distribution of access to nature across populations. “Tree equity isn’t about trees, it’s about people,” Daley said. “In neighborhoods with a lot of trees, people are healthier—including mental health benefits—and there’s less crime. People relate to each other differently.” This isn’t because trees cause prosperity, but because prosperous communities are more likely to invest in landscaping and tree cover, and to have the funds to do so.
The opposite side of the coin shows the drawbacks that non-green areas experience, all of which are only slated to worsen in coming years. “Today in America, extreme heat kills more than 12,000 people per year,” Daley said. Research projects that number could rise to 110,000 people per year by the end of this century, with the hardest-hit being those who don’t have air conditioning, don’t have good healthcare—and don’t have trees in their neighborhoods.
“Trees have incredible cooling power and every neighborhood needs that, but especially places where people are already most at risk,” Daley said. He pointed out that tree distribution maps are often also maps of income and race, with the lowest-income neighborhoods having 40 percent less tree coverage than the wealthiest neighborhoods.
In a future where we’ve succeeded in planting a trillion trees, cities will have equitable tree cover. There are already steps in this direction: the US Congress invested $1.5 billion in tree cover for cities as part of the Inflation Reduction Act.
Incentives Align With the Needs of the Natural WorldCapitalism likely won’t be replaced by another economic system anytime soon, but non-financial incentives will take on a larger role in influencing business and consumer decisions, and regulators will likely step in and change financial incentives too. Carbon credits are one early example of this (though there’s a lot of debate about their effectiveness), as are the subsidies around electric vehicles and solar and wind energy.
Could we implement similar subsidies or other means of incentive around reforestation? Some countries have already done so. Costa Rica, Rowe said, has been paying farmers to conserve and restore forests on their land for decades, making Costa Rica the first tropical country to reverse deforestation. “People are getting paid to do something that’s good for the Earth, and it has changed the relationship that a lot of the country has to nature,” she said. “So then it’s not just about the money; because we’ve created an economy that allows us to benefit from nature, we can love nature in a different way.”
A Shift in Consumerist CultureManufacturing—of everything from cars to cell phones to clothing—not only uses energy and creates emissions, it creates a lot of waste. When the newest iPhone comes out, millions of people tuck their old phone in the back of drawer and go out and buy the new one, even though the old one still worked perfectly. We give old clothes to Goodwill (or throw them away) and buy new ones long before the old clothes are unwearable or out of style. We trade in our 10-year-old cars for the new model, even though the car has 10 more years of drivability in it.
Having the newest things is a status symbol and a way to introduce some occasional novelty into our lives and routines. But what if we flipped that on its head, reversing what’s “cool” and high-status to align with the needs of the environment? What if we bragged about having an old car or phone or bike, and thereby not having contributed to the continuous manufacture and disposal of still-useful goods?
A shift to conscious consumerism has already begun, with people paying attention to the business practices of companies they buy from and seeking out brands that are more Earth-friendly. But this movement will need to grow far beyond its current state and include a much broader chunk of the population to really make a difference.
Rowe believes that in the not-too-distant future, products will have labeling with information about their supply chain and their impact on the local environment. “There are ways to weave forests into the daily fabric of our lives, and one of those is understanding what we consume,” she said. “Think about the cereal you had for breakfast. In 2050 the label will have information about the species of trees restored in the place where the wheat is grown, and the tons of carbon that were sequestered by the regenerative agriculture in this area.”
She envisions us gaining a completely new perspective on what we’re a part of and how we’re having impact. “We’re touching nature in every part of our lives, but we aren’t empowered to know it,” she added. “We don’t have the tools to take the action that we really want to take. In 2050, when we’ve reforested our planet, the way we have impact will be visible.”
Job Growth in Forestry and Related IndustriesPlanting a trillion trees—and making sure they’re healthy and growing—will require a massive mobilization of funds and people, and will spur creation of all sorts of jobs. Not to mention, reforestation will enable new industries to sprout where before there could be none. One example Lee gave was if you restore a mangrove, a shrimping industry can then be built there. “When we’re fostering a new forestry team, the lightbulb moment isn’t just about forests and trees,” he said. “There’s a whole economic livelihood that’s created. The blocker is often, how do we skill new communities and train them to have an entrepreneurial mindset?”
Parrish envisions the creation of “superhighways for nature,” an undertaking that would entail significant job creation in itself. “As the climate changes, as we get warmer, nature needs the ability to adapt and migrate and move around,” he said. “We need to create a network of connections with forests that provide for that and have a diverse ecological framework.” This would apply not only to primary forests, he said, but to suburban and even urban green spaces too.
Daley mentioned that his organization is seeing job creation on the front end of the reforestation pipeline, with one example being people who are employed to collect the seeds that’ll be used to plant trees. “We partner with the state of California and an organization called the Cone Core,” he said. “People collect cones to collect seeds they’ll use to reforest the burned acres in California.”
A Reforested WorldWill these visions become reality? We’re a long way from it right now, but planting a trillion trees isn’t impossible. In Daley’s opinion, the two variables that will most help the cause are innovation and mobilization, and both awareness and buy-in around reforestation are steadily growing. As more people feel empowered to take part, they’ll also find new ways to make a difference. “Hope comes from agency,” Daley said. To engage with a problem, “you need to feel like you can do something about it.”
Image Credit: Chris Lawton on Unsplash
Empowered by artificial intelligence technologies, computers today can engage in convincing conversations with people, compose songs, paint paintings, play chess and go, and diagnose diseases, to name just a few examples of their technological prowess.
These successes could be taken to indicate that computation has no limits. To see if that’s the case, it’s important to understand what makes a computer powerful.
There are two aspects to a computer’s power: the number of operations its hardware can execute per second and the efficiency of the algorithms it runs. The hardware speed is limited by the laws of physics. Algorithms—basically sets of instructions—are written by humans and translated into a sequence of operations that computer hardware can execute. Even if a computer’s speed could reach the physical limit, computational hurdles remain due to the limits of algorithms.
These hurdles include problems that are impossible for computers to solve and problems that are theoretically solvable but in practice are beyond the capabilities of even the most powerful versions of today’s computers imaginable. Mathematicians and computer scientists attempt to determine whether a problem is solvable by trying them out on an imaginary machine.
An Imaginary Computing MachineThe modern notion of an algorithm, known as a Turing machine, was formulated in 1936 by British mathematician Alan Turing. It’s an imaginary device that imitates how arithmetic calculations are carried out with a pencil on paper. The Turing machine is the template all computers today are based on.
To accommodate computations that would need more paper if done manually, the supply of imaginary paper in a Turing machine is assumed to be unlimited. This is equivalent to an imaginary limitless ribbon, or “tape,” of squares, each of which is either blank or contains one symbol.
The machine is controlled by a finite set of rules and starts on an initial sequence of symbols on the tape. The operations the machine can carry out are moving to a neighboring square, erasing a symbol, and writing a symbol on a blank square. The machine computes by carrying out a sequence of these operations. When the machine finishes, or “halts,” the symbols remaining on the tape are the output or result.
Computing is often about decisions with yes or no answers. By analogy, a medical test (type of problem) checks if a patient’s specimen (an instance of the problem) has a certain disease indicator (yes or no answer). The instance, represented in a Turing machine in digital form, is the initial sequence of symbols.
A problem is considered “solvable” if a Turing machine can be designed that halts for every instance whether positive or negative and correctly determines which answer the instance yields.
Not Every Problem Can Be SolvedMany problems are solvable using a Turing machine and therefore can be solved on a computer, while many others are not. For example, the domino problem, a variation of the tiling problem formulated by Chinese American mathematician Hao Wang in 1961, is not solvable.
The task is to use a set of dominoes to cover an entire grid and, following the rules of most dominoes games, matching the number of pips on the ends of abutting dominoes. It turns out that there is no algorithm that can start with a set of dominoes and determine whether or not the set will completely cover the grid.
Keeping It ReasonableA number of solvable problems can be solved by algorithms that halt in a reasonable amount of time. These “polynomial-time algorithms” are efficient algorithms, meaning it’s practical to use computers to solve instances of them.
Thousands of other solvable problems are not known to have polynomial-time algorithms, despite ongoing intensive efforts to find such algorithms. These include the traveling salesman problem.
The traveling salesman problem asks whether a set of points with some points directly connected, called a graph, has a path that starts from any point and goes through every other point exactly once, and comes back to the original point. Imagine that a salesman wants to find a route that passes all households in a neighborhood exactly once and returns to the starting point.
These problems, called NP-complete, were independently formulated and shown to exist in the early 1970s by two computer scientists, American Canadian Stephen Cook and Ukrainian American Leonid Levin. Cook, whose work came first, was awarded the 1982 Turing Award, the highest in computer science, for this work.
The Cost of Knowing ExactlyThe best-known algorithms for NP-complete problems are essentially searching for a solution from all possible answers. The traveling salesman problem on a graph of a few hundred points would take years to run on a supercomputer. Such algorithms are inefficient, meaning there are no mathematical shortcuts.
Practical algorithms that address these problems in the real world can only offer approximations, though the approximations are improving. Whether there are efficient polynomial-time algorithms that can solve NP-complete problems is among the seven millennium open problems posted by the Clay Mathematics Institute at the turn of the 21st century, each carrying a prize of a million dollars.
Beyond TuringCould there be a new form of computation beyond Turing’s framework? In 1982, American physicist Richard Feynman, a Nobel laureate, put forward the idea of computation based on quantum mechanics.
In 1995, Peter Shor, an American applied mathematician, presented a quantum algorithm to factor integers in polynomial time. Mathematicians believe that this is unsolvable by polynomial-time algorithms in Turing’s framework. Factoring an integer means finding a smaller integer greater than one that can divide the integer. For example, the integer 688,826,081 is divisible by a smaller integer 25,253, because 688,826,081 = 25,253 x 27,277.
A major algorithm called the RSA algorithm, widely used in securing network communications, is based on the computational difficulty of factoring large integers. Shor’s result suggests that quantum computing, should it become a reality, will change the landscape of cybersecurity.
Can a full-fledged quantum computer be built to factor integers and solve other problems? Some scientists believe it can be. Several groups of scientists around the world are working to build one, and some have already built small-scale quantum computers.
Nevertheless, like all novel technologies invented before, issues with quantum computation are almost certain to arise that would impose new limits.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Laura Ockel / Unsplash
ARTIFICIAL INTELLIGENCED-ID’s New Web App Gives a Face and Voice to OpenAI’s ChatGPT
Aisha Malik | TechCrunch“When you open up the web app on a desktop or mobile device, you’ll be greeted by an avatar named ‘Alice.’ You can then choose to either type out a question or click the microphone icon to say your query out loud. D-ID notes that Alice can answer almost anything. You can ask Alice to simulate a job interview or even host your family’s trivia night. …In a few weeks, the web app will let users generate a character, such as Dumbledore from Harry Potter, and talk to them.”
COMPUTINGTwo Oddball Ideas for a Megaqubit Quantum Computer
Samuel K. Moore | IEEE Spectrum“Experts say quantum computers might need at least a million qubits kept at near absolute zero to do anything computationally noteworthy. But connecting them all by coaxial cable to control and readout electronics, which work at room temperature, would be impossible. Computing giants such as IBM, Google, and Intel hope to solve that problem with cyrogenic silicon chips that can operate close to the qubits themselves. But researchers have recently put forward some more exotic solutions that could quicken the pace.”
SCIENCERoom-Temperature Superconductor Discovery Meets With Resistance
Charlie Wood and Zack Savitsky | Quanta“The results, published [this week] in Nature, appear to show that a conventional conductor—a solid composed of hydrogen, nitrogen and the rare-earth metal lutetium—was transformed into a flawless material capable of conducting electricity with perfect efficiency. While the announcement has been greeted with enthusiasm by some scientists, others are far more cautious, pointing to the research group’s controversial history of alleged research malfeasance.”
LONGEVITYSam Altman Invested $180 Million Into a Company Trying to Delay Death
Antonio Regalado | MIT Technology Review“[Altman] says he’s emptied his bank account to fund two other very different but equally ambitious goals: limitless energy and extended life span. One of those bets is on the fusion power startup Helion Energy, into which he’s poured more than $375 million, he told CNBC in 2021. The other is Retro, to which Altman cut checks totaling $180 million the same year. ‘It’s a lot. I basically just took all my liquid net worth and put it into these two companies,’ Altman says.”
ETHICSMeta’s Powerful AI Language Model Has Leaked Online—What Happens Now?
James Vincent | The Verge“Meta did not release LLaMA as a public chatbot (though the Facebook owner is building those too) but as an open-source package that anyone in the AI community can request access to. …However, just one week after Meta started fielding requests to access LLaMA, the model was leaked online. On March 3rd, a downloadable torrent of the system was posted on 4chan and has since spread across various AI communities, sparking debate about the proper way to share cutting-edge research in a time of rapid technological change.”
BIOTECHForget Designer Babies. Here’s How CRISPR Is Really Changing Lives
Antonio Regalado | MIT Technology Review“…there are now more than 50 experimental studies underway that use gene editing in human volunteers to treat everything from cancer to HIV and blood diseases, according to a tally shared with MIT Technology Review by David Liu, a gene-editing specialist at Harvard University. Most of these studies—about 40 of them—involve CRISPR, the most versatile of the gene-editing methods, which was developed only 10 years ago.”
ETHICSCould the Next Blockbuster Drug Be Lab-Rat Free?
Emily Anthes | The New York Times“…momentum is building for non-animal approaches, which could ultimately help speed drug development, improve patient outcomes and reduce the burdens borne by lab animals, experts said. ‘Animals are simply a surrogate for predicting what’s going to happen in a human,’ said Nicole Kleinstreuer, director of the National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods. ‘If we can get to a place where we actually have a fully human-relevant model,’ she added, ‘then we don’t need the black box of animals anymore.’i”
ENERGYThis Geothermal Startup Showed Its Wells Can Be Used Like a Giant Underground Battery
James Temple | MIT Technology Review“The results from the initial experiments…suggest Fervo can create flexible geothermal power plants, capable of ramping electricity output up or down as needed. Potentially more important, the system can store up energy for hours or even days and deliver it back over similar periods, effectively acting as a giant and very long-lasting battery. That means the plants could shut down production when solar and wind farms are cranking, and provide a rich stream of clean electricity when those sources flag.”
DIGITAL MEDIADetection Stays One Step Ahead of Deepfakes—For Now
Matthew Hutson | IEEE Spectrum“…as computer scientists devise better methods for algorithmically generating video, audio, images, and text—typically for more constructive uses such as enabling artists to manifest their visions—they’re also creating counter-algorithms to detect such synthetic content. Recent research shows progress in making detection more robust, sometimes by looking beyond subtle signatures of particular generation tools and instead utilizing underlying physical and biological signals that are hard for AI to imitate.”
ARTIFICIAL INTELLIGENCEGPT-4 Might Just Be a Bloated, Pointless Mess
Jacob Stern | The Atlantic“Will endless ‘scaling’ of our current language models really bring true machine intelligence? ...the scaling debate is representative of the broader AI discourse. It feels as though the vocal extremes have drowned out the majority. Either ChatGPT will completely reshape our world or it’s a glorified toaster. The boosters hawk their 100-proof hype, the detractors answer with leaden pessimism, and the rest of us sit quietly somewhere in the middle, trying to make sense of this strange new world.“
Image Credit: Laura Skinner / Unsplash
Despite impressive progress, today’s AI models are very inefficient learners, taking huge amounts of time and data to solve problems humans pick up almost instantaneously. A new approach could drastically speed things up by getting AI to read instruction manuals before attempting a challenge.
One of the most promising approaches to creating AI that can solve a diverse range of problems is reinforcement learning, which involves setting a goal and rewarding the AI for taking actions that work towards that goal. This is the approach behind most of the major breakthroughs in game-playing AI, such as DeepMind’s AlphaGo.
As powerful as the technique is, it essentially relies on trial and error to find an effective strategy. This means these algorithms can spend the equivalent of several years blundering through video and board games until they hit on a winning formula.
Thanks to the power of modern computers, this can be done in a fraction of the time it would take a human. But this poor “sample-efficiency” means researchers need access to large numbers of expensive specialized AI chips, which restricts who can work on these problems. It also seriously limits the application of reinforcement learning to real-world situations where doing millions of run-throughs simply isn’t feasible.
Now a team from Carnegie Mellon University has found a way to help reinforcement learning algorithms learn much faster by combining them with a language model that can read instruction manuals. Their approach, outlined in a pre-print published on arXiv, taught an AI to play a challenging Atari video game thousands of times faster than a state-of-the-art model developed by DeepMind.
“Our work is the first to demonstrate the possibility of a fully-automated reinforcement learning framework to benefit from an instruction manual for a widely studied game,” said Yue Wu, who led the research. “We have been conducting experiments on other more complicated games like Minecraft, and have seen promising results. We believe our approach should apply to more complex problems.”
Atari video games have been a popular benchmark for studying reinforcement learning thanks to the controlled environment and the fact that the games have a scoring system, which can act as a reward for the algorithms. To give their AI a head start, though, the researchers wanted to give it some extra pointers.
First, they trained a language model to extract and summarize key information from the game’s official instruction manual. This information was then used to pose questions about the game to a pre-trained language model similar in size and capability to GPT-3. For instance, in the game PacMan this might be, “Should you hit a ghost if you want to win the game?”, for which the answer is no.
These answers are then used to create additional rewards for the reinforcement algorithm, beyond the game’s built-in scoring system. In the PacMan example, hitting a ghost would now attract a penalty of -5 points. These extra rewards are then fed into a well-established reinforcement learning algorithm to help it learn the game faster.
The researchers tested their approach on Skiing 6000, which is one of the hardest Atari games for AI to master. The 2D game requires players to slalom down a hill, navigating in between poles and avoiding obstacles. That might sound easy enough, but the leading AI had to run through 80 billion frames of the game to achieve comparable performance to a human.
In contrast, the new approach required just 13 million frames to get the hang of the game, although it was only able to achieve a score about half as good as the leading technique. That means it’s not as good as even the average human, but it did considerably better than several other leading reinforcement learning approaches that couldn’t get the hang of the game at all. That includes the well-established algorithm the new AI relies on.
The researchers say they have already begun testing their approach on more complex 3D games like Minecraft, with promising early results. But reinforcement learning has long struggled to make the leap from video games, where the computer has access to a complete model of the world, to the messy uncertainty of physical reality.
Wu says he is hopeful that rapidly improving capabilities in object detection and localization could soon put applications like autonomous driving or household automation within reach. Either way, the results suggest that rapid improvements in AI language models could act as a catalyst for progress elsewhere in the field.
Image Credit: Kreg Steppe / Flickr
Earth’s surface is the “living skin” of our planet—it connects the physical, chemical, and biological systems. Over geological time, landscapes change as this surface evolves, regulating the carbon cycle and nutrient circulation as rivers carry sediment into the oceans.
All these interactions have far-reaching effects on ecosystems and biodiversity—the many living things inhabiting our planet.
As such, reconstructing how Earth’s landscapes have evolved over millions of years is a fundamental step towards understanding the changing shape of our planet, and the interaction of things like the climate and tectonics. It can also give us clues on the evolution of biodiversity.
Working with scientists in France (French National Center for Scientific Research, ENS Paris university, University of Grenoble, and University of Lyon), our team at the University of Sydney has now published a detailed geological model of Earth’s surface changes in the prestigious journal Science.
Ours is the first dynamic model—a computer simulation—of the past 100 million years at a high resolution down to ten kilometers. In unprecedented detail, it reveals how Earth’s surface has changed over time, and how that has affected the way sediment moves around and settles.
Broken into frames of a million years, our model is based on a framework that incorporates plate tectonic and climatic forces with surface processes such as earthquakes, weathering, changing rivers, and more.
Three Years in the MakingThe project started about three years ago when we began the development of a new global-scale landscape evolution model, capable of simulating millions of years of change. We also found ways to automatically add other information into our framework, such as paleogeography—the history of Earth’s landscapes.
For this new study, our framework used state-of-the-art plate tectonic reconstructions and simulations of past climates on a global scale.
Our advanced computer simulations used Australia’s National Computational Infrastructure, running on hundreds of computer processors. Each simulation took several days, building a complete picture to reconstruct the past 100 million years of Earth’s surface evolution.
All this computing power has resulted in global high-resolution maps that show the highs and lows of Earth’s landscapes (elevation), as well as the flows of water and sediment.
All of these fit well with existing geological observations. For instance, we combined data from present-day river sediment and water flows, drainage basin areas, seismic surveys, and long-term local and global erosion trends.
Our main outputs are available as time-based global maps at five-million-year intervals from the Open Science Framework.
Water and Sediment Flux Through Space and TimeOne of Earth’s fundamental surface processes is erosion, a slow process in which materials like soil and rock are worn and carried away by wind or water. This results in sediment flows.
Erosion plays an important role in Earth’s carbon cycle—the never-ending global circulation of one of life’s essential building blocks, carbon. Investigating the way sediment flows have changed through space and time is crucial for our understanding of how Earth’s climates have varied in the past.
We found that our model reproduces the key elements of Earth’s sediment transport, from catchment dynamics depicting river networks over time to the slow changes of large-scale sedimentary basins.
From our results, we also found several inconsistencies between existing observations of rock layers (strata), and predictions of such layers. This shows our model could be useful for testing and refining reconstructions of past landscapes.
Our simulated past landscapes are fully integrated with the various processes at play, especially the hydrological system—the movement of water—providing a more robust and detailed view of Earth’s surface.
Our study reveals more detail on the role that the constantly-evolving Earth’s surface has played in the movement of sediments from mountaintops to ocean basins, ultimately regulating the carbon cycle and Earth’s climate fluctuations through deep time.
As we explore these results in tandem with the geological record, we will be able to answer long-standing questions about various crucial features of the Earth system—including the way our planet cycles nutrients, and has given rise to life as we know it.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Sander Lenaerts on Unsplash
According to the IEA, there are currently 18 direct air capture plants in operation around the world. They’re located in Europe, Canada, or the US, and most of them use the CO2 for commercial purposes, with a couple storing it away for all eternity. Direct air capture (DAC) is a controversial technology, with opponents citing its high cost and energy usage. Indeed, when you consider the amount of CO2 in the atmosphere relative to the amount that any single DAC plant—or many of them collectively—can capture, and hold that up against their cost, it seems a bit silly to even be trying.
But given the lack of other great options available to stop the planet from bursting into flames, both the Intergovernmental Panel on Climate Change and the International Energy Agency say we shouldn’t discard DAC just yet—on the contrary, we should be trying to find ways to cut its costs and up its efficiency. A team from Lehigh University and Tianjin University have made one such breakthrough, developing a material they say can capture three times as much carbon as those currently in use.
Described in a paper published today in Science Advances, the material could make DAC a far more viable technology by eliminating some of its financial and practical obstacles, the team says.
Many of the carbon capture plants that are currently operational or under construction (including Iceland’s Orca and Mammoth and Wyoming’s Project Bison) use solid DAC technology: blocks of fans push air through sorbent filters that chemically bind with CO2. The filters need to be heated and placed under a vacuum to release the CO2, which must then be compressed under extremely high pressure.
These last steps are what drive carbon capture’s energy use and costs so high. The CO2 in Earth’s atmosphere is very diluted; according to the paper’s authors, its average concentration is about 400 parts per million. That means a lot of air needs to be blown through the sorbent filters for them to capture just a little CO2. Since it takes so much energy to separate the captured CO2 (called the “desorption” process), we want as much CO2 as possible to be getting captured in the first place.
The Lehigh-Tianjin team created what they call a hybrid sorbent. They started with a synthetic resin, which they soaked in a copper-chloride solution. The copper acts as a catalyst for the reaction that causes CO2 to bind to the resin, making the reaction go faster and use less energy. Besides being mechanically strong and chemically stable, the sorbent can be regenerated using salt solutions—including seawater—at temperatures lower than 90 degrees Celsius.
The team reported that one kilogram of their material was able to absorb 5.1 mol of CO2; in comparison, most solid sorbents currently in use for DAC have absorption capacities of 1.0 to 1.5 mol per kilogram. In between capture cycles they used seawater to regenerate the capture column, repeating the cycle 15 times without a noticeable decrease in the amount of CO2 the material was able to capture.
The main byproduct of the chemical reaction was carbonic acid, which the team noted can be easily neutralized into baking soda and deposited in the ocean. “Spent regenerant can be safely returned to the sea, an infinite sink for captured CO2,” they wrote. “Such a sequestration technique will also eliminate the energy needed for pressurizing and liquefying CO2 before deepwell injection.” This method would be most relevant in locations close to an ocean where geological storage—that is, injecting CO2 underground to turn it into rock—isn’t possible.
Using this newly-created material in large-scale carbon capture operations could be a game-changer. Not only would the manufacturing process for the sorbent be cheap and scalable, it would capture more CO2 and require less energy.
But would all that be enough to make direct air capture worthwhile, and truly put a dent in atmospheric CO2? To put it bluntly, probably not. Right now the world’s DAC facilities collectively capture 0.01 million metric tons of CO2. The IEA’s 2022 report on the technology estimates we’ll need to be capturing 85 million metric tons by 2030 to avoid the worst impacts of climate change.
No matter which way you do the math, it seems like a long shot; rather than a material that absorbs three times as much CO2 per unit, we need one that absorbs 3,000 times as much. But as we’ve witnessed throughout history, most scientific advances happen incrementally, not all at once. If we’re to reach a point where direct air capture is a true solution, it will take many more baby steps—like this one—to get there.
Image Credit: Michaela / Pixabay
The human brain is a master of computation. It’s no wonder that from brain-inspired algorithms to neuromorphic chips, scientists are borrowing the brain’s playbook to give machines a boost.
Yet the results—in both software and hardware—only capture a fraction of the computational intricacies embedded in neurons. But perhaps the major roadblock in building brain-like computers is that we still don’t fully understand how the brain works. For example, how does its architecture—defined by pre-established layers, regions, and ever-changing neural circuits—make sense of our chaotic world with high efficiency and low energy usage?
So why not sidestep this conundrum and use neural tissue directly as a biocomputer?
This month, a team from Johns Hopkins University laid out a daring blueprint for a new field of computing: organoid intelligence (OI). Don’t worry—they’re not talking about using living human brain tissue hooked up to wires in jars. Rather, as in the name, the focus is on a surrogate: brain organoids, better known as “mini-brains.” These pea-sized nuggets roughly resemble the early fetal human brain in their gene expression, wide variety of brain cells, and organization. Their neural circuits spark with spontaneous activity, ripple with brain waves, and can even detect light and control muscle movement.
In essence, brain organoids are highly-developed processors that duplicate the brain to a limited degree. Theoretically, different types of mini-brains could be hooked up to digital sensors and output devices—not unlike brain-machine interfaces, but as a circuit outside the body. In the long term, they may connect to each other in a super biocomputer trained using biofeedback and machine learning methods to enable “intelligence in a dish.”
Sound a bit creepy? I agree. Scientists have long debated where to draw the line; that is, when the mini-brain becomes too similar to a human one, with the hypothetical nightmare scenario of the nuggets developing consciousness.
The team is well aware. As part of organoid intelligence, they highlight the need for “embedded ethics,” with a consortium of scientists, bioethicists, and the public weighing in throughout development. But to senior author Dr. Thomas Hartung, the time for launching organoid intelligence research is now.
“Biological computing (or biocomputing) could be faster, more efficient, and more powerful than silicon-based computing and AI, and only require a fraction of the energy,” the team wrote.
A Brainy SolutionUsing brain tissue as computational hardware may seem bizarre, but there’ve been previous pioneers. In 2022, the Australian company Cortical Labs taught hundreds of thousands of isolated neurons in a dish to play Pong inside a virtual environment. The neurons connected with silicon chips powered by deep learning algorithms into a “synthetic biological intelligence platform” that captured basic neurobiological signs of learning.
Here, the team took the idea a step further. If isolated neurons could already support a rudimentary form of biocomputing, what about 3D mini-brains?
Since their debut a decade ago, mini-brains have become darlings for examining neurodevelopmental disorders such as autism and testing new drug treatments. Often grown from a patient’s skin cells—transformed into induced pluripotent stem cells (iPSCs)—the organoids are especially powerful for mimicking a person’s genetic makeup, including their neural wiring. More recently, human organoids partially restored damaged vision in rats after integrating with their host neurons.
In other words, mini-brains are already building blocks for a plug-and-play biocomputing system that readily connects with biological brains. So why not leverage them as processors for a computer? “The question is: can we learn from and harness the computing capacity of these organoids?” the team asked.
A Hefty BlueprintLast year, a group of biocomputing experts united in the first organoid intelligence workshop in an effort to form a community tackling the use and implications of mini-brains as biocomputers. The overarching theme, consolidated into “the Baltimore declaration,” was collaboration. A mini-brain system needs several components: devices to detect input, the processor, and a readable output.
In the new paper, Hartung envisions four trajectories to accelerate organoid intelligence.
The first focuses on the critical component: the mini-brain. Although densely packed with brain cells that support learning and memory, organoids are still difficult to culture on a large scale. An early key aim, explained the authors, is scaling up.
Microfluidic systems, which act as “nurseries,” also need to improve. These high-tech bubble baths provide nutrients and oxygen to keep burgeoning mini-brains alive and healthy while removing toxic waste, giving them time to mature. The same system can also pump neurotransmitters—molecules that bridge communication between neurons—into specific regions to modify their growth and behavior.
Scientists can then monitor growth trajectories using a variety of electrodes. Although most are currently tailored for 2D systems, the team and others are leveling up with 3D interfaces specifically designed for organoids, inspired by EEG (electroencephalogram) caps with multiple electrodes placed in a spherical shape.
Then comes the decoding of signals. The second trajectory is all about deciphering the whens and wheres of neural activity inside the mini-brains. When zapped with certain electrical patterns—for example, those that encourage the neurons to play Pong—do they output the expected results?
It’s another hard task; learning changes neural circuits on multiple levels. So what to measure? The team suggests digging into multiple levels, including altered gene expression in neurons and how they connect into neural networks.
Here is where AI and collaboration can make a splash. Biological neural networks are noisy, so multiple trials are needed before “learning” becomes apparent—in turn generating a deluge of data. To the team, machine learning is the perfect tool to extract how different inputs, processed by the mini-brain, transform into outputs. Similar to large-scale neuroscience projects such as the BRAIN Initiative, scientists can share their organoid intelligence research in a community workspace for global collaborations.
Trajectory three is further in the future. With efficient and long-lasting mini-brains and measuring tools in hand, it’s possible to test more complex inputs and see how the stimulation feeds back into the biological processor. For example, does it make its computation more efficient? Different types of organoids—say, those that resemble the cortex and the retina—can be interconnected to build more complex forms of organoid intelligence. These could help “empirically test, explore, and further develop neurocomputational theories of intelligence,” the authors wrote.
Intelligence on Demand?The fourth trajectory is the one that underlines the entire project: the ethics of using mini-brains for biocomputing.
As brain organoids increasingly resemble the brain—so much so that they can integrate and partially restore a rodent’s injured visual system—scientists are asking if they may gain a sort of awareness.
To be clear, there is no evidence that mini-brains are conscious. But “these concerns will mount during the development of organoid intelligence, as the organoids become structurally more complex, receive inputs, generate outputs, and—at least theoretically—process information about their environment and build a primitive memory,” the authors said. However, the goal of organoid intelligence isn’t to recreate human consciousness—rather, it’s to mimic the brain’s computational functions.
The mini-brain processor is hardly the only ethical concern. Another is cell donation. Because mini-brains retain their donor’s genetic makeup, there’s a chance of selection bias and limitation on neurodiversity.
Then there’s the problem of informed consent. As history with the famous cancer cell line HeLa cells has shown, cell donation can have multi-generational impacts. “What does the organoid exhibit about the cell donor?” the authors asked. Will researchers have an obligation to inform the donor if they discover neurological disorders during their research?
To navigate the “truly uncharted territory,” the team proposes an embedded ethics approach. At each step, bioethicists will collaborate with research teams to map out potential issues iteratively while gathering public opinions. The strategy is similar to other controversial topics, such as genetic editing in humans.
A mini-brain-powered computer is years away. “It will take decades before we achieve the goal of something comparable to any type of computer,” said Hartung. But it’s time to start—launching the program, consolidating multiple technologies across fields, and engaging in ethical discussions.
“Ultimately, we aim toward a revolution in biological computing that could overcome many of the limitations of silicon-based computing and AI and have significant implications worldwide,” the team said.
Image Credit: Jesse Plotkin/Johns Hopkins University
What would we do if we spotted a hazardous asteroid on a collision course with Earth? Could we deflect it safely to prevent the impact?
Last year, NASA’s Double Asteroid Redirection Test (DART) mission tried to find out whether a “kinetic impactor” could do the job: smashing a 600-kilogram spacecraft the size of a fridge into an asteroid the size of the Roman Colosseum.
Early results from this first real-world test of our potential planetary defense systems looked promising. However, it’s only now that the first scientific results are being published: five papers in Nature have recreated the impact, and analyzed how it changed the asteroid’s momentum and orbit, while two studies investigate the debris knocked off by the impact.
The conclusion: “Kinetic impactor technology is a viable technique to potentially defend Earth if necessary.”
Small Asteroids Could Be Dangerous, but Hard to SpotOur Solar System is full of debris, left over from the early days of planet formation. Today, some 31,360 asteroids are known to loiter around Earth’s neighborhood.
Asteroid statistics and the threats posed by asteroids of different sizes. Image Credit: NASA’s DART press briefAlthough we have tabs on most of the big, kilometer-sized ones that could wipe out humanity if they hit Earth, most of the smaller ones go undetected.
Just over 10 years ago, an 18-meter asteroid exploded in our atmosphere over Chelyabinsk, Russia. The shockwave smashed thousands of windows, wreaking havoc and injuring some 1,500 people.
A 150-meter asteroid like Dimorphos wouldn’t wipe out civilization, but it could cause mass casualties and regional devastation. However, these smaller space rocks are harder to find: we think we have only spotted around 40 percent of them so far.
The DART MissionSuppose we did spy an asteroid of this scale on a collision course with Earth. Could we nudge it in a different direction, steering it away from disaster?
Hitting an asteroid with enough force to change its orbit is theoretically possible, but can it actually be done? That’s what the DART mission set out to determine.
Specifically, it tested the “kinetic impactor” technique, which is a fancy way of saying “hitting the asteroid with a fast-moving object.”
The asteroid Dimorphos was a perfect target. It was in orbit around its larger cousin, Didymos, in a loop that took just under 12 hours to complete.
The impact from the DART spacecraft was designed to slightly change this orbit, slowing it down just a little so that the loop would shrink, shaving an estimated seven minutes off its round trip.
A Self-Steering SpacecraftFor DART to show the kinetic impactor technique is a possible tool for planetary defense, it needed to demonstrate two things: that its navigation system could autonomously maneuver and target an asteroid during a high-speed encounter, and that such an impact could change the asteroid’s orbit.
In the words of Cristina Thomas of Northern Arizona University and colleagues, who analyzed the changes to Dimorphos’ orbit as a result of the impact, “DART has successfully done both.”
The DART spacecraft steered itself into the path of Dimorphos with a new system called Small-body Maneuvering Autonomous Real Time Navigation (SMART Nav), which used the onboard camera to get into a position for maximum impact.
More advanced versions of this system could enable future missions to choose their own landing sites on distant asteroids where we can’t image the rubble-pile terrain well from Earth. This would save the trouble of a scouting trip first!
Dimorphos itself was one such asteroid before DART. A team led by Terik Daly of Johns Hopkins University has used high-resolution images from the mission to make a detailed shape model. This gives a better estimate of its mass, improving our understanding of how these types of asteroids will react to impacts.
Dangerous DebrisThe impact itself produced an incredible plume of material. Jian-Yang Li of the Planetary Science Institute and colleagues have described in detail how the ejected material was kicked up by the impact and streamed out into a 1,500-kilometer tail of debris that could be seen for almost a month.
The DART impact blasted a vast plume of dust and debris from the surface of the asteroid Dimorphos. Image Credit: CTIO / NOIRLab / SOAR / NSF / AURA / T. Kareta (Lowell Observatory), M. Knight (US Naval Academy)Streams of material from comets are well known and documented. They are mainly dust and ice and are seen as harmless meteor showers if they cross paths with Earth.
Asteroids are made of rockier, stronger stuff, so their streams could pose a greater hazard if we encounter them. Recording a real example of the creation and evolution of debris trails in the wake of an asteroid is very exciting. Identifying and monitoring such asteroid streams is a key objective of planetary defense efforts such as the Desert Fireball Network we operate from Curtin University.
A Bigger Than Expected ResultSo how much did the impact change Dimorphos’ orbit? By much more than the expected amount. Rather than changing by 7 minutes, it had become 33 minutes shorter!
This larger-than-expected result shows the change in Dimorphos’ orbit was not just from the impact of the DART spacecraft. The larger part of the change was due to a recoil effect from all the ejected material flying off into space, which Ariel Graykowski of the SETI Institute and colleagues estimated as between 0.3 percent and 0.5 percent of the asteroid’s total mass.
A First SuccessThe success of NASA’s DART mission is the first demonstration of our ability to protect Earth from the threat of hazardous asteroids.
At this stage, we still need quite a bit of warning to use this kinetic impactor technique. The earlier we intervene in an asteroid’s orbit, the smaller the change we need to make to push it away from hitting Earth. (To see how it all works, you can have a play with NASA’s NEO Deflection app.)
But should we? This is a question that will need answering if we ever do have to redirect a hazardous asteroid. In changing the orbit, we’d have to be sure we weren’t going to push it in a direction that would hit us in future too.
However, we are getting better at detecting asteroids before they reach us. We have seen two in the past few months alone: 2022WJ1, which impacted over Canada in November, and Sar2667, which came in over France in February.
We can expect to detect a lot more in future, with the opening of the Vera Rubin Observatory in Chile at the end of this year.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: CTIO / NOIRLab / SOAR / NSF / AURA/ T. Kareta (Lowell Observatory), M. Knight (US Naval Academy)
The enormous potential of AI to reshape the future has seen massive investment from industry in recent years. But the growing influence of private companies in the basic research that is powering this emerging technology could have serious implications for how it develops, say researchers.
The question of whether machines could replicate the kind of intelligence seen in animals and humans is almost as old as the field of computer science itself. Industry’s engagement with this line of research has fluctuated over the decades, leading to a series of AI winters as investment has flowed in and then back out again as the technology has failed to live up to expectations.
The advent of deep learning at the turn of the previous decade, however, has resulted in one of the most sustained runs of interest and investment from private companies. This is now beginning to yield some truly game-changing AI products, but a new analysis in Science shows that it’s also leading to industry taking an increasingly dominant position in AI research.
This is a doubled-edged sword, say the authors. Industry brings with it money, computing resources, and vast amounts of data that have turbo-charged progress, but it is also refocusing the entire field on areas that are of interest to private companies rather than those with the greatest potential or benefit to humanity.
“Industry’s commercial motives push them to focus on topics that are profit-oriented. Often such incentives yield outcomes in line with the public interest, but not always,” the authors write. “Although these industry investments will benefit consumers, the accompanying research dominance should be a worry for policy-makers around the world because it means that public interest alternatives for important AI tools may become increasingly scarce.”
The authors show that industry’s footprint in AI research has increased dramatically in recent years. In 2000, only 22 percent of presentations at leading AI conferences featured one or more co-authors from private companies, but by 2020 that had hit 38 percent. But the impact is most clearly felt at the cutting edge of the field.
Progress in deep learning has to a large extent been driven by the development of ever larger models. In 2010, industry accounted for only 11 percent of the biggest AI models, but by 2021 that had hit 96 percent. This has coincided with growing dominance on key benchmarks in areas like image recognition and language modeling, where industry involvement in the leading model has grown from 62 percent in 2017 to 91 percent in 2020.
A key driver of this shift is the much larger investments the private sector is able to make compared to public bodies. Excluding defense spending, the US government allocated $1.5 billion for spending on AI in 2021, compared to the $340 billion spent by industry around the world that year.
That extra funding translates to far better resources—both in terms of computing power and data access—and the ability to attract the best talent. The size of AI models is strongly correlated with the amount of data and computing resources available, and in 2021 industry models were 29 times larger than academic ones on average.
And while in 2004 only 21 percent of computer science PhDs that had specialized in AI went into industry, by 2020 that had jumped to almost 70 percent. The rate at which AI experts have been hired away from university by private companies has also increased eight-fold since 2006.
The authors point to OpenAI as a marker of the increasing difficulty of doing cutting-edge AI research without the financial resources of the private sector. In 2019, the organization transformed from a non-profit to a “capped for-profit organization” in order to “rapidly increase our investments in compute and talent,” the company said at the time.
This extra investment has had its perks, the authors note. It’s helped to bring AI technology out of the lab and into everyday products that can improve people’s lives. It’s also led to the development of a host of valuable tools used by industry and academia alike, such as software packages like TensorFlow and PyTorch and increasingly powerful computer chips tailored to AI workloads.
But it’s also pushing AI research to focus on areas with potential commercial benefits for its sponsors, and just as importantly, data-hungry and computationally-expensive AI approaches that dovetail nicely with the kind of things big technology companies are already good at. As industry increasingly sets the direction of AI research, this could lead to the neglect of competing approaches towards AI and other socially beneficial applications with no clear profit motive.
“Given how broadly AI tools could be applied across society, such a situation would hand a small number of technology firms an enormous amount of power over the direction of society,” the authors note.
There are models for how the gap between the private and public sector could be closed, say the authors. The US has proposed the creation of a National AI Research Resource made up of public research cloud and public datasets. China recently approved a “national computing power network system.” And Canada’s Advanced Research Computing platform has been running for almost a decade.
But without intervention from policymakers, the authors say that academics will likely be unable to properly interpret and critique industry models or offer public interest alternatives. Ensuring they have the capabilities to continue to shape the frontier of AI research should be a key priority for governments around the world.
Image Credit: DeepMind / Unsplash
ARTIFICIAL INTELLIGENCEMicrosoft Unveils AI Model That Understands Image Content, Solves Visual Puzzles
Benj Edwards | Ars Technica“On Monday, researchers from Microsoft introduced Kosmos-1, a multimodal model that can reportedly analyze images for content, solve visual puzzles, perform visual text recognition, pass visual IQ tests, and understand natural language instructions. The researchers believe multimodal AI—which integrates different modes of input such as text, audio, images, and video—is a key step to building artificial general intelligence (AGI) that can perform general tasks at the level of a human.”
ROBOTICSFigure Promises First General-Purpose Humanoid Robot
Evan Ackerman | IEEE Spectrum“Over the past year, the company has hired more than 40 engineers from institutions that include IHMC, Boston Dynamics, Tesla, Waymo, and Google X, most of whom have significant prior experience with humanoid robots or other autonomous systems. ‘It’s our view that this is the best humanoid robotics team out there,’ Adcock tells IEEE Spectrum.”
CRYPTOCURRENCYEthereum Moved to Proof of Stake. Why Can’t Bitcoin?
Amy Castor | MIT Technology Review“A single Bitcoin transaction uses the same amount of energy as a single US household does over the course of nearly a month. But does it have to be that way? The Bitcoin community has historically been fiercely resistant to change, but pressure from regulators and environmentalists fed up with Bitcoin’s massive carbon footprint may force them to rethink that stance.”
ARTIFICIAL INTELLIGENCEThe Inside Story of How ChatGPT Was Built From the People Who Made It
Will Douglas Heaven | MIT Technology Review“When OpenAI launched ChatGPT, with zero fanfare, in late November 2022, the San Francisco–based artificial-intelligence company had few expectations. Certainly, nobody inside OpenAI was prepared for a viral mega-hit. The firm has been scrambling to catch up—and capitalize on its success—ever since. …To get the inside story behind the chatbot—how it was made, how OpenAI has been updating it since release, and how its makers feel about its success—I talked to four people who helped build what has become one of the most popular internet apps ever.”
ETHICSFace Recognition Software Led to His Arrest. It Was Dead Wrong
Khari Johnson | Wired“The Alonzo Sawyer case adds to just a handful of known instances of innocent people getting arrested following investigations that involved face recognition misidentification—all have been Black men. Three cases came to light in 2019 and 2020 and another last month in which Georgia resident Randal Reid was released from jail after a judge recalled an arrest warrant linking him to thefts of designer purses in Louisiana.”
GOVERNANCEAs AI Booms, Lawmakers Struggle to Understand the Technology
Cecilia Kang and Adam Satariano | The New York Times“The problem is that most lawmakers do not even know what AI is, said Representative Jay Obernolte, a California Republican and the only member of Congress with a master’s degree in artificial intelligence. ‘Before regulation, there needs to be agreement on what the dangers are, and that requires a deep understanding of what AI is,’ he said. ‘You’d be surprised how much time I spend explaining to my colleagues that the chief dangers of AI will not come from evil robots with red lasers coming out of their eyes.’i”
SCIENCEKey Steps in Evolution on Earth Tell Us How Likely Intelligent Life Is Anywhere Else
Adam Frank | Big Think“There are trillions of planets where life could form. But what are the odds that intelligence could evolve on any of them? The Hard Steps Model identifies the unlikely accidents that led to intelligent life on Earth. It allows for the possibility of mathematically modeling the possibility of life emerging elsewhere. The model makes it seem like intelligence in the cosmos will be really, really rare.”
TECHStability AI, Hugging Face and Canva Back New AI Research Nonprofit
Kyle Wiggers | TechCrunch“Developing cutting-edge AI systems like ChatGPT requires massive technical resources, in part because they’re costly to develop and run. While several open source efforts have attempted to reverse-engineer proprietary, closed source systems created by commercial labs such as Alphabet’s DeepMind and OpenAI, they’ve often run into roadblocks—mainly due to a lack of capital and domain expertise. Hoping to avoid this fate, one community research group, EleutherAI, is forming a nonprofit foundation.”
Image Credit: Fernand De Canne / Unsplash
Hunter-gatherers took shelter from the ice age in Southwestern Europe, but were replaced on the Italian Peninsula according to two new studies, published in Nature and Nature Ecology & Evolution today.
Modern humans first began to spread across Eurasia approximately 45,000 years ago, arriving from the near east. Previous research claimed these people disappeared when massive ice sheets covered much of Europe around 25,000–19,000 years ago. By comparing the DNA of various ancient humans, we show this was not the case for all hunter-gatherer groups.
Our new results show the hunter-gatherers of Central and Southern Europe did disappear during the last ice age. However, their cousins in what is now France and Spain survived, leaving genetic traces still visible in the DNA of Western European peoples nearly 30,000 years later.
Two Studies With One Intertwining StoryIn our first study in Nature, we analyzed the genomes—the complete set of DNA a person carries—of 356 prehistoric hunter-gatherers. In fact, our study compared every available ancient hunter-gatherer genome.
In our second study in Nature Ecology & Evolution, we analyzed the oldest hunter-gatherer genome recovered from the southern tip of Spain, belonging to someone who lived approximately 23,000 years ago. We also analyzed three early farmers who lived roughly 6,000 years ago in southern Spain. This allowed us to fill an important sampling gap for this region.
By combining results from these two studies, we can now describe the most complete story of human history in Europe to date. This story includes migration events, human retreat from the effects of the ice age, long-lasting genetic lineages, and lost populations.
Post-Ice-Age Genetic ReplacementBetween 32,000 and 24,000 years ago, hunter-gatherer individuals (associated with what’s known as Gravettian culture) were widespread across the European continent. This critical time period ends at the Last Glacial Maximum. This was the coldest period of the last ice age in Europe, and took place 24,000 to 19,000 years ago.
Our data show that populations from Southwestern Europe (today’s France and Iberia), and Central and Southern Europe (today’s Italy and Czechia), were not closely genetically related. These two distinct groups were instead linked by similar weapons and art.
We could see that Central and Southern European Gravettian populations left no genetic signal after the Last Glacial Maximum—in other words, they simply disappeared. The individuals associated with a later culture (known as the Epigravettian) were not descendants of the Gravettian. According to one of my Nature co-authors, He Yu, they were “genetically distinct from the area’s previous inhabitants. Presumably, these people came from the Balkans, arrived first in northern Italy around the time of the Last Glacial Maximum, and spread all the way south to Sicily.”
In Central and Southern Europe, our data indicate people associated with the Epigravettian populations of the Italian peninsula later spread across Europe. This occurred approximately 14,000 years ago, following the end of the ice age.
Climate RefugeWhile the Gravettian populations of Central and Southern Europe disappeared, the fate of the Southwestern populations was not the same.
We detected the genetic profile of Southwestern Gravettian populations again and again for the next 20,000 years in Western Europe. We saw this first in their direct descendants (known as Solutrean and Magdalenian cultures). These were the people who took refuge and flourished in Southwestern Europe during the ice age. Once the ice age ended, the Magdalenians spread northeastward, back into Europe.
Remarkably, the 23,000-year-old remains of a Solutrean individual from Cueva de Malalmuerzo in Spain allowed us to make a direct link to the first modern humans that settled Europe. We could connect them to a 35,000-year-old individual from Belgium, and then to hunter-gatherers who lived in Western Europe long after the Last Glacial Maximum.
Sea levels during the ice age were lower, making it only 13 kilometers from the tip of Spain to Northern Africa. However, we observed no genetic links between individuals in southern Spain and northern Morocco from 14,000 years ago. This showed that while European populations retreated south during the ice age, they surprisingly stopped before reaching Northern Africa.
Our results show the special role the Iberian peninsula played as a safe haven for humans during the ice age. The genetic legacy of hunter-gatherers would survive in the region after more than 30,000 years, unlike their distant relatives further east.
Post Ice-Age InteractionSome 2,000 years after the end of the ice age, there were again two genetically distinct hunter-gatherer groups. There was the “old” group in Western and Central Europe, and the “more recent” group in Eastern Europe.
These groups showed no evidence of genetic exchange with southwestern hunter-gatherer populations for approximately 6,000 years, until roughly 8,000 years ago.
At this time, agriculture and a sedentary lifestyle had begun to spread with new peoples from Anatolia into Europe, forcing hunter-gatherers to retreat to the northern fringes of Europe.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Mauricio Anton/Wikimedia Commons
From burgers to sausages and steak tips to chicken nuggets, there’s no shortage of plant-based “meat” products on grocery store shelves and in restaurants these days. Companies like Beyond Meat and Impossible Foods have done an impressive job diversifying their offerings, with almost any processed meat you can think of now on their lists (even beef jerky and popcorn chicken). But a key cut of meat is still missing from these big names’ menus: a good old-fashioned filet, just like the cows make ‘em.
A lesser-known player in the industry has been working to fill this gap. Slovenian startup Juicy Marbles was co-founded in early 2021 by Y Combinator alums Luka Sincek, Tilen Travnik, and Maj Hrova. The company launched its first product, a thick-cut filet mignon, in early 2022, and more recently started selling a whole-cut loin as well.
I received a whole-cut loin to sample; it arrived on dry ice, uncooked, and I was instructed to freeze or cook it within ten days. As a lifelong and unrepenting carnivore, I was wary but curious. It looked like meat: light red, fibrous, slightly moist. It felt like meat: dense, not overly pliable, requiring some pressure to slice through. But how was it going to taste?
Vegan VentureSlicing up the steakI cut the steak into one-inch-thick slices and seasoned them with salt, pepper, paprika, and garlic powder, then pan-fried them in a little bit of vegetable oil for four minutes on each side. Upon flipping them over I was surprised to see that the edges had browned, much like real meat does.
Pan-frying the steakI served the steak with quinoa and sauteed veggies, and after a few bites, I couldn’t deny it was both tasty and had a pleasant texture. Did it taste or feel like a real steak? Not really. The real meat that it most reminded me of is rib meat, the kind that easily pulls off the bone when the ribs have been slow-cooked; the meat is soft and tender, but not dried out. The plant-based steak had a distinctly fatty-like mouthfeel without the excessive oiliness you sometimes get from animal fat.
Marbling MysteryAchieving this texture, and a realistic “marbling” effect, has been one of the biggest challenges for plant-based meat companies. How do you replicate—with plants—animal tissue that has thin ribbons of fat running through it?
Though they can’t reveal too many details of their proprietary technology, Juicy Marbles has disclosed that unlike many plant-based meat companies, they don’t use 3D printing in their production process. Rather, they use a grinder they call the Meat-O-Matic 9000, which layers plant protein fibers on top of each other in a way that resembles muscle fibers. Deposits of hardened sunflower oil help add a realistic fat marbling texture and mouthfeel.
“Our business is based around the concept of protein texture—this is the defining factor that draws people to steak, when compared to a cheaper cut,” the company told TechCrunch. “In the plant-based meat vertical, there has not been as much innovation in the whole cuts space, and no one has come close to inventing a steak that resembles anything high-end.”
Health HighlightsThe first few ingredients listed on the whole-cut loin package are plant structure (70 percent, made up of water, soy protein concentrate, and wheat protein isolate), sunflower oil, natural flavors, beetroot powder, and thickener. In terms of nutritional value, a four-ounce serving has 200 calories, 8 grams of fat, 7 grams of dietary fiber, no cholesterol, and 26 grams of protein.
A three-ounce filet mignon has 227 calories, 15 grams of fat, 6 grams of saturated fat, 82 milligrams of cholesterol, and 22 grams of protein. So the plant-based meat and the real meat are comparable in terms of some key parts of our daily diet.
One motivator for carnivores to go plant-based could be the slightly healthier profile of “meat” that doesn’t come from an animal. Juicy Marbles’ filet mignon got some press a little under a year ago when Lizzo posted a video to TikTok of herself cooking the filet with vegan eggs for breakfast, proclaiming after taking a bite, “It’s good!”
Though I enjoyed the steak and think it’s a high-quality product, I’m not certain I’d purchase it and regularly incorporate it into my meals, even if its price substantially undercut real steak.
Herein lies what I see as the fundamental problem with plant-based meat (and really any plant-based or vegan product that imitates an animal product, from eggs to milk to bacon): people who like the real thing aren’t likely to switch over to an imitation of the real thing, no matter how good it is. If I want meat, a plant-based approximation isn’t going to cut it.
Meanwhile, people who’ve made the choice to exclude meat from their diets may not be looking for a protein source that tastes and feels like meat.
Vladimir Mićković, Juicy Marbles’ Chief Business Officer, doesn’t feel that the company needs to target meat-eaters nor vegetarians. “It feels limiting to see people merely through the lens of their diet, so we don’t like to put them in such groups at all,” he told me in an email. “It doesn’t matter what diet our customers may follow, their taste buds and bodies will be the final judge.”
Food of the Future?Could plant-based meat make a difference for the environment in terms of water, land, and emissions? Sure—but it would need to be adopted on a massive scale, and that could take a while; the industry has seen some fluctuation, with fast food giants like Burger King, McDonald’s, and Kentucky Fried Chicken jumping on the bandwagon over the last three years, but plant-based meat more recently being called a fad and a flop as companies like Beyond Meat and Impossible Foods see declining sales.
If consumers didn’t have some amount of appetite for meat substitutes, though, we wouldn’t be seeing more of them appear on the market, so there must be something to the plant-based movement. And it’s undeniable that the way we produce meat needs to change.
It seems it’s still too early to say whether plant-based meat is a short-term fad or a long-term fix. But in the meantime, if you’re looking for a meat-like protein to add to your meal while keeping it vegan, Juicy Marbles steak is worth a try.
Mićković believes the product has endless possibilities. “I just adore it with bordelaise sauce, but most commonly, I like to cut it into strips and combine it with unusual flavors like chermoula, maple, fruit, or new spices,” he said. “The question we love to answer in the kitchen is, “Ok, hear me out, what if we…”
Image Credit: Juicy Marbles
One of my favorite childhood summertime memories is being surrounded by fireflies. As the sun set, their shimmering glow would spark up the backyard like delicate fairy lights. The fact that living beings could produce light felt like magic.
But it’s not magic. It’s enzymes.
Enzymes are the catalysts of life. They drive every step of our metabolism, power photosynthesis in plants, propel viruses to replicate—and in certain organisms, trigger bioluminescence so they shine like diamonds.
Unlike manmade catalysts, which help speed up chemical reactions but often require high heat, pressure, or both, enzymes are incredibly gentle. Similar in concept to yeast for baking, enzymes work at life-sustaining temperatures. All you need to do is give them a substrate and working conditions—for example, flour and water—and they’ll perform their magic.
It’s partially why enzymes are incredibly valuable. From brewing beer to manufacturing medications and breaking down pollutants, enzymes are nature’s expert chemists.
What if we can outperform nature?
This week, a new study in Nature tapped into AI to engineer enzymes from scratch. Using deep learning, Dr. David Baker’s team at the University of Washington designed a new enzyme that mimics the firefly’s ability to spark light, but inside human cells in Petri dishes. Overall, the AI “hallucinated” over 7,500 promising enzymes, which were further experimentally tested and optimized. The resulting light was bright enough to see with bare eyes.
Compared to its natural counterpart, the new enzyme was highly effective, requiring just a little bit of substrate to light up the dark. It was also highly specific, meaning that the enzyme only preferred one substrate. In other words, the strategy could design multiple enzymes, each never seen in nature, to simultaneously perform multiple jobs. For example, they could trigger multiple-colored bioluminescence like a disco ball for imaging different biochemical pathways inside cells. One day, the engineered enzymes could also “double-tap” medicine and, say, diagnose a condition and test a treatment at the same time.
“Living organisms are remarkable chemists. Rather than relying on toxic compounds or extreme heat, they use enzymes to break down or build up whatever they need under gentle conditions. New enzymes could put renewable chemicals and biofuels within reach,” said Baker.
Proteins by DesignAt their core, enzymes are just proteins. That’s great news for AI.
Back in 2021, the Baker lab developed an algorithm that accurately predicts protein structures based on the amino acid sequence alone. The team next nailed down functional sites in proteins using trRosetta, an AI architect that imagines and then hones in on hot spots that a drug, protein, or antibody can grab onto—paving the way for medications humans can’t dream up.
So why not use the same strategy to design enzymes and fundamentally rewire nature’s biochemistry?
Enzyme 2.0The team focused on luciferase as their first target—the enzyme that makes fireflies sparkle.
It’s not for childhood nostalgia: luciferase is widely used in biological research. With the right partner substrate, luminescent photons shine through the dark without the need for an external light source, allowing scientists to directly peek inside a cell’s inner workings. So far, scientists have only identified a few types of these valuable enzymes, with many unsuitable for mammalian cells. This makes the enzyme a perfect candidate for AI-driven design, the team said.
They set out with several goals. One, the new light-emitting enzyme should be small and stable in higher temperatures. Two, it needed to play well with cells: when coded as DNA letters and delivered into living human cells, it could hijack the cell’s internal protein-making factory and fold into accurate 3D structures without causing stress or damage to its host. Three, the candidate enzyme had to be selective for its substrate to emit light.
Selecting the substrates was easy: the team focused on two chemicals already useful for imaging. Both are in a family dubbed “luciferin,” but they differ in their exact chemical structure.
Then they ran into problems. A critical factor to train an AI is tons of data. Most previous studies used open-sourced databases such as the Protein Data Bank to screen for possible protein scaffolds—the backbone that makes up a protein. Yet DTZ (diphenylterazine), their first luciferin of choice, had few entries. Even worse, changes to their sequence caused unpredictable results in their ability to emit light.
As a workaround, the team generated their own database of protein scaffolds. Their backbone of choice started from a surrogate protein, dubbed NTF2 (nuclear transport factor 2). It’s a wild bet: NTF2 has nothing to do with bioluminescence, but contained multiple pockets in size and structure feasible for DTZ to bind to—and potentially emit light.
The adoption strategy worked. With a method called “family-wide hallucination,” the team used deep learning to hallucinate over two thousand potential enzyme structures based on NTF2-like protein backbones. The algorithm then optimized the core regions of the binding pocket, while allowing creativity in more flexible regions of the protein.
In the end, the AI hallucinated over 1,600 protein scaffolds, each better suited for DTZ than the original NTF2 protein. Next, with the help of RosettaDesign—a suite of AI and other computational tools for protein design—the team further screened for active sites for DTZ while keeping the scaffold stable. Overall, over 7,600 designs were selected for screening. In a matchmaker’s dream (and a grad student’s nightmare), the designs were encoded into DNA sequences and inserted into bacteria to test their enzymatic strengths.
One winner reigned. Dubbed LuxSit (from the Latin for “let light exist”), it’s compact—smaller than any known luciferases—and incredibly stable, retaining full structure at 95 degrees Celsius (203 Fahrenheit). And it works: when given its substrate, DTZ, the testing apparatus glowed.
The Race for Designer EnzymesWith LuxSit in hand, the team next set out to optimize its ability. Focusing on its binding pocket, they generated a library of mutants in which each amino acid was mutated one at a time to see if these “letter” changes affected its performance.
Spoiler: they did. Screening for the most active enzyme, the team found LuxSit-i, which pumps out 100 more photons every second onto the same area compared to LuxSit. The new enzyme also triumphed over natural luciferases, lighting up cells 40 percent more than naturally-occurring luciferase from the sea pansy—a species that glow on the luminescent beaches on the warm shores of Florida.
Compared to its natural counterparts, LuxSit-i also had an “exquisite” ability to target its substrate molecule, DTZ, with a 50-fold selectivity over another substrate. This means the enzyme played well with other luciferases, allowing researchers to monitor multiple events inside cells simultaneously. In a proof-of-concept the team proved just that, tracking two critical cellular pathways involved in metabolism, cancer, and immune system function using LuxSit-i and another luciferase enzyme. Each enzyme grabbed onto their substrate, emitting a different color of light.
Overall, the study further illustrates the power of AI for altering existing biochemical processes—and potentially designing synthetic life. It’s not the first to hunt for enzymes with additional, or more efficient, abilities. Back in 2018, a team at Princeton engineered a new enzyme by experimentally mutating each “hotspot” amino acid at a time—a tedious, if rewarding attempt. Flash forward and deep learning is, cough, catalyzing the entire design process.
“This breakthrough means that custom enzymes for almost any chemical reaction could, in principle, be designed,” said study author Dr. Andy Hsien-Wei Yeh.
Image Credit: Joshua Woroniecki from Pixabay
There are growing fears around the impact automation could have on jobs, but there’s been much less focus on how it could impact unpaid labor. New research suggests close to half of the time-consuming domestic work people do for free could be automated within a decade.
The prospect of “technological unemployment” has been a central part of the public discourse around AI and robotics ever since an influential 2013 study from the University of Oxford reported that around 47 percent of US employment was at risk of automation.
While there’s been considerable debate about the scale of the problem, it is now widely accepted that emerging technologies could dramatically reshape the world of work in ways unseen since the industrial revolution. More often than not, this is framed in a negative light, with the focus on the economic impact for displaced workers.
But the authors of a new study in PLOS ONE point out that much of the work humans do isn’t related to their employment and is instead dedicated to household chores or caring for relatives, with women carrying the bulk of this burden. And it turns out many of these tasks are just as amenable to automation as our day jobs, with the study predicting that 39 percent of the time currently spent on this kind of work could be automated within a decade
“If it is true that robots are taking our jobs, then it appears that they are also capable of taking out the trash for us,” the authors write. “Considering that people currently spend almost similar amounts of time on unpaid work as they do on paid work, the social and economic implications of this future of unpaid work could be significant.”
The authors reached their conclusions by asking a panel of 65 AI experts from the UK and Japan to predict about how much of the time spent on 17 domestic tasks—such as cooking, doing laundry, and car maintenance—would be automated in the next five to ten years, and how much it would cost users of those technologies.
These experts were pulled in roughly equal numbers from academia, corporate research and development, and business backgrounds and almost evenly split between men and women. To account for varying attitudes towards automation between different cultures, the team drew 29 of these experts from the UK and 36 from Japan.
After getting a smaller group to help them narrow down the tasks to consider, the authors then put their questions to the wider group. The experts who gave the highest and lowest answers were asked to give explanations, and then the entire group was allowed to use these explanations and statistics from the first round to revise their estimates.
Grocery shopping was seen as the most susceptible, with the researchers predicting on average that 59 percent of the time spent on it would be automated within 10 years. In contrast, the hardest to automate was physical childcare at 21 percent. Care work was generally seen as more difficult to automate, with an average score of 29 percent across several tasks, while housework was seen to be much simpler with a score of 44 percent.
Interestingly, despite the researchers instructing the experts to focus purely on the technical feasibility of automating each task, in many cases their explanations suggested they had also considered more human factors. In particular, much of the reasoning for why care-related tasks were less amenable to technological solutions was a lack of societal acceptance.
The authors also broke down the responses based on the demographics of the experts to see how different cultural factors could impact the forecast. They found that male experts from the UK were much more optimistic than their female counterparts, which the authors say fits previous research showing men are typically more optimistic about technology.
However, the situation was reversed when it came to the Japanese experts, which the team suggests could be due to much greater gender disparity in who does housework in Japan. They point to surveys showing that only 52 percent of Japanese men aged 20 to 59 do any domestic work, compared to 88 percent in the UK.
The authors say that discrepancies due to cultural differences show the potential limitations to this kind of forecasting study, but also suggest that more carefully taking these factors into account could help boost their validity.
Regardless of how accurate the results are, though, the study shines a light on an important and much-overlooked aspect of automation. While it is likely to cause significant disruption to the world of work as we know it, it could also free us from much of the domestic drudgery that currently occupies our free time.
Image Credit: Photos Hobby / Unsplash
You can easily picture yourself riding a bicycle across the sky even though that’s not something that can actually happen. You can envision yourself doing something you’ve never done before—like water skiing—and maybe even imagine a better way to do it than anyone else.
Imagination involves creating a mental image of something that is not present for your senses to detect, or even something that isn’t out there in reality somewhere. Imagination is one of the key abilities that make us human. But where did it come from?
I’m a neuroscientist who studies how children acquire imagination. I’m especially interested in the neurological mechanisms of imagination. Once we identify what brain structures and connections are necessary to mentally construct new objects and scenes, scientists like me can look back over the course of evolution to see when these brain areas emerged—and potentially gave birth to the first kinds of imagination.
From Bacteria to MammalsAfter life emerged on Earth around 3.4 billion years ago, organisms gradually became more complex. Around 700 million years ago, neurons organized into simple neural nets that then evolved into the brain and spinal cord around 525 million years ago.
Eventually dinosaurs evolved around 240 million years ago, with mammals emerging a few million years later. While they shared the landscape, dinosaurs were very good at catching and eating small, furry mammals. Dinosaurs were cold-blooded, though, and, like modern cold-blooded reptiles, could only move and hunt effectively during the daytime when it was warm. To avoid predation by dinosaurs, mammals stumbled upon a solution: hide underground during the daytime.
Not much food, though, grows underground. To eat, mammals had to travel above the ground—but the safest time to forage was at night, when dinosaurs were less of a threat. Evolving to be warm-blooded meant mammals could move at night. That solution came with a trade-off, though: Mammals had to eat a lot more food than dinosaurs per unit of weight in order to maintain their high metabolism and to support their constant inner body temperature around 99 degrees Fahrenheit (37 degrees Celsius).
Our mammalian ancestors had to find 10 times more food during their short waking time, and they had to find it in the dark of night. How did they accomplish this task?
To optimize their foraging, mammals developed a new system to efficiently memorize places where they’d found food: linking the part of the brain that records sensory aspects of the landscape—how a place looks or smells—to the part of the brain that controls navigation. They encoded features of the landscape in the neocortex, the outermost layer of the brain. They encoded navigation in the entorhinal cortex. And the whole system was interconnected by the brain structure called the hippocampus. Humans still use this memory system for remembering objects and past events, such as your car and where you parked it.
Groups of neurons in the neocortex encode these memories of objects and past events. Remembering a thing or an episode reactivates the same neurons that initially encoded it. All mammals likely can recall and re-experience previously encoded objects and events by reactivating these groups of neurons. This neocortex-hippocampus-based memory system that evolved 200 million years ago became the first key step toward imagination.
The next building block is the capability to construct a “memory” that hasn’t really happened.
Involuntary Made-Up ‘Memories’The simplest form of imagining new objects and scenes happens in dreams. These vivid, bizarre involuntary fantasies are associated in people with the rapid eye movement (REM) stage of sleep.
Scientists hypothesize that species whose rest includes periods of REM sleep also experience dreams. Marsupial and placental mammals do have REM sleep, but the egg-laying mammal the echidna does not, suggesting that this stage of the sleep cycle evolved after these evolutionary lines diverged 140 million years ago. In fact, recording from specialized neurons in the brain called place cells demonstrated that animals can “dream” of going places they’ve never visited before.
In humans, solutions found during dreaming can help solve problems. There are numerous examples of scientific and engineering solutions spontaneously visualized during sleep.
The neuroscientist Otto Loewi dreamed of an experiment that proved nerve impulses are transmitted chemically. He immediately went to his lab to perform the experiment—later receiving the Nobel Prize for this discovery.
Elias Howe, the inventor of the first sewing machine, claimed that the main innovation, placing the thread hole near the tip of the needle, came to him in a dream.
Dmitri Mendeleev described seeing in a dream “a table where all the elements fell into place as required. Awakening, I immediately wrote it down on a piece of paper.” And that was the periodic table.
These discoveries were enabled by the same mechanism of involuntary imagination first acquired by mammals 140 million years ago.
Imagining on PurposeThe difference between voluntary imagination and involuntary imagination is analogous to the difference between voluntary muscle control and muscle spasm. Voluntary muscle control allows people to deliberately combine muscle movements. Spasm occurs spontaneously and cannot be controlled.
Similarly, voluntary imagination allows people to deliberately combine thoughts. When asked to mentally combine two identical right triangles along their long edges, or hypotenuses, you envision a square. When asked to mentally cut a round pizza by two perpendicular lines, you visualize four identical slices.
This deliberate, responsive and reliable capacity to combine and recombine mental objects is called prefrontal synthesis. It relies on the ability of the prefrontal cortex located at the very front of the brain to control the rest of the neocortex.
When did our species acquire the ability of prefrontal synthesis? Every artifact dated before 70,000 years ago could have been made by a creator who lacked this ability. On the other hand, starting about that time there are various archeological artifacts unambiguously indicating its presence: composite figurative objects, such as lion-man; bone needles with an eye; bows and arrows; musical instruments; constructed dwellings; adorned burials suggesting the beliefs in afterlife, and many more.
Multiple types of archaeological artifacts unambiguously associated with prefrontal synthesis appear simultaneously around 65,000 years ago in multiple geographical locations. This abrupt change in imagination has been characterized by historian Yuval Harari as the “cognitive revolution.” Notably, it approximately coincides with the largest Homo sapiens‘ migration out of Africa.
Genetic analyses suggest that a few individuals acquired this prefrontal synthesis ability and then spread their genes far and wide by eliminating other contemporaneous males with the use of an imagination-enabeled strategy and newly developed weapons.
So it’s been a journey of many millions of years of evolution for our species to become equipped with imagination. Most nonhuman mammals have potential for imagining what doesn’t exist or hasn’t happened involuntarily during REM sleep; only humans can voluntarily conjure new objects and events in our minds using prefrontal synthesis.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Jr Korpa / Unsplash
ARTIFICIAL INTELLIGENCEI Made an AI Clone of Myself
Chloe Xiang | Motherboard“To create my AI clone, Synthesia told me that we would have to clone my voice and body, and it would take a total of a little over two hours to do so. Before the shoot, I was given a schedule of ‘Voice Clone,’ ‘Prep [Hair and Makeup],’ and ‘Video Performance.’ No details beyond that. Entering the studio the day of, I had no idea what to expect, other than that I was like an actress responding to a call sheet, ready to do my best improv.”
SPACECalifornia Company Sets Launch Date for World’s First 3D-Printed Rocket
Passant Rabie | Gizmodo“On Wednesday, Relativity Space announced that it had secured its launch license from the Federal Aviation Administration and is ready to blast its Terran 1 rocket into space. …Terran 1 is a two-stage, 110-foot-tall (33 meters) rocket that’s 85% 3D printed, making it the ‘largest 3D printed object to exist and to attempt orbital flight,’ according to the company. Relativity Space is working towards its goal of making the rocket 95% 3D printed.”
COMPUTINGGoogle’s Improved Quantum Processor Good Enough for Error Correction
John Timmer | Ars Technica“…getting quantum error correction isn’t really the news—they’d managed to get it to work a couple of years ago. Instead, the signs of progress are a bit more subtle. In earlier generations of processors, qubits were error-prone enough that adding more of them to an error-correction scheme caused problems that were larger than the gain in corrections. In this new iteration, adding more qubits and getting the error rate to go down is possible.”
TECHChatGPT-Style Search Represents a 10x Cost Increase for Google, Microsoft
Ron Amadeo | Ars Technica“A ChatGPT-style search engine would involve firing up a huge neural network modeled on the human brain every time you run a search, generating a bunch of text and probably also querying that big search index for factual information. …All that extra processing is going to cost a lot more money. After speaking to Alphabet Chairman John Hennessy (Alphabet is Google’s parent company) and several analysts, Reuters writes that ‘an exchange with AI known as a large language model likely costs 10 times more than a standard keyword search’ and that it could represent ‘several billion dollars of extra costs.’i”
SPACEIngenious Technique Could Make Moon Farming Possible
Kevin Hurler | Gizmodo“The idea is that astronauts can extract nutrients in lunar regolith to create fertilizer for hydroponic farming. These nutrients could be pulled from soil using a processing plant and then dissolved into water, all on the Moon’s surface. The resulting nutrient-rich water can then be pumped into a greenhouse for hydroponic farming, a crucial part of maintaining a long-term human presence on the Moon.”
GOVERNANCEThe US Copyright Office Says You Can’t Copyright Midjourney AI-Generated Images
Richard Lawler | The Verge“A copyright registration granted to the Zarya of the Dawn comic book has been partially canceled, because it included ‘non-human authorship’ that hadn’t been taken into account. …To justify the decision, the Copyright Office cites previous cases where people weren’t able to copyright words or songs that listed ‘non-human spiritual beings’ or the Holy Spirit as the author—as well as the infamous incident where a selfie was taken by a monkey.”
ROBOTICSAlphabet Layoffs Hit Trash-Sorting Robots
Paresh Dave | Wired“Just over a year after graduating from Alphabet’s X moonshot lab, the team that trained over a hundred wheeled, one-armed robots to squeegee cafeteria tables, separate trash and recycling, and yes, open doors, is shutting down as part of budget cuts spreading across the Google parent, a spokeswoman confirmed. …Everyday Robots emerged from the rubble of at least eight robotics acquisitions by Google a decade ago. Google cofounders Larry Page and Sergey Brin expected machine learning would reshape robotics, and Page in particular wanted to develop a consumer-oriented robot, a former employee involved at the time says, speaking anonymously to discuss internal deliberations.”
Image Credit: Abhishek Tiwari / Unsplash
“Look at this,” says Erica’s message. She is poring over the very first images from the brand new James Webb Space Telescope (JWST). It is July 2022, barely a week after those first images from the revolutionary super telescope were released. Twenty-five years in the making, a hundred to a thousand times more powerful than any previous telescope, one of the biggest and most ambitious scientific experiments in human history: it is hard to not speak in superlatives, and it is all true.
The telescope took decades to build, because it had to be made foldable to fit on top of a rocket and be sent into the coldness of space, 1.5 million kilometers from Earth. Here, far from the heat glow of the Earth, JWST can detect the faintest infrared light from the distant universe.
Little did I know that among the pictures is a small red dot that will shake up our understanding of how the first galaxies formed after the Big Bang. After months of analysis, my colleagues and I just published our results in Nature.
Hunting New Kinds of GalaxiesErica and I are on the hunt to discover new types of galaxies. Galaxies that the venerable Hubble Space Telescope had missed, even after decades of surveying the sky.
She and I go back 15 years. We met when she was a first-year student at a Californian liberal arts college and I was a freshly minted PhD straight out of university, just starting my first gig as a researcher in Los Angeles. JWST was only a distant rumor.
Somehow, many years later, our paths crossed again, and now Assistant Professor Erica Nelson of the University of Colorado and I are finding ourselves at the tip of the spear attacking the first data of a very real JWST.
“UFOs,” she calls the new galaxies, and I can read a giant grin between the lines: “Ultra-red Flattened Objects,” because they all look like flying saucers. In the color images they appear very red because all the light is coming out in the infrared, while the galaxies are invisible at wavelengths humans can see.
Infrared is JWST’s superpower, allowing it to spy the most distant galaxies. Ultraviolet and visible light from the first stars and galaxies that formed after the Big Bang is stretched out by the expansion of the universe as it travels towards us, so by the time the light reaches us we see it as infrared light.
Impossibly Early, Impossibly Massive GalaxiesAll of Erica’s galaxies look like saucers, except one. I stare at the little red dot on the screen. That is no UFO. And then it hits me: this is something very different. Much more important.
I run the analysis software on the little pinprick and it spits out two numbers: distance 13.1 billion light years, mass 100 billion stars, and I nearly spit out my coffee. We just discovered the impossible. Impossibly early, impossibly massive galaxies.
At this distance, the light took 13 billion years to reach us, so we are seeing the galaxies at a time when the universe was only 700 million years old, barely 5 percent of its current age of 13.8 billion years. If this is true, this galaxy has formed as many stars as our present-day Milky Way. In record time.
And where there is one, there are more. One day later I had found six.
Images of six candidate massive galaxies, seen 500–800 million years after the Big Bang. Image Credit: NASA / ESA / CSA / I. Labbe, Author providedAstronomy’s Missing Link?Could we have discovered astronomy’s missing link? There has been a long-standing puzzle in galaxy formation. As we look out in space and back in time, we see the “corpses” of fully formed, mature galaxies appear seemingly out of nowhere around 1.5 billion years after the Big Bang.
These galaxies have stopped forming stars. Dead galaxies, we call them, and some astronomers are obsessed with them. The stellar ages of these dead galaxies suggest they must have formed much earlier in the universe, but Hubble has never been able to spot their earlier, living stages.
Early dead galaxies are truly bizarre creatures, packing as many stars as the Milky Way, but in a size 30 times smaller. Imagine an adult, weighing 100 kilos, but standing 6 centimeters tall. Our little red dots are equally bizarre. They look like baby versions of the same galaxies, also weighing in at 100 kilos, with a height of 6cm.
Too Many Stars, Too EarlyThere is a problem, however. These little red dots have too many stars, too early. Stars form out of hydrogen gas, and fundamental cosmological (“Big Bang”) theory makes hard predictions on how much gas is available to form stars.
To produce these galaxies so quickly, you almost need all the gas in the universe to turn into stars at near 100 percent efficiency. And that is very hard, which is the scientific term for impossible. This discovery could transform our understanding of how the earliest galaxies in the universe formed.
The six galaxies and their surroundings in the sky. Image Credit: NASA / ESA / CSA / I. Labbe, Author providedThe implication is that there is different channel, a fast track, that produces monster galaxies very quickly, very efficiently. A fast track for the top one percent.
In a way, each of these candidates can be considered a “black swan.” The confirmation of even one would rule out our current “all swans are white” model of galaxy formation, in which all early galaxies grow slowly and gradually.
Checking the FingerprintsThe first step to solve this mystery is to confirm the distances with spectroscopy, where we put the light of each of these galaxies through a prism and split it into its rainbow-like fingerprint. This will tell us the distance to 0.1 percent accuracy.
It will also tell us what is producing the light, whether it is stars or something else more exotic.
By chance, about a month ago, JWST already targeted one of the six candidate massive galaxies and it turned out to be a distant baby quasar. A quasar is a phenomenon that occurs when gas falls into a supermassive black hole at the center of a galaxy and starts to shine brightly.
This is really exciting on the one hand, because the origin of supermassive black holes in galaxies is not understood either, and finding baby quasars might just hold the key. On the other hand, quasars can outshine their entire host galaxy, so it is impossible to tell how many stars are there and whether the galaxy is really that massive.
Could that be the answer for all of them? Baby quasars everywhere? Probably not, but it will take another year to investigate the remaining galaxies and find out.
One black swan down, five to go.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA / ESA / CSA / I. Labbe, Author provided
From the Axe to the Ryse Recon, Aska A5, or Jetson One, there’s no shortage of electric vertical takeoff and landing (eVTOL) aircraft in the works, ramping up to fill the skies with easy-breezy transit in the who-knows-how-distant future. But despite an across the board wow-factor, what all these vehicles are missing is range. The Aska A5 leads the pack with a 250-mile flight range, but it’s not all-electric; it has lithium-ion battery packs and a gas engine.
Enter a new player: AMSL Aero’s Vertiia. The eVTOL is designed to go up to 250 kilometers (155 miles) on electric batteries and up to 1,000 km (621 miles) with a hydrogen fuel cell power train.
AMSL Aero is an Australian company that was founded in 2017 by a husband and wife team: Andrew Moore, who previously worked as an aeronautical engineer at Yamaha, and Siobhan Lyndon, a former director of people operations at Google. They raised $23 million in Series B funding last year, and just completed the Vertiia’s first successful test flight.
The flight was remote-controlled and tethered, with a prototype aircraft that’s slightly smaller than the planned production version. The Vertiia will have space for five people—four passengers and a pilot—or up to 500 kilograms (1,100 pounds) of cargo. It will have eight sets of motors and propellers on a box wing design. This design reduces wingtip drag and ups aerodynamic efficiency, meaning the Vertiia’s relatively small size—its wingspan is 23 feet—shouldn’t stop it from operating smoothly in an urban environment with compact spaces.
The Vertiia prototype that completed a test flight this month. Image Credit: AMSL AeroThe box wing design will provide an additional advantage once the aircraft starts running on hydrogen: AMSL plans to store the hydrogen fuel in tanks that connect the ends of the wings. Hydrogen’s energy density is better than that of batteries, but worse than jet fuel; planes will need about four times as much liquid hydrogen as they do jet fuel to fly the same distance.
That’s a lot more weight to carry, and the challenges don’t end there. Liquid hydrogen also needs to be stored under pressure and at very cold temperatures, meaning the tanks that hold it will be more complex than conventional jet fuel tanks.
But Moore believes that, in a sense, all of this can work to the Vertiia’s advantage. “One of the great advantages of the box wing is that if you do it right, what you put in the wingtips actually helps you aerodynamically,” he said. “So if you’re putting underslung tanks under your wings, that’ll give you a significant drag penalty—whereas for us, it can actually help reduce drag.”
Moore and Lyndon are purpose-building the Vertiia for the aeromedical, emergency services, and passenger markets, describing it as “a safe aircraft that could ease the burden of traffic congestion, provide greater choice over where people can work and live, improve medical access and patient outcomes for Australia’s rural and remote communities and most importantly, assist in de-carbonizing the transport industry.”
They’re not the only ones hoping to use hydrogen as an aircraft fuel. Rolls-Royce tested its hydrogen-powered jet engine late last year, and Airbus is working on a “megawatt-class” hydrogen fuel cell jet engine. Competitor Lilium isn’t aiming to use hydrogen fuel, but is targeting a similarly long range as the Vertiia with its unique fixed-wing eVTOL design.
AMSL is aiming for the Vertiia to enter service in 2026. Given the regulatory approvals, test flights, and production steps left to complete, that seems like an ambitious timeline, but the team feel that once they see success in their home country, bringing their product to the rest of the world will be an easy next step.
“Vertiia…was developed for the harsh long-distance conditions in Australia,” said Lyndon. “If it can work in Australia, it can work anywhere.”
Image Credit: AMSL Aero
3D printing has been slowly but surely ramping up as a viable construction technology, with communities of 3D printed homes being built in California, Virginia, Texas, and Mexico, among others. Now a new development on the other side of the Atlantic is joining this list.
Last week 14Trees, a joint venture between Swiss sustainable construction company Holcim and British International Investment, announced completion of the first 10 units of a 3D printed housing project in Kilifi, Kenya. The community is called Mvule Gardens, and it will eventually consist of 52 single-family homes.
The first homes to be completed include six three-bedroom houses that are 836 square feet each and four two-bedrooms measuring 616 square feet. Printing started in October 2022, with the goal of putting up the walls of one house each week for 10 weeks. According to Holcim, printing the walls of the two-bedroom homes took just 18 hours.
Aerial view of the Mvule Gardens development in progress, with 3D printed walls of the new homes. Image Credit: HolcimThe printer being used for the project is a BOD2 from Danish company COBOD. The company says the BOD2 is the fastest 3D construction printer on the market, laying down one meter’s worth of material per second. The gantry-style printer moves between three axes on a metal frame, and requires two human operators to run it. The BOD2 was used in 2020 to print an apartment building in Germany, and in 2021 for a school in Malawi.
Kilifi is in south-east Kenya, on the coast about 35 miles northeast of Mombasa. Prices for the two-bedroom houses at Mvule Gardens start at 3,610,000 Kenyan shillings, about $28,620. While this would be quite affordable for the average American, it’s less so for the average Kenyan.
However, 14Trees says one of its main goals is to continue lowering construction costs with each phase of the project, ultimately offering homes for 20 percent less than standard houses. They’re also planning to let future homeowners design their own floor plans, taking advantage of the customizable and modular nature of 3D printing construction.
As you’ve probably heard, or perhaps experienced first-hand, there’s a massive housing shortage not just in the US, but all over the world, with millions more people in need of homes than there are homes available. The best way to solve this crisis is to build, but we have to do so in a way that’s smart, sustainable, and affordable—not to mention fast. According to 14Trees, there’s a deficit of nearly two million homes in Kenya alone.
There’s some debate about whether 3D printing is truly a viable and scalable solution to the housing shortage. Critics point out that the cost savings likely won’t be as significant as all the hype implies, nor will the eco-friendliness, since the homes use cement and cement is a major source of carbon emissions. Also, 3D printing is most practical for single-family homes, and less applicable for the dense, multi-story housing that will go furthest in closing the supply gap.
Will 3D printed homes really become the “Teslas of housing”? That remains to be seen, but as companies and governments work towards solving the housing crisis in the US and across the world, 3D printed communities are better than the alternative, which in many cases is nothing.
The first 10 Mvule Gardens homes are now being roofed and painted, and will be turned over to their new owners upon completion as the next 10 homes get underway.
Image Credit: Holcim
Death comes for us all. Aging, maybe not.
It sounds preposterous, but plenty of animals—from the lowly jellyfish to naked mole rats and giant tortoises—show negligible signs of aging. Some animals are even “biologically immortal,” escaping the gradual deterioration of physiological functions as the clock ticks on.
Why?
One theory, the geroscience hypothesis, proposes that aging is due to a myriad of molecular changes that accumulate over time. Dubbed the hallmarks of aging, these “red flags” range from genetic mutations to chronic inflammation. As we age, the genome gradually breaks down. Telomeres, the DNA “caps” that protect chromosomes, waste away. The cell’s energy factory, the mitochondria, slowly disintegrate.
But it’s not all bad news: by hunting down contributors to aging, we can develop more sophisticated methods to combat these molecular processes. In turn, the treatments can also potentially reverse aging at the molecular level.
This week, a new analysis from one of the largest anti-aging studies to date found that cutting calories by 25 percent for two years slowed the pace of aging. Called CALERIE, or the Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy trial, the study was the first randomized controlled study—a gold standard—to examine one of the most prominent theories in longevity: that cutting calories without sacrificing nutrients promotes healthy longevity.
Initial results from the trial found that the diet rewired multiple metabolic and immune responses to promote health. The new results went further, asking: can a two-year modest cut in calories alter your biological age?
Spoiler alert: yes and no. Although reducing calories didn’t change the volunteers’ biological age compared to people who ate to their hearts’ desire, it slowed the rate of aging—that is, how rapidly a person ages based on biological measures.
Don’t brush those results off. Even slowing aging by just two percent corresponds to a 10-15 percent reduction in mortality risk, which is similar to quitting smoking, the authors said.
“Our study found evidence that calorie restriction slowed the pace of aging in humans” said study author Dr. Calen Ryan at Columbia’s Butler Aging Center.
Tick Tock Goes the ClockWe all know people who look and behave younger—or older—than their age. Scientists have long known that your chronological age—that is, the years you count on your birthday—is often different than your biological age. Recent studies show that peoples’ biological age is more predictive of their chances of getting age-related diseases, such as hypertension, diabetes, heart disease, cancer, and dementia.
The question is, how do you measure your biological age?
One popular solution is using DNA methylation (DNAm) clocks. As we age, parts of our DNA become dotted with a chemical group that silences the gene, in a process called methylation. A decade ago, scientists found that DNA methylation can closely predict a person’s chronological age. These first-generation clocks used machine learning to compare samples spanning from teenagers to the elderly to extract patterns from DNA methylation as a proxy for aging.
But the results weren’t helpful. The clocks struggled to predict age-related diseases or the risk of death, making them inept for early intervention, the authors explained.
Flash forward five years, and second-generation DNAm clocks rocked the geroscience field. Rather than chronological age, these clocks aimed to better quantify biological age by analyzing mortality risk. For example, the PhenoAge clock, developed by a team at the University of California, Los Angeles (UCLA) added clinical biomarkers such as white blood cell counts—a reflection of immune system health—into the DNA methylation aging model.
GrimAge, another DNAm clock developed at UCLA, also honed in on age-related diseases. Using machine learning, the clock was trained on DNA methylation patterns specifically associated with smoking, cardiovascular disease, and cancer—essentially focusing the algorithm on hunting down age-related diseases. Compared to first-generation clocks, both PhenoAge and GrimAge were far more powerful predictors of mortality and age-related diseases.
But they weren’t perfect. Although they had improved potential for testing aging interventions, they struggled with reliability.
Enter the third wave of DNAm clocks. If PhenoAge and GrimAge were odometers—capturing the biological aging already experienced—these clocks are speedometers. DunedinPACE (Pace of Aging Computed from the Epigenome) is a popular one: it captures the pace of aging rather than age itself. Developed in a longitudinal study in New Zealand, the algorithm uses an exceptionally long list of health measures to capture each person’s health deterioration as they age.
CALERIE ConundrumThe new analysis used all three clocks—PhenoAge, GrimAge, and DunedinPACE—to see if reducing calorie intake delayed biological aging.
The data came from blood samples of 200 volunteers in the CALERIE Phase 2 trial. The multi-center randomized controlled study was the largest yet examining caloric restriction as an anti-aging intervention. The volunteers were a diverse bunch, ranging from 21 to 50 years old and comprised of different genders and ethnicities.
The control group had it easy: they could go about their daily eating habits. Those in the restriction arm cut a quarter of their daily calorie intake and attended behavioral counseling sessions to help sustain their diet.
Perhaps unsurprisingly, not everyone stuck to their regime—the average calorie cut was roughly 12 percent, about a muffin every day. Even so, people on the restricted diet decreased their rate of aging by two to three percent as measured with DunedinPACE. It doesn’t sound like much, but according to one estimate it cuts mortality risk up to 15 percent, boosting cardiovascular and metabolic health while slowing age-related bodily changes.
Then came the shocker: restricting calories didn’t impact peoples’ biological age, as measured with both PhenoAge and GrimAge clocks. There were multiple reasons: for one, the trial lasted for only two years, and these clocks measure aging factors up to a specific point in time. In other words, the intervention may be too brief to change a lifetime of dietary habits and history, which are etched into the DNA epigenome. The team was also unable to follow up with the participants beyond the two-year mark, when the study ended, which may have revealed longer-term health benefits.
“This is an interesting study…it suggests that measures of aging from DNA may slow, but does not report on any physical or functional changes in aging,” said Dr. Duane Mellor at Aston Medical School in Birmingham, U.K., who was not involved in the study.
Feast or Fast?To the authors, the study is just the first step in hunting down why people age—and how we can potentially slow or reverse the process.
“The purpose of DNAm analysis in CALERIE was to evaluate intervention effects at the molecular level, where aging processes are posited to originate,” said the authors.
A follow-up trial is in the works to see if cutting calories has long-term effects on healthy aging. But perhaps more impactful is the use of DNAm clocks to assess aging interventions. Scientists have long identified multiple therapies that could improve healthspan in animal models. But because human aging takes decades to cause diseases, it’s difficult to assess the efficacy of potential treatments.
“Humans live a long time,” said study author Dr. Daniel Belsky, “so it isn’t practical to follow them until we see differences in aging-related disease or survival. Instead, we rely on biomarkers developed to measure the pace and progress of biological aging over the duration of the study.”
For now, the study showed that DNAm clocks can efficiently tag-team with anti-aging interventions to assess their efficacy.
“Our findings are important because they provide evidence from a randomized trial that slowing human aging may be possible. They also give us a sense of the kinds of effects we might look for in trials of interventions that could appeal to more people, like intermittent fasting or time-restricted eating,” said Ryan.
Image Credit: fancycrave1 from Pixabay
It’s been impossible to miss the latest collision of AI and mainstream culture.
The cycle started in earnest last year with the release of OpenAI’s DALL-E 2, a machine learning algorithm that concocts photorealistic images from text prompts. The hype ramped up even further with the company’s release of ChatGPT in November. But things really went off the rails last week, when Microsoft—a big investor in OpenAI with nearly unfettered access to its algorithms—blended a specialized version of ChatGPT into its Bing search engine in the form of a chatbot.
While the capability of these algorithms is undoubtedly advancing quickly, it seems recent leaps have been as much about what they can do as the fact average people can now access them.
Now, another AI is getting a mainstream release: Sony just announced its superhuman AI driver, GT Sophy, is officially joining Gran Turismo.
As of today, any player who downloads the latest update can race GT Sophy in the new “Gran Turismo Sophy Race Together” mode. The mode, which Sony says is a special event, will be available through the end of March. Players can compete with Sophy over four races, each more difficult than the last, or go head-to-head with the same car and settings to see how fast the AI can go.
GT Sophy first made headlines last year when the team published a paper in Nature outlining how the algorithm beat top Gran Turismo players. By pushing the envelope right to its physical limits and also observing etiquette—reckless drivers are penalized—the algorithm outpaced human players by seconds in a competition usually decided by milliseconds.
The plan was always to incorporate GT Sophy into the game. But the latest announcement should be viewed as a first step. Much like other recent AI deployments, Sony will seek feedback from players to improve the AI further.
While the recent round of AI releases has been in the category of generative AI—models trained to produce text or images after poring over billions of examples scraped from the internet—GT Sophy was trained a little differently.
Like other game-playing algorithms, it uses deep reinforcement learning, where the algorithm is fed the rules of the game and conditions of its environment, then plays millions of rounds, scoring its own performance and making iterative improvements. At least 10 times a second, GT Sophy takes stock of its position relative to other cars on the track and the various forces acting on the car and makes split-second decisions based on the data.
The approach has yielded algorithms that beat humans at games like Go, Starcraft, Stratego, and Diplomacy.
One way these algorithms beat humans is by developing surprising strategies. DeepMind’s AlphaGo famously outfoxed world champion Go player Lee Sedol with a move no human would make. At first dubbed a mistake, it later proved a turning point in the match.
Kazunori Yamauchi, the creator and CEO of Gran Turismo, said last year that GT Sophy gains a speed advantage thanks, in part, to an aggressive strategy driving through curves. Whereas drivers often brake into a curve and accelerate out of it, GT Sophy brakes during the curve, shifting the load from two to three tires. “We notice that, actually, top drivers such as [Formula One champions] Lewis Hamilton or Max Verstappen actually are doing that, using three tires, going fast in and fast out, all these things that we thought were unique to GT Sophy,” he said.
By racing the AI, Sony hopes players can improve. The top players in chess, for example, aren’t computer or human alone, but the two playing together.
“From the beginning, Gran Turismo Sophy was always about more than just being superhuman; we aspired to create an AI agent that would enhance the experience of players of all levels, and to make this experience available to everyone,” said Michael Spranger, Chief Operating Officer, Sony AI.
Algorithms like GT Sophy may eventually have real-world applications in robotics and self-driving cars. But for now, it’ll be more about the fun of getting schooled by an AI driver.
Image Credit: Sony
One of the biggest challenges for humanity as we move further out into the solar system will be learning to “live off the land” rather than lugging materials with us. Blue Origin now says it’s made major progress in that direction by making solar panels out of moon dust.
Establishing a more permanent human presence beyond Earth’s orbit will require huge amounts of material, both to build infrastructure and provide life support for astronauts. Given the enormous cost of space launches, using Earth-bound resources for this is likely to be unsustainable.
That’s led to a growing focus on “in-situ resource utilization” (ISRU), which refers to making use of materials found in space or on other celestial bodies to do things like build shelters, generate oxygen, or provide water. One key challenge is generating enough electricity to support long-term settlements without having to ship bulky power equipment from Earth.
Blue Origin, the space technology company founded by Jeff Bezos, says it’s closer to solving this problem after demonstrating that it can make solar cells out of simulated moon dust. The company’s approach, which it dubs “Blue Alchemist,” uses a process known as “molten regolith electrolysis” to generate all of the key ingredients needed for a working solar panel.
“To make long-term presence on the moon viable, we need abundant electrical power,” the company said in a blog post. “Our approach, Blue Alchemist, can scale indefinitely, eliminating power as a constraint anywhere on the moon.”
The idea isn’t particularly new. The fine dust found on the surface of the moon, known as regolith, contains all of the key ingredients required for making solar panels, including silicon, iron, magnesium, and aluminum.
But moon dust isn’t easy to come by, so to develop their approach the researchers first had to make their own. They created a simulated lunar soil that is chemically and mineralogically the same as the real thing, and even accounts for the variable size of grains.
They then used molten regolith electrolysis, which is an established process, to extract the key ingredients they were interested in. This involves first melting the lunar soil by heating it to above 1,600 degrees Celsius (2,912 degrees Fahrenheit) and then sticking a probe into it that passes a current through the molten mass.
This causes the iron to separate out first, followed by silicon and then aluminum. Because most of these metals are found as oxides in the regolith, it also creates oxygen as a byproduct, which could be used for both astronaut life support or to help power rockets.
Crucially, Blue Origin’s approach produces silicon with 99.99 percent purity, which is critical if it is to be used in solar panels. Most interestingly though, they’ve found a way to use the byproducts of the molten regolith electrolysis process to create glass covers to protect the solar cells from the harsh lunar environment.
The blog announcing the news revealed that the company has been able to produce solar cells this way since 2021. And they aren’t the only ones—space manufacturing company Lunar Resources told The Verge that they’ve been doing the same for several years now.
But while proving that the concept works using simulated moon dust on Earth is an impressive step, actually doing it in space presents a lot of other challenges. One of the biggest is simply getting the required equipment there in the first place. Lunar Resources chief technology officer Alex Ignatiev told The Verge that the reactor they use to heat the regolith weighs about a ton.
That’s still likely to be much more weight-efficient than shipping hundreds of solar panels from Earth, though. So while it may take some time to get the idea off the ground, this could be a major step towards enabling a more sustainable human presence on the lunar surface.
Image Credit: NASA
ARTIFICIAL INTELLIGENCEAI Is Dreaming Up Drugs That No One Has Ever Seen. Now We’ve Got to See if They Work.
Will Douglas Heaven | MIT Technology Review“There are now hundreds of startups exploring the use of machine learning in the pharmaceutical industry, says Nathan Benaich at Air Street Capital, a VC firm that invests in biotech and life sciences companies: ‘Early signs were exciting enough to attract big money.’ Today, on average, it takes more than 10 years and billions of dollars to develop a new drug. The vision is to use AI to make drug discovery faster and cheaper.”
AUTOMATIONThe US Air Force Successfully Tested This AI-Controlled Jet Fighter
Jonathan Gitlin | Ars Technica“An autonomous jet fighter has now completed 17 hours of flight testing, including advanced fighter maneuvers and beyond-visual-range engagements, according to the United States Air Force. …’We conducted multiple sorties [takeoffs and landings] with numerous test points performed on each sortie to test the algorithms under varying starting conditions, against various simulated adversaries, and with simulated weapons capabilities,’ said Air Force Lt. Col. Ryan ‘Hal’ Hefron, the DARPA program manager for ACE.”
ARTIFICIAL INTELLIGENCEIntroducing the AI Mirror Test, Which Very Smart People Keep Failing
James Vincent | The Verge“In behavioral psychology, the mirror test is designed to discover animals’ capacity for self-awareness. There are a few variations of the test, but the essence is always the same: do animals recognize themselves in the mirror or think it’s another being altogether? Right now, humanity is being presented with its own mirror test thanks to the expanding capabilities of AI—and a lot of otherwise smart people are failing it.”
ENVIRONMENTMIT Team Makes a Case for Direct Carbon Capture From Seawater, Not Air
Loz Blain | New Atlas“As atmospheric carbon concentrations rise, carbon dioxide begins to dissolve into seawater. The ocean currently soaks up some 30-40% of all humanity’s annual carbon emissions, and maintains a constant free exchange with the air. Suck the carbon out of the seawater, and it’ll suck more out of the air to re-balance the concentrations. Best of all, the concentration of carbon dioxide in seawater is more than 100 times greater than in air.”
3D PRINTINGThis Startup Can 3D Print a Battery Into Any Shape You Want
Adele Peters | Fast Company“The technique, which prints using thin layers of powder, can change what the batteries look like—imagine an e-bike battery that curves to fit the frame of a bike, or a cellphone battery that’s shaped to fill every gap around the circuit board, making the phone last longer before it needs another charge. But 3D printing also enables what’s often called the holy grail of the industry: Solid-state batteries.”
VIRTUAL REALITYWelcome to the Oldest Part of the Metaverse
John-Clark Levin | MIT Technology Review“Today’s headlines treat the metaverse as a hazy dream yet to be built, but if it’s defined as a network of virtual worlds we can inhabit, its oldest extant corner has been already running for 25 years. It’s a medieval fantasy kingdom created for the online role-playing game Ultima Online—and it has already endured a quarter-century of market competition, economic turmoil, and political strife. So what can this game and its players tell us about creating the virtual worlds of the future?”
SCIENCECan Black Holes Really Cause Dark Energy?
Ethan Siegel | Big Think“One of the greatest mysteries in the universe is that of the accelerated expansion of the cosmos, often described as an unknown form of energy dubbed ‘dark energy.’ While many potential explanations have been offered for why dark energy exists, no one has yet been able to calculate its value, or offer a compelling reason for why it possesses the value it does. In a brand new study put forth in February of 2023, a team of scientists put forth the idea, backed by some very suggestive evidence, that black holes might be the culprit. How does the idea stack up?”
CRYPTOCURRENCYConfusion Spirals in Crypto as the US Cracks Down
Joel Khalili | Wired“[Last] weekend, The Wall Street Journal reported that the agency intends to sue crypto firm Paxos for issuing BUSD, a stablecoin developed in partnership with the world’s largest crypto exchange, Binance. …The concern is that a ruling against the issuing or use of BUSD will set a precedent that could be applied to all stablecoins, striking down a crucial piece of infrastructure in many crypto markets. ‘If the supply suddenly dried up, the crypto economy would collapse,’ says economist Frances Coppola, who previously worked for HSBC and other banks.”
BIOTECHFor the First Time, Genetically Modified Trees Have Been Planted in a US Forest
Gabriel Popkin | The New York Times“On Monday, in a low-lying tract of southern Georgia’s pine belt, a half-dozen workers planted row upon row of twig-like poplar trees. These weren’t just any trees, though: Some of the seedlings being nestled into the soggy soil had been genetically engineered to grow wood at turbocharged rates while slurping up carbon dioxide from the air. …Living Carbon, a San Francisco-based biotechnology company that produced the poplars, intends for its trees to be a large-scale solution to climate change.“
Image Credit: Vimal S / Unsplash
Sailing cargo ships are making a genuine comeback.
Japanese bulk carrier MOL is operating a wind-assisted ship. American food giant Cargill is working with Olympic sailor Ben Ainslie to deploy WindWings on its routes. Swedish shipping company Wallenius is aiming for Oceanbird to cut emissions by up to 90 percent. The French startup Zephyr & Borée has built the Canopée, which will transport parts of European Space Agency’s Ariane 6 rocket this year.
I researched the decarbonization of the shipping industry. While doing fieldwork aboard the Avontuur, a wind-propelled cargo ship, I even got stuck at sea for five months (because of the pandemic, not because the winds failed).
Sailing Towards Zero EmissionsLike every other sector, the shipping industry needs to decarbonize in line with the Paris Agreement, but its emissions continue to grow. In 2018 the International Maritime Organization (IMO) set a first-ever target of halving shipping emissions between 2008 and 2050.
It was an important, but inadequate, first step. Climate Action Tracker calculates that halving emissions is not nearly enough to keep global warming below 1.5℃.
And yet the scientific consensus is that 1.5℃ is the real upper limit we can risk. Beyond that, dangerous tipping points could spell even more frequent disasters.
Luckily, the IMO will revise its strategy this July. I and many others expect far more ambition—because zero shipping emissions by 2050 is a necessity to keep the 1.5℃ limit credible. That gives us less than three decades to clean up an industry whose ships have an average life of 25 years. The 2050 timeline conceals that our carbon budget will likely run out far more quickly—requiring urgent action for all sectors, including shipping.
Research has confirmed the potential of wind propulsion. The maths is simple. Shipping accounts for one billion tons of carbon dioxide a year, almost three percent of global greenhouse gas emissions. If wind propulsion saves fossil fuels today, the dwindling carbon budget stretches a little further. This, in turn, buys more time to develop alternative fuels, which most ships will need to some extent. Once these fuels are widely available, we’ll need less of them because the wind can provide anything from 10 percent to 90 percent of the power a ship needs.
Some commentators aren’t easily convinced, but I found most objections to wind-propelled shipping are based on four myths that can be easily debunked.
Myth 1: Wind ships are a thing of the past, for good reason
Wind ships may remind us of 19th-century tea clippers and, worse, of the slave trade and colonial exploitation. But returning to wind propulsion doesn’t mean going back in time.
New wind-powered ships use a blend of new and old technology to harness the wind where it is most common: at sea. This reduces the need for fossil fuels and for new alternative fuels that will require investment and space for new landside infrastructure, both to generate electricity and to transform this power into fuel.
Even if research into sailing cargo ships all but stopped in the late 19th century, engineering, materials science, yacht racing, and aerospace design have yielded major innovations that are being used for cargo ships.
Myth 2: The wind is unreliable, so ships won’t arrive on time
The wind may seem fickle when standing on the beach. But at sea the trade winds that powered globalization have remained stable. Indeed, the most common trade routes are still well-served by the prevailing winds.
Weather forecasting has also improved massively since the last days of sail. And weather routing software helps find the best course to take better than anyone could in the 19th century.
While the wind may not be as predictable as a steady flow of heavy fuel oil, technological advances have taken a lot of uncertainty out of sailing. The wind is also free and unaffected by fluctuating oil prices.
Myth 3: Sails cannot work on all types of ships
It’s true not all types of ships would work with sails, rotors, or kites mounted on their decks. This can be due to the type of ship, as the largest container ships can’t easily accommodate sails, for example. It can also be because of where or how vessels operate; the windless waters of the doldrums and tight ferry schedules do pose challenges.
However, the argument that wind propulsion isn’t viable because some ships can’t use it is like claiming that commuting by bike is not a realistic option because not everyone can do so.
Meanwhile, the race between Veer Voyage and Windcoop to build the first wind-powered container ship is on. So perhaps such ships can use sails after all.
Myth 4: If it makes so much sense, we’d already be doing it
The 1970s oil crisis drove an upswing of interest in wind propulsion. Conferences in Delft (1980) and Manila (1985) heralded a new dawn for wind ships. But as oil prices dropped, interest waned.
Wind has had a hard time competing with cheap heavy fuel oil—the toxic sludge that refineries have no other use for. Wind propulsion has remained a niche part of the sector because shipping companies don’t have to pay the real environmental and societal costs of burning fossil fuels.
But a global carbon price is likely to be applied soon to international shipping (the European Union’s Emissions Trading Scheme already includes shipping). This creates a financial incentive for non-polluting means of propulsion.
What Are We Waiting For?The added complexity of using wind propulsion and weather routing software is a small trade-off to decarbonize shipping.
The International Windship Association reports that more than 20 commercial cargo ships already use “wind-assist” technologies that are retrofitted on existing vessels. The first purpose-built modern sailing cargo ship, Canopée, will start operations this year.
Canopée has been launched and wingsails will soon be installed before her first journey transporting parts of the Ariane 6 rocket launcher. Image Credit: Kapitel/Wikimedia Commons, CC BY-SAWhile shipping is a conservative industry, with few companies willing to be first movers, many more wind-propelled vessels will be launched in the next years.
For shipping companies, the biggest risk now isn’t making a daring investment; it’s not investing in a sustainable future at all.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Concept image from Oceanbird, CC BY-SA
Late last year, a Virginia-based startup called Contraline started the initial clinical trial of its male birth control, a hydrogel that’s injected directly into mens’ vas deferens (the tube that carries sperm from the testes to the urethra) to block sperm from passing. Trial participants will be monitored for the next three years, with doctors keeping an eye on the gel’s effectiveness and safety. Though the method seems promising, it will require men to both plan ahead and undergo a slightly invasive procedure (if getting a single shot can be considered invasive).
A new method that was successfully tested in mice is quicker, simpler, and could eventually be used in humans. It comes in the form of a pill that men would take 30 minutes to an hour before sex. Its effects are strongest for the ensuing three hours, and it fully wears off within 24 hours.
Described in a paper published this week in Nature Communications, the pill works by inhibiting the function of a protein called soluble adenylyl cyclase (sAC), which is essential for sperm motility and maturation. Dr. Melanie Balbach, a postdoc at Cornell, discovered that mice that were given a drug to inactivate sAC produced sperm that couldn’t propel themselves forward.
A small percentage of men lack the gene that encodes for sAC, and these men are infertile but otherwise healthy. Knowing there were no negative side effects of inhibiting sAC allowed the team to move forward with investigating it as a contraceptive option; since it’s non-hormonal, there’s no risk of it impacting testosterone or causing male hormone deficiencies.
The team partnered with drug discovery scientists to develop a sAC inhibitor called TDI-11861. Male mice who were given a dose of the drug didn’t impregnate any females even after 52 mating attempts. In the control group, meanwhile, about a third of the females became pregnant from male mice not given the drug.
A major advantage of this method, if it were to translate to humans, is how quickly it works and how quickly it stops working. “Our inhibitor works within 30 minutes to an hour,” Dr. Balbach said. “Every other experimental hormonal or nonhormonal male contraceptive takes weeks to bring sperm count down or render them unable to fertilize eggs.” It also takes weeks to reverse their effects, while sAC inhibitors wear off within hours. Men could take it only when needed, making day-to-day decisions about their fertility rather than committing to a multi-month or multi-year option.
But getting men to share the burden of birth control will likely take more than having options available, even if those options are as straightforward as the sAC inhibitor seems. The predominant line of thinking is that birth control is womens’ responsibility, never mind the severe side effects that come with many forms of female birth control. What might it take for men to not only get used to the idea of male birth control, but willingly partake in it?
Allan Pacey, a professor of andrology at the University of Sheffield who was not involved in the study, thinks the sAC inhibitor could be a gateway solution. “There is a pressing need for an effective, reversible, oral contraceptive for men and although many different approaches have been tested over the years, none has yet reached the market,” he said. “If the trials on mice can be replicated in humans with the same degree of efficacy, then this could well be the male contraceptive approach we have been looking for.”
Image Credit: Mohamed Hassan from Pixabay
Construction is a major carbon emitter. The manufacture of cement alone accounts for eight percent of the world’s emissions. But humanity certainly isn’t about to stop building things—in fact, fixing the housing shortage should be near the top of our list of problems to solve. So we need to find more sustainable ways to build, and if they can be cheaper to boot, even better.
Companies are working on all kinds of solutions in the sustainable construction space, from 3D printed homes to carbon-negative concrete. A new potential solution is joining their ranks, and a recent infusion of funding indicates this unique idea could have a lot of promise.
Plantd is a startup that makes engineered building materials out of grass. That’s right—grass! The fastest-growing perennial grass on Earth, according to the company.
Based in Durham, North Carolina, Plantd closed its Series A funding round to the tune of $10 million in January. Two of its three cofounders are former SpaceX engineers.
What’s Great About GrassLike trees, grass captures and stores carbon as it grows, and some varieties can even capture more carbon than trees do. This is mostly because of their growth speed. Think about it: a tree takes at least 10 years—if not 20 or more—to reach a mass large enough to be used for lumber, whereas grass can be harvested multiple times in one season (though there is at least one startup out there that’s engineering trees that grow faster and capture more carbon).
After testing several different types of grass and other raw materials, Plantd settled on a perennial (meaning it grows back every year and doesn’t need to be re-planted) long grass that can grow 20 to 30 feet and absorb up to 30 tons of carbon in a year.
Though grass is obviously softer than wood, it contains a similar cellulose fiber that can be broken down then reconstituted and engineered in such a way that the final product is even stronger than wood (check out this video that made the rounds on LinkedIn last year: a regular wood panel and a Plantd panel are subjected to a sledgehammer, and just one of the two withstands the test).
Image Credit: PlantdPlantd makes structural building panels for wall sheathing, roof decking, and subflooring, and they say their product outcompetes wood on every metric: it’s stronger, cheaper, lighter, more moisture-resistant, and captures more carbon—all for the same cost as wood. The panels are meant to be a replacement for a plywood-like material called traditional oriented strand board, or OSB. Custom-built machinery uses heat and pressure to press shredded grass into panels, with a standard four-by-eight-foot panel using about 50 pounds of grass.
The DifferencePlantd’s CEO and co-founder Josh Dorfman told Forbesthat the panels will allow contractors to build homes using fewer two-by-fours. “Our panels are strong enough to maintain more of the structural integrity of a home compared to the conventional wood-based panels used today,” he said. “Instead of using two two-by-fours per panel when constructing a wall, a builder will only have to use one. This leads to clear cost savings, and the savvy contractor will also recognize that fewer two-by-fours means fewer thermal gaps in a building’s envelope where air can escape.” That means better energy efficiency, and ultimately lower electric and gas bills for homeowners.
Plantd is partnering with a local farmer near Durham to start growing several acres of its grass this year. Dorfman expects that scaling up and finding more farmers won’t be difficult, because many farmers in North Carolina who have historically grown tobacco are looking to replace it with a more sustainable, better-paying crop.
For processing the grass and turning it into super-strong panels, Plantd is developing its production technology in-house, and says it will be automated, modular, electric, and low-emissions, with 80 percent of the carbon that enters their factory in the form of grass getting locked away in the building materials that leave it. They estimate they’ll be able to produce the same amount of material that’s produced from wood using nine times less land (15,000 acres versus 140,000 acres).
Going ForwardDorfman got the idea for his company while trying to build a sustainable furniture business during the pandemic. As supply chains crumbled, the cost of building materials went up even as their quality went down. “That frustration led to my interest in launching a materials company using alternative biomass to trees,” Dorfman said.He and his cofounders launched Plantd in the spring of 2021.
The company plans to use its Series A funding to establish its agricultural supply chain and build its first pieces of manufacturing equipment. They hope to make their panels a standard in the industry, and their ultimate vision is to build the “factory of the future” to enable mass adoption of the product. They’ll have their work cut out for them; according to the company’s website, the US market for building panels is $26 billion, and the global market for engineered wood products is worth $280 billion.
“Our value proposition is about affordability, durability, and sustainability,” Dorfman said. “We believe the future is going to be abundant. With Plantd, the way to solve climate change is to build more, not less, because every new home and building is an opportunity to lock away atmospheric carbon.”
Image Credit: Plantd
With a massive beak, googly eyes, a rotund body, and a disproportionately small feathered tail, the dodo is iconic for all the wrong reasons. The flightless bird vanished in the seventeenth century, and has since been the poster child for human-caused extinction.
But what if we can bring the iconic bird back?
Last week, a biotech company based in Dallas, Texas called Colossal Biosciences announced an audacious plan to “de-extinct” the dodo. Founded by Harvard geneticist George Church and tech entrepreneur Ben Lamm in 2021, the company has ongoing projects to recreate the woolly mammoth and the thylacine, a Tasmanian tiger.
The dodo has now joined this lineup. Similar to previous projects, resurrecting the iconic bird requires huge advances in genetic engineering, stem cell biology, artificial wombs, and animal husbandry. Whether they can fit into a whole new world—300 years later—is hotly debated. Even if the technology works, the resulting “hacked” species would raise a big philosophical question: at what point does resembling a dodo genetically equate to resurrecting the species?
But for Colossal Biosciences, the challenge is worth it.
“A goal here is to create an animal that can be physically and psychologically well in the environment in which it lives,” said Dr. Beth Shapiro, a scientific advisory board member at Colossal Biosciences. A professor of ecology and evolutionary biology at the University of California, Santa Cruz, Shapiro has had a decades-long fascination with the extinct bird.
Other experts in the field are cautiously optimistic, if just for the attention brought to conservation. “It’s incredibly exciting that there’s that kind of money available,” said Dr. Thomas Jensen, a cell and molecular reproductive physiologist at Wells College, to Nature. Whether it’ll work out, he added, remains to be seen.
A Genetic EgghuntThe de-extinction playbook is already laid out.
Step one, decode the extinct animal’s genome. Step two, find its nearest living cousin. Step three, screen for genetic differences, and replace the living animal’s DNA code with that of the extinct species. Finally, produce an embryo that can be brought to life in a surrogate species.
Yeah, it’s not exactly a walk in the park.
Thanks to Shapiro, Colossal has already nailed the first two steps. Back in 2002, her team sequenced a chunk of the bird’s mitochondrial DNA (mtDNA), which live inside the cell’s energy-producing factory, the mitochondria. These genetic codes are passed down solely through the maternal line. Comparing the dodo’s mtDNA with that of their living cousins, the team honed in on the Nicobar pigeon, a peacock-colored bird that roams around the Indian Andamans to the Solomons and New Guinea, as their closest living relative. The two birds shared a common ancestor roughly 30 to 50 million years ago, wrote Shapiro in a 2016 study.
Early last year, she announced that her team has sequenced the entire dodo genome from a museum sample, although the results have yet to be published in a scientific journal. By comparing the dodo’s genome sequences to that of the Nicobar, it’s now possible to hunt down DNA changes that define the dodo—and pinpoint genetic changes needed to transform a Nicobar into its long-extinct cousin.
An Avian HeadacheHere’s where the playbook changes.
In mammals, the edited genome—one that resembles the extinct species—is transplanted into an egg cell of its closet cousin and developed into an embryo. The embryo is then brought to life inside the surrogate womb of a living species, a method akin to cloning.
It doesn’t work for birds.
Cloning a species requires access to an egg cell that’s sufficiently developed so it can be fertilized. This stage is hard to capture in avian species. Then there’s the problem of reintroducing a cloned egg back into the body.
“To implant a cloned embryo, one would have to take out the developing embryo from within a developing hard-shelled egg within the female’s body and replace it with the cloned embryo—and hope that the embryo integrates into the yolk of the egg and that all the puncturing doesn’t deform the egg or harm the female,” explained Dr. Ben Novak, lead scientist and program manager for biotechnology for bird conservation at Revive & Restore, company focused on genetically rescuing endangered and extinct species.
Colossal honed in on a different approach for assisted reproduction: utilizing primordial germ cells (PGCs). True to their name, these cells can transform into both sperm and egg-producing cells. The company plans to extract these flexible reproductive “blank slates” from developing Nicobars and edit their DNA sequences to better match those of the dodo using tools such as CRISPR.
It’s a hard task. Most genetic tools are optimized for mammalian species, but those for birds are sorely lacking. So far, scientists have struggled to introduce just a single genetic change into quails. Editing the Nicobar will require thousands of precise DNA changes simultaneously.
Then comes the surrogate challenge. “Dodo eggs are much, much larger than Nicobar pigeon eggs, you couldn’t grow a dodo inside of a Nicobar egg,” said Jensen. He would know: his team inserted PGCs into chicken eggs, creating chimeric chickens that can generate quail sperm (but not eggs). Finding a potential surrogate match for a wild, extinct species is far more challenging.
That said, the rest of the process may be relatively smooth sailing.
In mammals, fetuses are heavily influenced by signals and molecules from the mother’s womb. We can’t yet predict how an extinct species interacts with its surrogate modern mom during pregnancy. In contrast, the bird egg is a relatively insulated environment and the process should be simpler, predicts Shapiro, because “everything happens in an egg.”
What if It Works?With rapid progress in genome editing and reproductive technologies, Colossal’s moonshot project may just work out. But would the resulting animal actually be a dodo?
To Dr. Mikkel Sinding at the University of Copenhagen, we need to consider both nature and nurture. Genetics is just one aspect that defines a species; societal interactions and the environment further shape a species’ behavior. But for a “resurrected” dodo, “there is nobody around to teach the dodo how to be a dodo,” he said.
Then there are ecological concerns. Even if the dodo retains its natural instincts, it would be brought back into a world that hasn’t existed for 300 years. The bird originally thrived in Mauritius. Today, the island is facing deteriorating forests, oil leaks, and plastics in its surrounding waters. Would an engineered dodo survive in that ecosystem? And if not, is it ethical to raise the creatures solely inside a zoo or otherwise controlled environment purely for our enjoyment?
These questions don’t yet have an answer. However, scientists hope the dodo may highlight environmental issues due to their superstar power. The project could help propel efforts to restore the island’s natural ecosystem, including endemic plants and other animals. In terms of technology, lessons learned along the way could cross over into biotech and medicine—for example, PGC-aided reproduction—ultimately with a far wider reach than de-extinction.
“There’s a new set of potential tools here, a new set of possibilities and opportunities,” said Dr. Ronald Sandler, the director of the Ethics Institute at Northeastern University in Boston.
Image Credit: Rawpixel.com/Henrik Gronvold
Geoengineering the planet to reverse the worst effects of climate change is a controversial idea that has been largely rejected by the scientific establishment. But what if we did it out in space instead?
Despite growing efforts to reduce carbon emissions worldwide, the consensus is that we’re doing too little too late. This has led to growing interest in geoengineering approaches, which either attempt to remove carbon dioxide from the atmosphere or manage the amount of solar radiation entering it.
Both would involve tinkering with critical Earth systems though, which is why most scientists say the risks outweigh the potential rewards. But an alternative that has been floated several times over the last few decades is to instead deflect the sun’s rays before they even reach the planet.
Now researchers have devised a new twist on this idea by proposing to fire dust from the moon’s surface towards a gravitationally stable point between the Earth and the sun where it could act as a solar shield.
“Our strategy could be an option in addressing climate change,” lead author Ben Bromley, from the University of Utah, said in a press release. “We aren’t experts in climate change, or the rocket science needed to move mass from one place to the other. We’re just exploring different kinds of dust on a variety of orbits to see how effective this approach might be. We do not want to miss a game changer for such a critical problem.”
Previous suggestions for creating solar shields in space have involved building giant mirrors or fleets of spacecraft in orbit, or most recently a giant raft of bubbles made of silicon. But a major problem with most of these approaches is the cost and complexity of doing construction in outer space.
An alternative is to use simple clouds of dust to reflect the light, but this raises the question of where to source it from and how to get it there. To have a significant impact on climate change you would need approximately 10 billion tons of the stuff, the researchers say in a paper in PLOS Climate outlining their idea.
The team is well suited to solving the problem, because they specialize in studying how planets form from the clouds of dust orbiting around stars. They used techniques from their regular line of work to analyze the best positions for a dust-based solar shield, the best ways to get it there, and how long it would stay put.
They found the most effective approach would be to launch dust from Earth to a space-based platform at the Lagrange point between the sun and Earth, where the gravitational pull of the two bodies cancels each other out. The dust would then be released and disperse to create an effective solar shield. However, the cost and effort involved in launching dust from Earth would be astronomical, and the particles would quickly be blown away by the solar wind.
That’s why the researchers suggest using lunar dust instead. The low gravity of the moon makes launching material far less costly, and the team also found that the regolith on the surface of the moon is surprisingly effective at reflecting light. What’s more, they found that firing it along a trajectory between the moon and the Lagrange point led to a sun shield that would last considerably longer.
There are clearly still plenty of gaps in the plan. For a start, you would need to build mining infrastructure on the moon to harvest the dust, and some kind of high-powered gun to launch it into space. Also, while it would hang around for longer, those 10 billion kilograms of dust would still need to be replenished at regular intervals. Perhaps most disconcertingly, any failure in the system would lead to a “termination shock” that could cause a rapid and deadly spike in solar radiation.
Given the enormous cost and potential pitfalls, it seems unlikely the idea will get off the ground anytime soon. But given our slow action on climate change so far, it can’t hurt to have some people thinking about potential moonshots.
Image Credit: NASA
From pocket knives to smartphones, humans keep inventing ever-more-sophisticated tools. However, the notion that tool use is an exclusively human trait was shattered in the 1960s when Jane Goodall observed our closest living relatives, chimpanzees, retrieving termites from holes with stripped twigs.
Tool use among non-human animals is hotly debated. It’s often thought a big brain is needed to understand the properties of objects, how to finely manipulate them, and how to teach this to other members of a species.
Until recently, humans and chimps stood out among tool-using species. They were considered the only species that used “toolsets,” wherein a collection of different tools is used to achieve a task. They were also thought to be the only animals that carried toolsets in anticipation of needing them later.
A third species joined the exclusive club of toolset makers in 2021, when scientists in Indonesia saw wild Goffin’s cockatoos using three distinct types of tools to extract seeds from fruit. And in research published this week, researchers have shown Goffin’s cockatoos can also take the next leap of logic, by carrying a set of tools they’ll need for a future task.
Bright, Enigmatic CreaturesParrots have proven to be something of an enigma. They’re known to be highly intelligent creatures, yet they’ve rarely been observed using tools in the wild.
Curiously, the only parrot species known to use tools regularly in the wild is Australia’s own palm cockatoo, which uses them in a very unusual way. Males in northern Australia “manufacture” drumsticks and seedpod tools to use during their complex mating displays. They grasp the drumstick or seedpod in the left foot and beat it against a hollow trunk in a rhythmic performance, with all the hallmarks of human instrumental music.
The 2021 study of wild Goffin’s cockatoos was particularly significant as it showed the birds’ tools were similar in complexity to those made by chimps, meaning their cognitive skills could be directly compared.
A small number of Goffin’s cockatoos were seen crafting a set of tools designed for three different purposes—wedging, cutting, and spooning—and using them sequentially to access seeds in fruits. This requires similar brain power to a chimp’s method of using multiple tools when fishing for termites.
Anticipating ProblemsAn initial stumbling block in interpreting chimps’ use of toolsets was that nobody could show whether they visualized a collection of small tasks as one problem, or used single tools to solve separate problems.
Researchers finally solved this when they observed chimpanzees not only carrying their toolsets with them, but doing this flexibly and according to the exact problems they faced. They must have been thinking it through from start to finish!
This is precisely what Goffin’s cockatoos have now been shown to do (albeit in a captive setting). They’ve been confirmed as the third species that can not only use tools, but can carry toolsets in anticipation of needing them later on.
This panel of photos shows Figaro the cockatoo flying with two tools towards a box with a cashew. Image Credit: Thomas Suchanek, CC BY-NC-SAInspired by the toolsets chimpanzees use and transport in the wild for extracting termites from the ground, the authors of the study designed clever experiments to test Goffin’s cockatoos under similar circumstances.
The birds, initially ten in total, had to extract cashews from boxes that required either one or two tool types. They were tested in various ways to examine their flexibility and innovation, but the pièce de resistance came when reaching the box with the tools required additional movement, including climbing a ladder and horizontal and vertical flight.
Though only five of the ten birds made it through the earlier experiments, four of those that did tended to transport both tools in one go, in anticipation of needing them to open the two-tool box. In other words, these birds could categorize both tools as a “toolset” and use it accordingly. Mission accomplished!
Nothing Wrong With a Bird BrainBut what about needing a big brain for complex tasks?
Like primates, some bird species have enlarged forebrains that provide them enhanced cognitive abilities including insight and innovation, understanding of others’ mental states, symbolic communication, episodic memory, and future planning.
Parrots are especially well endowed with these abilities, so we shouldn’t be surprised they can use toolsets as easily as chimpanzees. Rather, what’s surprising is that more parrots haven’t been seen transporting toolsets for future use.
One has to conclude it’s because wild parrots are rarely presented with problems that require this. Parrots have powerful feet and beaks that allow them to reach the most difficult places and break the hardest fruits and seeds. Yet bright individuals in captivity can spontaneously invent new tools to solve new problems—so there’s no doubting how capable they are.
This new study is further proof parrots belong in the animal world’s exclusive version of Mensa. Between the considered planning shown by Goffin’s cockatoos, and the palm cockatoo’s ability to play instruments, it seems we’ve only scratched the surface of what these remarkable birds can achieve.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Thomas Suchanek, CC BY-SA
ARTIFICIAL INTELLIGENCEThe Original Startup Behind Stable Diffusion Has Launched a Generative AI for Video
Will Douglas Heaven | MIT Technology Review“In a demo reel posted on its website, Runway shows how its software, called Gen-1, can turn clips of people on a street into claymation puppets, or books stacked on a table into a cityscape at night. Runway hopes that Gen-1 will do for video what Stable Diffusion did for images. ‘We’ve seen a big explosion in image-generation models,’ says Runway CEO and cofounder Cristóbal Valenzuela. ‘I truly believe that 2023 is going to be the year of video.’i”
ENERGYA Bold Plan to Beam Solar Energy Down From Space
Ramin Skibba | Wired“Whether you’re covering deserts, ugly parking lots, canals, or even sunny lakes with solar panels, clouds will occasionally get in the way—and every day the sun must set. No problem, says the European Space Agency: Just put the solar arrays in space. The agency recently announced a new exploratory program called Solaris, which aims to figure out if it is technologically and economically feasible to launch solar structures into orbit, use them to harness the sun’s power, and transmit energy to the ground.”
TECH7 Problems Facing Bing, Bard, and the Future of AI Search
James Vincent | The Verge“Satya Nadella, Microsoft’s CEO, describes the changes as a new paradigm—a technological shift equal in impact to the introduction of graphical user interfaces or the smartphone. And with that shift comes the potential to redraw the landscape of modern tech—to dethrone Google and drive it from one of the most profitable territories in modern business. Even more, there’s the chance to be the first to build what comes after the web. But each new era of tech comes with new problems, and this one is no different.”
TRANSPORTATIONElectric Vehicles Could Match Gasoline Cars on Price This Year
Jack Ewing | The New York Times“Increased competition, government incentives and falling prices for lithium and other battery materials are making electric vehicles noticeably more affordable. The tipping point when electric vehicles become as cheap as or cheaper than cars with internal combustion engines could arrive this year for some mass market models and is already the case for some luxury vehicles.”
ARTIFICIAL INTELLIGENCEResearchers Discover a More Flexible Approach to Machine Learning
Steven Nadis | Quanta“Apart from applications like autonomous driving and flight, liquid networks seem well suited to the analysis of electric power grids, financial transactions, weather and other phenomena that fluctuate over time. In addition, Hasani said, the latest version of liquid networks can be used “to perform brain activity simulations at a scale not realizable before.’i”
SPACERolls-Royce Nuclear Engine Could Power Quick Trips to the Moon and Mars
Kevin Hurler | Gizmodo“The British aerospace engineering company says it’s developing a micro-nuclear reactor that the company hopes could be a source of fuel for long trips to the Moon and Mars. …Since the nuclear reactor won’t have to carry as much fuel as a chemical propulsion rocket, the entire system will be lighter allowing for faster travel or increased payloads.”
ENVIRONMENTThe Generative AI Race Has a Dirty Secret
Chris Stokel-Walker | Wired“The race to build high-performance, AI-powered search engines is likely to require a dramatic rise in computing power, and with it a massive increase in the amount of energy that tech companies require and the amount of carbon they emit. …Martin Bouchard, cofounder of Canadian data center company QScale, believes that, based on his reading of Microsoft and Google’s plans for search, adding generative AI to the process will require ‘at least four or five times more computing per search’ at a minimum.”
ENERGYWe Were Promised Smaller Nuclear Reactors. Where Are They?
Casey Crownhart | MIT Technology Review“For over a decade, we’ve heard that small reactors could be a big part of nuclear power’s future. “Because of their size, small modular reactors (SMRs) could solve some of the major challenges of traditional nuclear power, making plants quicker and cheaper to build and safer to operate. “That future may have just gotten a little closer.”
SCIENCEHow Our Reality May Be a Sum of All Possible Realities
Charlie Wood | Quanta“The most powerful formula in physics starts with a slender S, the symbol for a sort of sum known as an integral. Further along comes a second S, representing a quantity known as action. Together, these twin S’s form the essence of an equation that is arguably the most effective diviner of the future yet devised. The oracular formula is known as the Feynman path integral. As far as physicists can tell, it precisely predicts the behavior of any quantum system—an electron, a light ray or even a black hole.”
Image Credit: Miti / Unsplash
As technology advances and starts to push the idea of designer babies from the realm of science fiction into reality, concern is rising around the murky ethics involved. Scientists and government bodies have started laying out guidelines around human enhancement and germline editing.
But besides these extreme scenarios, where embryos could be tweaked using genetic engineering tools like CRISPR, there are similar technologies already being used—and their ethical implications are no less complex, particularly given their accessibility. A recent study found that a substantial portion of Americans would be interested in using genetics tech to make their babies smarter.
The study was supported by the National Institutes of Health and published yesterday in Science. The team asked survey respondents who may conceive using in vitro fertilization (IVF) how likely they were to use polygenic screening or CRISPR-style gene editing to increase their kids’ chances of getting into a top-100 ranked college.
The researchers told respondents that for purposes of the study they should assume the screening and editing options would be both free and safe. Neither of these assumptions are reality; the technologies haven’t been proven to be fully safe (particularly using CRISPR on embryos), and they’re certainly not free. Since a high cost and unproven safety would both substantially detract from peoples’ openness to the tech, though, simply gauging their attitudes was simplified by operating under these assumptions.
28 percent of respondents said they were more likely than not to use gene editing to make their babies smarter, and 38 percent said they’d use polygenic screening. The researchers also noted what they called a bandwagon effect, where people who were told something along the lines of “everyone else is doing it” were more likely to say they’d do it too. This is logical; our comfort with decisions is buoyed by a sense that others in our shoes would choose similarly.
It’s important to note, though, that the survey made it clear that genetically enhancing embryos didn’t come with a guaranteed result of a smarter kid. “In this study, we stipulated a realistic effect—that each service would increase the odds of having a child who attends a top-100 college by 2 percentage points, from 3 percent to 5 percent odds—and lots of people are still interested,” said Michelle N. Meyer, chair of the Department of Bioethics and Decision Sciences at Geisinger and first author of the article.
28 and 38 percent don’t seem like high numbers—that’s a little below and a little above one-third of total respondents who would use the technologies. But imagine walking around in a world where one out of every three people had had their genes tweaked before birth. Unsettling, no? The researchers said their results point to substantial and growing interest in genetic technologies for offspring enhancement, and that now is the time to get a national conversation going around regulations.
They emphasized the danger of relying on polygenic embryo screening as a trait-prediction tool. Polygenic risk scores are based on your genes and can give you an estimate of your and your kids’ risk for diseases like diabetes, cancer, Alzheimer’s, or schizophrenia. Analyzing an embryo’s genes can give some indication of their risk for these conditions, and companies are already offering polygenic screening to people trying to conceive through IVF. If multiple embryos are screened, would-be parents can choose to implant the one with the best scores.
It’s already gone a couple steps beyond screening for optimal health outcomes, though—people have provided their embryos’ genomic data to services that use it to make predictions about non-medical traits. It’s not only a slippery slope, but there’s not enough evidence showing clear links between these predictions and real-life outcomes.
“Polygenic indexes are already only weak predictors for most individual adult outcomes, especially for social and behavioral traits, and there are several factors that lower their predictive power even more in the context of embryo selection,” said senior author Patrick Turley, assistant research professor of economics at the USC Dornsife College of Letters, Arts and Sciences.
The team noted the importance of a person’s environment in their gene expression—epigenetics—as well as the disparities that exist between the data available for people of European ancestry versus those of other heritage.
Economic disparities should be kept in mind too; since these technologies are far from free, the wealthy would have exclusive access to them, further widening gaps in equality that have already brought negative impacts on society.
Everyone wants to give their child the best possible chance at a healthy, happy life. Now that gene editing and polygenic screening are already “out of the box,” so to speak, they’re not going back in. But as this study emphasizes, they should be carefully studied, considered, and regulated sooner rather than later.
Image Credit: www.picjumbo.com from Pixabay
Black holes are bizarre things, even by the standards of astronomers. Their mass is so great, it bends space around them so tightly that nothing can escape, even light itself.
And yet, despite their famous blackness, some black holes are quite visible. The gas and stars these galactic vacuums devour are sucked into a glowing disc before their one-way trip into the hole, and these discs can shine more brightly than entire galaxies.
Stranger still, these black holes twinkle. The brightness of the glowing discs can fluctuate from day to day, and nobody is entirely sure why.
My colleagues and I piggy-backed on NASA’s asteroid defense effort to watch more than 5,000 of the fastest-growing black holes in the sky for five years, in an attempt to understand why this twinkling occurs. In a new paper in Nature Astronomy, we report our answer: a kind of turbulence driven by friction and intense gravitational and magnetic fields.
Gigantic Star-EatersWe study supermassive black holes, the kind that sit at the centers of galaxies and are as massive as millions or billions of suns.
Our own galaxy, the Milky Way, has one of these giants at its center, with a mass of about four million suns. For the most part, the 200 billion or so stars that make up the rest of the galaxy (including our sun) happily orbit around the black hole at the center.
However, things are not so peaceful in all galaxies. When pairs of galaxies pull on each other via gravity, many stars may end up tugged too close to their galaxy’s black hole. This ends badly for the stars: they are torn apart and devoured.
We are confident this must have happened in galaxies with black holes that weigh as much as a billion suns, because we can’t imagine how else they could have grown so large. It may also have happened in the Milky Way in the past.
Black holes can also feed in a slower, more gentle way: by sucking in clouds of gas blown out by geriatric stars known as red giants.
Feeding TimeIn our new study, we looked closely at the feeding process among the 5,000 fastest-growing black holes in the universe.
In earlier studies, we discovered the black holes with the most voracious appetites. Last year, we found a black hole that eats an Earth’s-worth of stuff every second. In 2018, we found one that eats a whole sun every 48 hours.
But we have lots of questions about their actual feeding behavior. We know material on its way into the hole spirals into a glowing “accretion disc” that can be bright enough to outshine entire galaxies. These visibly feeding black holes are called quasars.
Most of these black holes are a long, long way away—much too far for us to see any detail of the disc. We have some images of accretion discs around nearby black holes, but they are merely breathing in some cosmic gas rather than feasting on stars.
Five Years of Flickering Black HolesIn our new work, we used data from NASA’s ATLAS telescope in Hawaii. It scans the entire sky every night (weather permitting), monitoring for asteroids approaching Earth from the outer darkness.
These whole-sky scans also happen to provide a nightly record of the glow of hungry black holes, deep in the background. Our team put together a five-year movie of each of those black holes, showing the day-to-day changes in brightness caused by the bubbling and boiling glowing maelstrom of the accretion disc.
The twinkling of these black holes can tell us something about accretion discs.
In 1998, astrophysicists Steven Balbus and John Hawley proposed a theory of “magneto-rotational instabilities” that describes how magnetic fields can cause turbulence in the discs. If that is the right idea, then the discs should sizzle in regular patterns. They would twinkle in random patterns that unfold as the discs orbit. Larger discs orbit more slowly with a slow twinkle, while tighter and faster orbits in smaller discs twinkle more rapidly.
But would the discs in the real world prove this simple, without any further complexities? (Whether “simple” is the right word for turbulence in an ultra-dense, out-of-control environment embedded in intense gravitational and magnetic fields where space itself is bent to its breaking point is perhaps a separate question).
Using statistical methods, we measured how much the light emitted from our 5,000 discs flickered over time. The pattern of flickering in each one looked somewhat different.
But when we sorted them by size, brightness, and color, we began to see intriguing patterns. We were able to determine the orbital speed of each disc—and once you set your clock to run at the disc’s speed, all the flickering patterns started to look the same.
This universal behavior is indeed predicted by the theory of “magneto-rotational instabilities.” That was comforting! It means these mind-boggling maelstroms are “simple” after all.
And it opens new possibilities. We think the remaining subtle differences between accretion discs occur because we are looking at them from different orientations.
The next step is to examine these subtle differences more closely and see whether they hold clues to discern a black hole’s orientation. Eventually, our future measurements of black holes could be even more accurate.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: EHT Collaboration
The price of eggs has more than doubled in the last year due to inflation, avian flu outbreaks, and the war in Ukraine. But demand for the breakfast and baking staple hasn’t gone down much; people like eggs, and there aren’t many viable substitutes that truly taste, look, and perform like the real thing.
An Israeli startup called Yo Egg thinks it has a solution in the form of vegan eggs. The product doesn’t share much with real eggs in terms of composition, but the company says it’s achieved a near-exact match in taste and texture.
Yo eggs are made primarily of water, vegetable oil, soy protein, and chickpea protein, with small amounts of other ingredients including potato starch, yeast, and seaweed extract. One egg has 40 calories, 1 gram of fat, no cholesterol, and 3 grams of protein.
A large Grade-A chicken egg, meanwhile, has 70 calories, 5 grams of fat, 195 milligrams of cholesterol, 6 grams of protein, and at least 10 percent of the recommended daily value of vitamins A, D, E, and B12, among other important nutrients. So while you get less bad stuff—namely fat and cholesterol—from a Yo egg, you get less good stuff too.
Yo egg was first served at an Israeli breakfast chain called Benedict, and the company unveiled its sunny-side-up eggs in the US at a trade show last May. The eggs are also served at Google and Facebook’s corporate offices in Israel. Besides scaling up production of their existing products, Yo Egg wants to create vegan versions of a hard-boiled egg and a scrambled egg.
The company’s website doesn’t show what sort of packaging the eggs come in, but they’re likely individually packaged in some sort of imitation shell (even if that “shell” is a plastic cube, like an ice-cube tray), unlike egg substitutes that can be poured out of a milk-like container; since Yo Egg aims to give consumers the “whole egg experience,” it’s important that the product is served with distinct white and yolk components.
There are several other plant-based egg substitutes on the market, and more in the works. Evo Foods has a liquid egg alternative in India; Swiss grocery company Migros makes a soy-protein-based hard-boiled egg; Singapore-based OsomeFood has a hard-boiled egg made of mycoprotein; and San Francisco-based Eat Just’s Just Egg is made of mung bean protein. There’s also Every Company’s Every EggWhite, made using precision fermentation and a protein recipe from real chickens.
The global plant-based egg market is predicted to reach just under $800 million in value by 2027, up from $148 million in 2020; increasing demand for vegetarian and vegan foods are expected to be the biggest factor in that growth. More people are starting to look for animal-free options due to health, environmental, and animal rights reasons (though the recent struggles of the plant-based meat industry, which is being called both a fad and a flop, indicate otherwise).
Yo Egg has a production facility in Israel and recently opened a second one in Los Angeles, which they say can produce thousands of eggs per day. Their poached egg will start being offered this week at six different restaurants in LA.
“Our vision is to create the world’s largest egg company, not egg alternative company, and not the largest plant-based egg company, but the largest egg company without using chickens,” Yo Egg’s CEO, Eran Groner, told TechCrunch.
They’ll have their work cut out for them; they’re aiming to get the vegan eggs into grocery stores and reach price parity with traditional eggs within the next few years (whether that means the inflated prices we’re seeing now or the prices of two years ago is unclear; hopefully the latter).
“It will work in our benefit to remove the animals from the food system,” Groner said. “Because then we won’t see price hikes, we’ll use way less natural resources, and we’ll minimize the use of antibiotics and the danger of animal-borne diseases.”
Let’s be honest: the odds of fully removing animals from the food system anytime in the foreseeable future are slim to none. But for consumers who want alternatives, the available options are steadily growing.
Image Credit: Yo! Egg
Nearly a decade ago, mini-brains shot onto the neuroscience scene with a hefty promise: understanding the developing brain and restoring injured brains.
Known as brain organoids, these tiny clumps of brain tissue—roughly the size of a lentil—look nothing like the three-pound organ piloting our lives. Yet under the surface, they behave eerily similar to the brain of a human fetus. Their neurons spark with electrical activity. They readily integrate with—and subsequently control—muscles, at least in a dish. Similar to full-blown brains, they give birth to new neurons. Some even develop the six-layered structure of the human cortex—the wrinkly, outermost layer of the brain that supports thought, reasoning, judgment, speech, and perhaps even consciousness.
Yet a critical question haunts neuroscientists: can these Frankenstein bits of brain tissue actually restore an injured brain?
A study published in Cell Stem Cell this month concluded that they can. Using brain organoids made from human cells, a team led by Dr. Han-Chiao Isaac Chen at the University of Pennsylvania transplanted the mini-brains into adult rats with substantial damage to their visual cortex—the area that supports vision.
In just three months, the mini-brains merged with the rats’ brains. When the team shone flashing lights for the animals, the organoids spiked with electrical activity. In other words, the human mini-brain received signals from the rats’ eyes.
It’s not just random noise. Similar to our visual cortex, some of the mini-brain’s neurons gradually developed a preference for light shone at a particular orientation. Imagine looking at a black and white windmill blow toy as your eyes adjust to the different moving stripes. It sounds simple, but the ability of your eyes to adjust—dubbed “orientation selection”—is a sophisticated level of visual processing that’s critical to how we perceive the world.
The study is one of the first to show that mini-brain tissue can integrate with an injured adult host and perform its intended function. Compared to previous attempts at stem cell transplants, the artificial tissues could substitute an injured or degenerating piece of the brain in the future—but many caveats remain.
“Neural tissues have the potential to rebuild areas of the injured brain,” said Chen. “We haven’t worked everything out, but this is a very solid first step.”
A Mini-Brain’s Mini-LifeBrain organoids have had a hell of a ride. First engineered in 2014, they immediately captured the interest of neuroscientists as an unprecedented model of the brain.
The quasi-brains are made from multiple sources to mimic different areas of the brain. One immediate use was to combine the technology with iPSCs (induced pluripotent stem cells) to study neurodevelopmental disorders, such as schizophrenia or autism.
Here, a patient’s skin cells are transformed back into a stem-cell-like state, which can be further grown into a 3D tissue of their brain. Because the person and the mini-brain share the same genes, it’s possible to partially duplicate the person’s brain during development—and potentially hunt down new cures.
Since their birth, mini-brains have now expanded in size, age, and sophistication. One major leap was a consistent blood supply. Our brains are intimately intertwined with blood vessels, feeding our neurons and neural networks with oxygen and nutrients to supply energy. The breakthrough came in 2017, when several teams showed that transplanting human organoids into rodent brains triggered the host’s blood vessels to integrate and “feed” the structured brain tissue, allowing it to further develop into the intricate brain architecture inside the host. The studies sparked a firestorm of discussion within the field, with bioethicists and researchers alike wondering if human organoids could change a rodent’s perception or behavior.
Chen had a different, if more challenging idea. Most previous studies transplanted mini-brains into infant rodents to nurture the organoids and ease their merger with the developing brain.
Adult brains, in contrast, are far more ridged. Highly intertwined neural circuits—including their signaling and functions—are already established. Even when injured, when the brain is ready for repair, shoving in extra bits of human organoid grafts like a Band-Aid could support broken neural circuits—or interfere with established ones.
Chen’s new study put the theory to the test.
An Unexpected MergerTo start, the team cultivated brain organoids with a renewable human stem cell line. Using a previously validated chemical recipe, the cells were coaxed into mini-brains that mimic the frontal parts of the cortex (around the forehead).
By day 80, the team saw rudimentary cortical layers in the organoid, along with cells organized in a way that resembled a developing brain. They then transplanted the organoids into the damaged visual cortex of young adult rats.
Just one month after transplant, the host’s blood vessels merged with the human tissue, supplying it with much-needed oxygen and nutrients and allowing it to further grow and mature. The mini-brains developed a myriad of different brain cells—not just neurons, but also “supporting” brain cells such as astrocytes and specialized immune cells dubbed microglia. The latter two are far from dispensable: they have been implicated in brain aging, Alzheimer’s disease, inflammation, and cognition.
But can the transplanted human mini-brain function inside a rat?
In a first test, the team used a popular tracer to map the connections between the organoid and the animal’s eye. Similar to a dye, the tracer is a virus that hops between neural connections—dubbed synapses—while carrying a protein that glows a bright green under a fluorescent microscope. Like a highlighted route on Google Maps, the light stream clearly connected all the way to the transplanted mini-brain, meaning that its circuitry linked up, through multiple synapses, to the rats’ eyes.
Second question: could the transplanted tissue help the rat “see”? In six out of eight animals, turning the lights on or off triggered an electrical response, suggesting the human neurons responded to outside stimulation. The pattern of the electrical activity resembled natural ones seen in the visual cortex, “suggesting that organoid neurons have a comparable potential for light responsiveness to visual cortex neuron,” the authors said.
In another test, the grafts developed “picky” neurons that preferred a specific orientation selectivity for light—a quirk embedded inside our ability to perceive the world. When tested with different light gratings that flickered from black to white, the grafted neurons’ overall preference mimicked that of normal, healthy neurons.
“We saw that a good number of neurons within the organoid responded to specific orientations of light, which gives us evidence that these organoid neurons were able to not just integrate with the visual system, but they were able to adopt very specific functions of the visual cortex,” said Chen.
Plug-and-Play Brain Tissue?The study shows that mini-brains can rapidly establish neural networks with the host’s brain, at a rate far faster than transplanting individual stem cells. It suggests a powerful use for the technology: repairing damaged brains at unprecedented speed.
Many questions remain. For one, the study was conducted in rats dosed with immunosuppressants to inhibit rejection. The hope for mini-brains is that they’ll be cultured from a patient’s own cells, eliminating the need for immunosuppressant drugs—a hope yet to be fully tested. Another problem is how to best match the mini-brain’s “age” to its host’s, so as not to disrupt the person’s intrinsic neural signals.
The team’s next step is to support other damaged brain regions using mini-brains, particularly damage due to degeneration from age or disease. Adding non-invasive technologies, such as neuromodulation or visual “rehabilitation” of the neurons, could further help the transplant integrate into the host’s circuit and potentially elevate their function.
“Now, we want to understand how organoids could be used in other areas of the cortex, not just the visual cortex, and we want to understand the rules that guide how organoid neurons integrate with the brain so that we can better control that process and make it happen faster,” said Chen.
Image Credit: Jgamadze et al.
A common concern about the green energy transition is whether we’ll have enough materials to build all the wind turbines and solar panels required. A new analysis suggests planet Earth has more than enough to go around.
The technologies that will be crucial if we are to switch from fossil fuels to renewable energy require some highly specific materials. For instance, wind turbines need considerable amounts of fiberglass to build their blades, and solar panels require huge amounts of high-grade polysilicon. Rare earth metals are also needed in a wide range of renewable power technologies.
A rapid expansion in green energy roll-outs will massively increase demand for these key ingredients, not to mention boosting requirements for structural materials like steel, cement, and common metals used in electronics like copper and nickel. But efforts to assess whether we have the capacity to meet that demand have been piecemeal.
That prompted researchers to carry out the most comprehensive account to date of the materials needed for the green energy transition. In a paper published in Joule, they show that even under the most ambitious scenarios Earth’s geological reserves of these key ingredients are more than sufficient. And while extracting them will result in significant carbon emissions, these will be a fraction of what will be produced if we don’t switch to renewable power.
The analysis estimated demand for 17 key materials used in renewable energy generation technology between 2020 and 2050. The authors consider 75 different scenarios that differ in how quickly certain technologies are deployed and the speed with which emissions are reduced.
Crucially, the scenarios they considered took into account the fact that renewable generation will not only need to replace existing fossil fuel plants, but also expand overall capacity to meet the expected growth in demand for electricity over the coming decades. They also looked at a range of different projections about how much material will be required for each of these technologies.
For almost all the materials, they found that total demand only represented a small fraction of “geological reserves,” which refers to global reserves that can be recovered economically. The one outlier was the rare earth metal tellurium, which is used in emerging thin-film solar panels. Under some scenarios, demand could outstrip total reserves, which the authors concede could limit the rollout of this technology.
Even if the reserves are there, though, the analysis found that we will need to significantly expand the rate at which we produce or extract these resources. Yearly fiberglass requirements could be as high as 66.4 percent of today’s total production capacity, while annual demand for the rare earth metals dysprosium, neodymium, and tellurium will be 309.4 percent, 271.4 percent, and 372.4 percent of what we currently manage.
This will inevitably lead to a significant jump in emissions from the industries involved in providing these materials. However, the researchers found that even in the worst case, the emissions would total 29 gigatons of CO2 equivalent, which is a tiny fraction of the 320 gigatons we can still emit and have a good chance of avoiding more than 1.5 degrees of warming.
The analysis does miss out on a key source of future demand for many of these materials: batteries. Given the expected worldwide transition to electric vehicles and probable need for grid-scale storage, that could change the math considerably.
In addition, experts have pointed out that increased emissions and total reserves aren’t the only concern when it comes to a massive increase in material extraction and production. Mining these materials is often highly damaging to the environment, and many of these resources are concentrated in countries that have exploitative labor practices, such as the Democratic Republic of Congo, or tense relations with the West, such as China.
Increasing extraction to the levels required by these scenarios would also raise fundamental logistical issues. Demetrios Papathanasiou, global director for energy and extractives at the World Bank, told MIT Technology Review that over the next three decades we’ll need to mine the same amount of copper as humanity has mined to date.
We may get some help from innovations in recycling and efforts to reduce the amount of materials required by these technologies, but the authors warn this is unlikely to make a significant difference in total demand. “With the power sector becoming a sizable industrial consumer of some inputs, the mining and mineral processing sector will consequently play a crucial role in supporting the clean energy transition,” they conclude.
Image Credit: jaroslava V/Shutterstock.com
TECHChatGPT May Be the Fastest Growing App in History
Lauren Leffer | Gizmodo“In January, just two months after the program’s public launch, ChatGPT reached 100 million monthly active users (MAUs), UBS noted, based on data amassed from Similarweb. …For comparison, it took TikTok about nine months from its worldwide launch to net 100 million users, per a report from Reuters. Instagram didn’t reach that benchmark for two and a half years, the outlet added.”
ARTIFICIAL INTELLIGENCEThe Generative AI Revolution Has Begun—How Did We Get Here?
Haomiao Huang | Ars Technica“You may be familiar with the latest happenings in the world of AI. You’ve seen the prize-winning artwork, heard the interviews between dead people, and read about the protein-folding breakthroughs. …There’s a reason all of this has come at once. The breakthroughs are all underpinned by a new class of AI models that are more flexible and powerful than anything that has come before. …Where did these foundation models come from, and how have they broken out beyond language to drive so much of what we see in AI today?”
TRANSPORTATIONStartup’s Bladeless Flying Car Is Designed to Reach Mach 0.8
Kristin Houser | Big Think“[Jetoptera is replacing] standard spinning propellers with a ‘Fluidic Propulsion System’ (FPS) that it describes as ‘a bladeless fan on steroids.’ …The FPS has no moving parts that passengers could come in contact with, and based on research funded by the DoD, Jetoptera says the system is ‘the most silent propulsion method in the skies.’ ‘In places like New York, Los Angeles, and London, our aircraft wouldn’t be heard until it was about 200 feet away,’ Andrei Evulet, Jetoptera’s CEO and CTO, told Future Flight in 2021.”
BIOTECHA De-Extinction Company Is Trying to Resurrect the Dodo
Antonio Regalado | MIT Technology Review“It’s the third species picked by Colossal Biosciences, of Austin, Texas, for what it calls a process of technological ‘de-extinction.’ The company is also working on using large-scale genome engineering to morph modern elephants back into woolly mammoths and resurrect the Tasmanian tiger.”
ROBOTICSRoboticists Want to Give You a Third Arm
Dario Farina, Etienne Burdet, Carsten Mehring, and Jaime Ibàñez | IEEE Spectrum“What could you do with an extra limb? Consider a surgeon performing a delicate operation, one that needs her expertise and steady hands—all three of them. As her two biological hands manipulate surgical instruments, a third robotic limb that’s attached to her torso plays a supporting role. Or picture a construction worker who is thankful for his extra robotic hand as it braces the heavy beam he’s fastening into place with his other two hands. …Such scenarios may seem like science fiction, but recent progress in robotics and neuroscience makes extra robotic limbs conceivable with today’s technology.”
SENSORSThis Tiny Sensor Is About to Change Your Phone Camera Forever
Andy Boxall | Digital Trends“i‘We believe there’s a real opportunity to develop and use a spectral imager in a smartphone. Despite all the progress which has been made with different cameras and the computing power of a smartphone, none can really identify the true color of a picture.’ This is how Spectricity CEO Vincent Mouret described the company’s mission to Digital Trends in a recent interview, as well as the reason why it’s making a miniaturized spectral image sensor that’s ready for use in a smartphone.“
ETHICSResearchers Prove AI Art Generators Can Simply Copy Existing Images
Kyle Barr | Gizmodo“One of the main defenses used by those who are bullish on AI art generators is that although the models are trained on existing images, everything they create is new. AI evangelists often compare these systems to real life artists. Creative people are inspired by all those who came before them, so why can’t AI be similarly evocative of previous work? New research may put a damper on that argument, and could even become a major sticking point for multiple ongoing lawsuits regarding AI-generated content and copyright.”
NEUROSCIENCEThe Difference Between Speaking and Thinking
Matteo Wong | The Atlantic“Although ChatGPT can generate fluent and sometimes elegant prose, easily passing the Turing-test benchmark that has haunted the field of AI for more than 70 years, it can also seem incredibly dumb, even dangerous. It gets math wrong, fails to give the most basic cooking instructions, and displays shocking biases. In a new paper, cognitive scientists and linguists address this dissonance by separating communication via language from the act of thinking: Capacity for one does not imply the other.”
Image Credit: Eugene Production / Unsplash
Some 540 million years ago, diverse life forms suddenly began to emerge from the muddy ocean floors of planet Earth. This period is known as the Cambrian Explosion, and these aquatic critters are our ancient ancestors.
All complex life on Earth evolved from these underwater creatures. Scientists believe all it took was an ever-so-slight increase in ocean oxygen levels above a certain threshold.
We may now be in the midst of a Cambrian Explosion for artificial intelligence (AI). In the past few years, a burst of incredibly capable AI programs like Midjourney, DALL-E 2, and ChatGPT have showcased the rapid progress we’ve made in machine learning.
AI is now used in virtually all areas of science to help researchers with routine classification tasks. It’s also helping our team of radio astronomers broaden the search for extraterrestrial life, and results so far have been promising.
Discovering Alien Signals With AIAs scientists searching for evidence of intelligent life beyond Earth, we have built an AI system that beats classical algorithms in signal detection tasks. Our AI was trained to search through data from radio telescopes for signals that couldn’t be generated by natural astrophysical processes.
When we fed our AI a previously studied dataset, it discovered eight signals of interest the classic algorithm missed. To be clear, these signals are probably not from extraterrestrial intelligence, and are more likely rare cases of radio interference.
Nonetheless, our findings—published today in Nature Astronomy—highlight how AI techniques are sure to play a continued role in the search for extraterrestrial intelligence.
Not So IntelligentAI algorithms do not “understand” or “think.” They do excel at pattern recognition, and have proven exceedingly useful for tasks such as classification—but they don’t have the ability to problem solve. They only do the specific tasks they were trained to do.
So although the idea of an AI detecting extraterrestrial intelligence sounds like the plot of an exciting science fiction novel, both terms are flawed: AI programs are not intelligent, and searches for extraterrestrial intelligence can’t find direct evidence of intelligence.
Instead, radio astronomers look for radio “technosignatures.” These hypothesized signals would indicate the presence of technology and, by proxy, the existence of a society with the capability to harness technology for communication.
For our research, we created an algorithm that uses AI methods to classify signals as being either radio interference, or a genuine technosignature candidate. And our algorithm is performing better than we’d hoped.
What Our AI Algorithm DoesTechnosignature searches have been likened to looking for a needle in a cosmic haystack. Radio telescopes produce huge volumes of data, and in it are huge amounts of interference from sources such as phones, WiFi, and satellites.
Search algorithms need to be able to sift out real technosignatures from “false positives,” and do so quickly. Our AI classifier delivers on these requirements.
It was devised by Peter Ma, a University of Toronto student and the lead author on our paper. To create a set of training data, Peter inserted simulated signals into real data, and then used this dataset to train an AI algorithm called an autoencoder. As the autoencoder processed the data, it “learned” to identify salient features in the data.
In a second step, these features were fed to an algorithm called a random forest classifier. This classifier creates decision trees to decide if a signal is noteworthy, or just radio interference—essentially separating the technosignature “needles” from the haystack.
After training our AI algorithm, we fed it more than 150 terabytes of data (480 observing hours) from the Green Bank Telescope in West Virginia. It identified 20,515 signals of interest, which we then had to manually inspect. Of these, eight signals had the characteristics of technosignatures, and couldn’t be attributed to radio interference.
Eight Signals, No Re-DetectionsTo try and verify these signals, we went back to the telescope to re-observe all eight signals of interest. Unfortunately, we were not able to re-detect any of them in our follow-up observations.
We’ve been in similar situations before. In 2020 we detected a signal that turned out to be pernicious radio interference. While we will monitor these eight new candidates, the most likely explanation is they were unusual manifestations of radio interference: not aliens.
Sadly the issue of radio interference isn’t going anywhere. But we will be better equipped to deal with it as new technologies emerge.
Narrowing the SearchOur team recently deployed a powerful signal processor on the MeerKAT telescope in South Africa. MeerKAT uses a technique called interferometry to combine its 64 dishes to act as a single telescope. This technique is better able to pinpoint where in the sky a signal comes from, which will drastically reduce false positives from radio interference.
If astronomers do manage to detect a technosignature that can’t be explained away as interference, it would strongly suggest humans aren’t the sole creators of technology within the galaxy. This would be one of the most profound discoveries imaginable.
At the same time, if we detect nothing, that doesn’t necessarily mean we’re the only technologically-capable “intelligent” species around. A non-detection could also mean we haven’t looked for the right type of signals, or our telescopes aren’t yet sensitive enough to detect faint transmissions from distant exoplanets.
We may need to cross a sensitivity threshold before a Cambrian Explosion of discoveries can be made. Alternatively, if we really are alone, we should reflect on the unique beauty and fragility of life here on Earth.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: ESO/José Francisco Salgado
Last summer, the biggest four-day work week trial in the world kicked off in the UK. 3,300 people started working 80 percent of their regular hours for 100 percent of their pay. Feedback from employees and companies was overwhelmingly positive; people felt they were more productive and less stressed, and some businesses even saw their financial performance improve.
Meanwhile, a similar trial was taking place in the US and other English-speaking countries (Australia, Ireland, the UK, New Zealand, and Canada), with 903 employees across 33 companies getting a day of the week back in exchange for consistent work output. This pilot was also a resounding success, with 96.9 percent of participants voting to stick with a four-day week rather than going back to five days. Employees’ self-assessed work performance improved, as did their “satisfaction across multiple domains of life.”
The message is clear: a four-day work week works. People like it. Companies like it. Everyone’s happier, and there’s no decrease in productivity or hit to financial performance. So now that we’re all in agreement, what comes next?
The state of Maryland is the first in the US to take a step towards standardizing the four-day week. A proposed bill would give tax credits to companies that implement a 32-hour work week without reducing their employees’ pay. They’d get credits of $750,000 per year for up to two years if they have at least 30 employees scale down to a shorter work week.
The tax credit would be used in part to help businesses cover the cost of collecting data about the trial and reporting it to the state. The state would have to pay the cost of administering the program, which could be as much as $250,000 a year.
So what’s in it for the state? It seems a bit counter-intuitive for a state government to incentivize its citizens to work less. What about growing the economy and staying competitive?
As we’ve unfortunately learned through the chaotic labor market of the last couple years, it’s hard to grow the economy when millions of people are unhappy with their jobs and voluntarily leave them. The instability and worker shortages brought by this state of affairs must be more harmful than working one less day per week—especially if that one day is making a difference in employee satisfaction.
That’s job satisfaction and overall life satisfaction. Less time behind a desk means more time doing whatever you please, be it spending time with family, exercising, or working on personal projects—and ideally, that means a happier you, one who’s more motivated to perform at work and less likely to quit in a flurry of frustration and stress.
“We have a real opportunity here to create a win-win,” said Vaughn Stewart, the Maryland state delegate who sponsored the bill in the House after learning about the global trial. “We can make a shift toward reducing working hours without harming productivity, and possibly even boosting companies’ bottom line because they not only have improved productivity but retention and recruitment.”
The Maryland Legislature will hold hearings on the bill this month. If it passes, it would be the first of its kind in the US, and would be the first official change to the work week since 1940, when the federal government changed the minimum standard from 44 hours to 40.
Stewart is cautiously optimistic, noting that he’s gotten more interest in this bill than in all the other bills he’s sponsored combined since he became a member of Maryland’s House of Delegates four years ago.
If it’s signed into law, Maryland’s four-day work week pilot would go into effect on July 1.
Image Credit: David from Pixabay
Flight. Invisibility. Mind-reading. Super-strength. These powers have mostly been limited to the realms of science fiction and fantasy, though we’re starting to see robots and computers replicate some of them. Now a small robot built by an international team has a new superpower: shape-shifting. Or perhaps a more accurate name would be… state-shifting.
Described in a paper published last week in Matter, the robot can go from a solid state to a liquid state based on manipulation of the magnetic fields around it. The team developed the robot as an attempt to get the best of both worlds in terms of robotic properties and capabilities. Hard robots often can’t access certain spaces because of their inflexible bodies, while flexible robots lack strength and durability. Why not make a bot that can do it all?
Credit: Wang and Pan et al. under CC BY-SAVideos show the robot “escaping” from a cage and extracting a ball from a model of a human stomach. The researchers say it could have all kinds of real-world applications, from performing tasks in tight spaces (like soldering a circuit board) to accessing parts of the body that are hard to reach (like the inside of the intestines) to acting as a universal screw by melting and reforming into a screw socket.
The robot is made primarily of gallium, a soft, silvery metal that’s used in electronic circuits, semiconductors, and LEDs. Its most useful feature in this case is its very low melting point: gallium melts at a cool 85.57 degrees Fahrenheit (29.76 degrees Celsius). That’s just slightly above room temperature (in a warm room, admittedly), or the outdoor temperature on a midsummer day.
The team sprinkled magnetic particles throughout the gallium, and these are key to the robot’s functionality.
“The magnetic particles here have two roles,” said the paper’s senior author and mechanical engineer Carmel Majidi of Carnegie Mellon University. “One is that they make the material responsive to an alternating magnetic field, so you can, through induction, heat up the material and cause the phase change. But the magnetic particles also give the robots mobility and the ability to move in response to the magnetic field.”
Gallium’s low melting point meant that exposing it to a rapidly-changing magnetic field generated enough electricity within the metal to cause it to heat up and melt. In the “prison break” experiment the researchers set up, the robot escaped its cell and re-solidified into its original form on the other side. It should be noted that the bot isn’t yet able to re-assume its original form without help, though; there was a mold waiting for it outside the cell.
Despite not having quite reached Terminator status, the ease with which the robot’s state can be manipulated could give it a major advantage over existing phase-shifting materials, which tend to require heat guns or electrical currents to go from solid to liquid. The gallium-based bot is also more fluid in its liquid form than similar materials.
When they used it in a model of a human stomach to remove an object, the solid robot was able to move quickly to the object, melt down, surround the object, coalesce back into a solid, and move out of the stomach with the object. The team noted that although the robot worked well in the model, pure gallium would quickly melt inside a real human body; they’d have to add metals like bismuth and tin to raise the material’s melting point for use in biomedical applications.
Manipulated by magnetic fields, the robot removes a foreign object from a model human stomach. Credit: Wang and Pan et al. under CC BY-SA“What we’re showing are just one-off demonstrations, proofs of concept, but much more study will be required to delve into how this could actually be used for drug delivery or for removing foreign objects,” said Majidi.
The team got some of their inspiration for the robot from sea cucumbers, which can quickly change their stiffness back and forth. They call their invention a “magnetoactive solid-liquid phase transitional machine.” Using magnetic fields, the robots were also able to jump over moats, climb walls, and support heavy weight.
The next step is for the team to search for more real-world applications for their technology, and tweak its properties accordingly. Chengfeng Pan, an engineer at the Chinese University of Hong Kong who led the study, said, “Now we’re pushing this material system in more practical ways to solve some very specific medical and engineering problems.”
Image Credit: Q. Wang et al/Matter 2023 (CC BY-SA)
We speak at a rate of roughly 160 words every minute. That speed is incredibly difficult to achieve for speech brain implants.
Decades in the making, speech implants use tiny electrode arrays inserted into the brain to measure neural activity, with the goal of transforming thoughts into text or sound. They’re invaluable for people who lose their ability to speak due to paralysis, disease, or other injuries. But they’re also incredibly slow, slashing word count per minute nearly ten-fold. Like a slow-loading web page or audio file, the delay can get frustrating for everyday conversations.
A team led by Drs. Krishna Shenoy and Jaimie Henderson at Stanford University is closing that speed gap.
Published on the preprint server bioRxiv, their study helped a 67-year-old woman restore her ability to communicate with the outside world using brain implants at a record-breaking speed. Known as “T12,” the woman gradually lost her speech from amyotrophic lateral sclerosis (ALS), or Lou Gehrig’s disease, which progressively robs the brain’s ability to control muscles in the body. T12 could still vocalize sounds when trying to speak—but the words came out unintelligible.
With her implant, T12’s attempts at speech are now decoded in real time as text on a screen and spoken aloud with a computerized voice, including phrases like “it’s just tough,” or “I enjoy them coming.” The words came fast and furious at 62 per minute, over three times the speed of previous records.
It’s not just a need for speed. The study also tapped into the largest vocabulary library used for speech decoding using an implant—at roughly 125,000 words—in a first demonstration on that scale.
To be clear, although it was a “big breakthrough” and reached “impressive new performance benchmarks” according to experts, the study hasn’t yet been peer-reviewed and the results are limited to the one participant.
That said, the underlying technology isn’t limited to ALS. The boost in speech recognition stems from a marriage between RNNs—recurrent neural networks, a machine learning algorithm previously effective at decoding neural signals—and language models. When further tested, the setup could pave the way to enable people with severe paralysis, stroke, or locked-in syndrome to casually chat with their loved ones using just their thoughts.
We’re beginning to “approach the speed of natural conversation,” the authors said.
Loss for WordsThe team is no stranger to giving people back their powers of speech.
As part of BrainGate, a pioneering global collaboration for restoring communications using brain implants, the team envisioned—and then realized—the ability to restore communications using neural signals from the brain.
In 2021, they engineered a brain-computer interface (BCI) that helped a person with spinal cord injury and paralysis type with his mind. With a 96 microelectrode array inserted into the motor areas of the patient’s brain, the team was able to decode brain signals for different letters as he imagined the motions for writing each character, achieving a sort of “mindtexting” with over 94 percent accuracy.
The problem? The speed was roughly 90 characters per minute at most. While a large improvement from previous setups, it was still painfully slow for daily use.
So why not tap directly into the speech centers of the brain?
Regardless of language, decoding speech is a nightmare. Small and often subconscious movements of the tongue and surrounding muscles can trigger vastly different clusters of sounds—also known as phonemes. Trying to link the brain activity of every single twitch of a facial muscle or flicker of the tongue to a sound is a herculean task.
Hacking SpeechThe new study, a part of the BrainGate2 Neural Interface System trial, used a clever workaround.
The team first placed four strategically located electrode microarrays into the outer layer of T12’s brain. Two were inserted into areas that control movements around the mouth’s surrounding facial muscles. The other two tapped straight into the brain’s “language center,” which is called Broca’s area.
In theory, the placement was a genius two-in-one: it captured both what the person wanted to say, and the actual execution of speech through muscle movements.
But it was also a risky proposition: we don’t yet know whether speech is limited to just a small brain area that controls muscles around the mouth and face, or if language is encoded at a more global scale inside the brain.
Enter RNNs. A type of deep learning, the algorithm has previously translated neural signals from the motor areas of the brain into text. In a first test, the team found that it easily separated different types of facial movements for speech—say, furrowing the brows, puckering the lips, or flicking the tongue—based on neural signals alone with over 92 percent accuracy.
The RNN was then taught to suggest phonemes in real time—for example, “huh,” “ah,” and “tze.” Phenomes help distinguish one word from another; in essence, they’re the basic element of speech.
The training took work: every day, T12 attempted to speak between 260 and 480 sentences at her own pace to teach the algorithm the particular neural activity underlying her speech patterns. Overall, the RNN was trained on nearly 11,000 sentences.
Having a decoder for her mind, the team linked the RNN interface with two language models. One had an especially large vocabulary at 125,000 words. The other was a smaller library with 50 words that’s used for simple sentences in everyday life.
After five days of attempted speaking, both language models could decode T12’s words. The system had errors: around 10 percent for the small library and nearly 24 percent for the larger one. Yet when asked to repeat sentence prompts on a screen, the system readily translated her neural activity into sentences three times faster than previous models.
The implant worked regardless if she attempted to speak or if she just mouthed the sentences silently (she preferred the latter, as it required less energy).
Analyzing T12’s neural signals, the team found that certain regions of the brain retained neural signaling patterns to encode for vowels and other phonemes. In other words, even after years of speech paralysis, the brain still maintains a “detailed articulatory code”—that is, a dictionary of phonemes embedded inside neural signals—that can be decoded using brain implants.
Speak Your MindThe study builds upon many others that use a brain implant to restore speech, often decades after severe injuries or slowly-spreading paralysis from neurodegenerative disorders. The hardware is well known: the Blackrock microelectrode array, consisting of 64 channels to listen in on the brain’s electrical signals.
What’s different is how it operates; that is, how the software transforms noisy neural chatter into cohesive meanings or intentions. Previous models mostly relied on decoding data directly obtained from neural recordings from the brain.
Here, the team tapped into a new resource: language models, or AI algorithms similar to the autocomplete function now widely available for Gmail or texting. The technological tag-team is especially promising with the rise of GPT-3 and other emerging large language models. Excellent at generating speech patterns from simple prompts, the tech—when combined with the patient’s own neural signals—could potentially “autocomplete” their thoughts without the need for hours of training.
The prospect, while alluring, comes with a side of caution. GPT-3 and similar AI models can generate convincing speech on their own based on previous training data. For a person with paralysis who’s unable to speak, we would need guardrails as the AI generates what the person is trying to say.
The authors agree that, for now, their work is a proof of concept. While promising, it’s “not yet a complete, clinically viable system,” for decoding speech. For one, they said, we need to train the decoder with less time and make it more flexible, letting it adapt to ever-changing brain activity. For another, the error rate of roughly 24 percent is far too high for everyday use—although increasing the number of implant channels could boost accuracy.
But for now, it moves us closer to the ultimate goal of “restoring rapid communications to people with paralysis who can no longer speak,” the authors said.
Image Credit: Miguel Á. Padriñán from Pixabay
Last July, New York-based startup MyForest foods announced the opening of a vertical farm that would grow three million pounds of mycelium a year, all for plant-based bacon. Now competitor Meati Foods is blowing them out of the water with a facility that will be able to produce more than 45 million pounds of product once it’s fully scaled up. The company announced the opening of a factory it’s calling “Mega Ranch” in Thornton, Colorado (a suburb north of Denver) last week.
Meati makes a variety of plant-based imitation meat products, or “animal-free whole-food proteins,” including a classic steak, carne asada, a classic cutlet, and a crispy cutlet. The meats are made of 95 percent mushroom root, with additional ingredients including oat fiber, seasonings, fruit and vegetable juices, and lycopene (for color). With up to 17 grams of protein and 12 grams of dietary fiber per serving, the company says the meats are comparable to their animal-derived counterparts in nutritional value.
Mushroom roots are called mycelium, and they’re a different sort of root than what you typically see at the bottom of most plants and trees. Mycelium is a root-like structure of fungus made of a mass of branching, thread-like strands called hyphae. The hyphae absorb nutrients from soil or another substrate so the fungus can grow.
Companies are using mycelium as a base for all sorts of vegan materials, from packaging to leather to biomedical scaffolds. It’s a viable ingredient both because it’s easy to manipulate—the nutrients in the substrate it’s grown on can be tweaked to yield different properties, like making it stiffer or more flexible—and because it grows fast; Meati says its proprietary growth formula can turn a teaspoon of spores into the equivalent of hundreds of cows’ worth of whole-food protein in just a few days.
The mycelium is grown in stainless steel vats (similar to fermentation tanks at breweries), where it’s fed a liquid rich in sugar and nutrients that helps it grow faster than it would in the wild. Meati harvests mycelium fibers from the vats, then must assemble them in such a way that the texture resembles animal muscle.
Meati’s new Colorado plant will occupy 100,000 square feet, and will enable the company to produce tens of millions of pounds of its products by the end of this year. The products are already sold through retail and foodservice partners that include Sprouts Farmers Market, Sweetgreen, and Birdcall, and Meati’s aiming to get start selling at 7,000 new locations by the end of this year.
The company’s total funding to date is over $250 million. They expect to bring in tens of millions in revenue this year and hundreds of millions in 2024. Despite the Mega Ranch opening this year, they’re already scouting out a location for a “Giga Ranch” that will be able to produce hundreds of millions of pounds of product annually.
Despite the somewhat ailing state of the plant-based meat industry, Meati’s co-founder and CEO, Tyler Huggins, sees nothing but growth in his company’s future. “There is no shortage of stuff coming out, and we have no lack of demand,” he told TechCrunch. “Our pipeline is robust, and everything we produce in the next year or more is already pre-sold. It’s now about unlocking capacity to get the product out there.”
Image Credit: Meati
Technology can change the world in ways that are unimaginable, until they happen. Switching on an electric light would have been unimaginable for our medieval ancestors. In their childhood, our grandparents would have struggled to imagine a world connected by smartphones and the internet.
Similarly, it is hard for us to imagine the arrival of all those technologies that will fundamentally change the world we are used to.
We can remind ourselves that our own future might look very different from the world today by looking back at how rapidly technology has changed our world in the past. That’s what this article is about.
One insight I take away from this long-term perspective is how unusual our time is. Technological change was extremely slow in the past—the technologies that our ancestors got used to in their childhood were still central to their lives in their old age. In stark contrast to those days, we live in a time of extraordinarily fast technological change. For recent generations, it was common for technologies that were unimaginable in their youth to become common later in life.
The Long-Run Perspective on Technological ChangeThe big visualization offers a long-term perspective on the history of technology.1
The timeline begins at the center of the spiral. The first use of stone tools, 3.4 million years ago, marks the beginning of this history of technology.2 Each turn of the spiral then represents 200,000 years of history. It took 2.4 million years—12 turns of the spiral—for our ancestors to control fire and use it for cooking.3
To be able to visualize the inventions in the more recent past—the last 12,000 years—I had to unroll the spiral. I needed more space to be able to show when agriculture, writing, and the wheel were invented. During this period, technological change was faster, but it was still relatively slow: several thousand years passed between each of these three inventions.
From 1800 onwards, I stretched out the timeline even further to show the many major inventions that rapidly followed one after the other.
The long-term perspective that this chart provides makes it clear just how unusually fast technological change is in our time.
You can use this visualization to see how technology developed in particular domains. Follow, for example, the history of communication: from writing, to paper, to the printing press, to the telegraph, the telephone, the radio, all the way to the internet and smartphones.
Or follow the rapid development of human flight. In 1903, the Wright brothers took the first flight in human history (they were in the air for less than a minute), and just 66 years later, we landed on the moon. Many people saw both within their lifetimes: the first plane and the moon landing.
This large visualization also highlights the wide range of technology’s impact on our lives. It includes extraordinarily beneficial innovations, such as the vaccine that allowed humanity to eradicate smallpox, and it includes terrible innovations, like the nuclear bombs that endanger the lives of all of us.
What will the next decades bring?
The red timeline reaches up to the present and then continues in green into the future. Many children born today, even without any further increases in life expectancy, will live well into the 22nd century.
New vaccines, progress in clean, low-carbon energy, better cancer treatments—a range of future innovations could very much improve our living conditions and the environment around us. But, as I argue in a series of articles, there is one technology that could even more profoundly change our world: artificial intelligence.
One reason why artificial intelligence is such an important innovation is that intelligence is the main driver of innovation itself. This fast-paced technological change could speed up even more if it’s not only driven by humanity’s intelligence, but artificial intelligence too. If this happens, the change that is currently stretched out over the course of decades might happen within very brief time spans of just a year. Possibly even faster.4
I think AI technology could have a fundamentally transformative impact on our world. In many ways it is already changing our world, as I documented in this companion article. As this technology is becoming more capable in the years and decades to come, it can give immense power to those who control it (and it poses the risk that it could escape our control entirely).
Such systems might seem hard to imagine today, but AI technology is advancing very fast. Many AI experts believe that there is a very real chance that human-level artificial intelligence will be developed within the next decades, as I documented in this article.
Technology Will Continue to Change the World—We Should All Make Sure That It Changes It for the BetterWhat is familiar to us today—photography, the radio, antibiotics, the internet, or the International Space Station circling our planet—was unimaginable to our ancestors just a few generations ago. If your great-great-great grandparents could spend a week with you they would be blown away by your everyday life.
What I take away from this history is that I will likely see technologies in my lifetime that appear unimaginable to me today.
In addition to this trend towards increasingly rapid innovation, there is a second long-run trend. Technology has become increasingly powerful. While our ancestors wielded stone tools, we are building globe-spanning AI systems and technologies that can edit our genes.
Because of the immense power that technology gives those who control it, there is little that is as important as the question of which technologies get developed during our lifetimes. Therefore I think it is a mistake to leave the question about the future of technology to the technologists. Which technologies are controlled by whom is one of the most important political questions of our time, because of the enormous power that these technologies convey to those who control them.
We all should strive to gain the knowledge we need to contribute to an intelligent debate about the world we want to live in. To a large part this means gaining the knowledge, and wisdom, on the question of which technologies we want.
Acknowledgements: I would like to thank my colleagues Hannah Ritchie, Bastian Herre, Natasha Ahuja, Edouard Mathieu, Daniel Bachler, Charlie Giattino, and Pablo Rosado for their helpful comments to drafts of this essay and the visualization. Thanks also to Lizka Vaintrob and Ben Clifford for a conversation that initiated this visualization.
This article was originally published on Our World in Data and has been republished here under a Creative Commons license. Read the original article.
Image Credit: Pat Kay / Unsplash
ARTIFICIAL INTELLIGENCEAI Has Designed Bacteria-Killing Proteins From Scratch—and They Work
Karmela Padavic-Callaghan | New Scientist“The AI, called ProGen, works in a similar way to AIs that can generate text. ProGen learned how to generate new proteins by learning the grammar of how amino acids combine to form 280 million existing proteins. Instead of the researchers choosing a topic for the AI to write about, they could specify a group of similar proteins for it to focus on. In this case, they chose a group of proteins with antimicrobial activity.”
DIGITAL MEDIABuzzFeed to Use ChatGPT Creator OpenAI to Help Create Quizzes and Other Content
Alexandra Bruell | The Wall Street Journal“BuzzFeed Inc. said it would rely on ChatGPT creator OpenAI to enhance its quizzes and personalize some content for its audiences, becoming the latest digital publisher to embrace artificial intelligence. In a memo to staff sent Thursday morning, which was reviewed by The Wall Street Journal, Chief Executive Jonah Peretti said he intends for AI to play a larger role in the company’s editorial and business operations this year.“
ROBOTICSMetal Robot Can Melt Its Way Out of Tight Spaces to Escape
Karmela Padavic-Callaghan | New Scientist“A miniature, shape-shifting robot can liquefy itself and reform, allowing it to complete tasks in hard-to-access places and even escape cages. It could eventually be used as a hands-free soldering machine or a tool for extracting swallowed toxic items.”
FUTUREDon’t Be Sucked in by AI’s Head-Spinning Hype Cycles
Devin Coldewey | TechCrunch“[AI] certainly can outplay any human at chess or go, and it can predict the structure of protein chains; it can answer any question confidently (if not correctly) and it can do a remarkably good imitation of any artist, living or dead. But it is difficult to tease out which of these things is important, and to whom, and which will be remembered as briefly diverting parlor tricks in 5 or 10 years, like so many innovations we have been told are going to change the world.”
SPACENASA Announces Successful Test of New Propulsion Technology for Treks to Deep Space
Kevin Hurler | Gizmodo“The rotating detonation rocket engine, or RDRE, generates thrust with detonation, in which a supersonic exothermic front accelerates to produce thrust, much the same way a shockwave travels through the atmosphere after something like TNT explodes. NASA says that this design uses less fuel and provides more thrust than current propulsion systems and that the RDRE could be used to power human landers, as well as crewed missions to the Moon, Mars, and deep space.“
ARTIFICIAL INTELLIGENCEThe Best Use for AI Eye Contact Tech Is Making Movie Stars Look Straight at the Camera
James Vincent | The Verge“This tech comes with a bunch of interesting questions, of course. Like: is constant unbroken eye contact good or a bit creepy? Are these tools useful for people who don’t naturally like eye contact? …But forget that high-brow trash for now, because here’s the stupidest and best use case of this technology yet: editing movie scenes so actors make eye contact with the camera.”
SCIENCEResearchers Look a Dinosaur in Its Remarkably Preserved Face
Jeanne Timmons | Ars Technica“Borealopelta markmitchelli found its way back into the sunlight in 2017, millions of years after it had died. This armored dinosaur is so magnificently preserved that we can see what it looked like in life. Almost the entire animal—the skin, the armor that coats its skin, the spikes along its side, most of its body and feet, even its face—survived fossilization. It is, according to Dr. Donald Henderson, curator of dinosaurs at the Royal Tyrrell Museum, a one-in-a-billion find.”
TECHGoogle, Not OpenAI, Has the Most to Gain From Generative AI
Mark Sullivan | Fast Company“After spending billions on artificial intelligence R&D and acquisitions, Google finds itself ceding the AI limelight to OpenAI, an upstart that has captured the popular imagination with the public beta of its startlingly conversant chatbot, ChatGPT. Now Google reportedly fears the ChatGPT AI could reinvent search, its cornerstone business. But Google, which declared itself an ‘AI-first’ company in 2017, may yet regain its place in the sun. Its AI investments, which date back to the 2000s, may pay off, and could even power the company’s next quarter century of growth (Google turns 25 this year). Here’s why.”
BIOTECHCRISPR Wants to Feed the World
Jennifer Doudna | Wired“A great deal of the attention surrounding CRISPR has focused on the medical applications, and for good reason: The results are promising, and the personal stories are uplifting, offering hope to many who have suffered from long-neglected genetic diseases. In 2023, as CRISPR moves into agriculture and climate, we will have the opportunity to radically improve human health in a holistic way that can better safeguard our society and enable millions of people around the world to flourish.“
ETHICSA Watermark for Chatbots Can Expose Text Written by an AI
Melissa Heikkilä | MIT Technology Review“Hidden patterns purposely buried in AI-generated texts could help identify them as such, allowing us to tell whether the words we’re reading are written by a human or not. These ‘watermarks’ are invisible to the human eye but let computers detect that the text probably comes from an AI system. If embedded in large language models, they could help prevent some of the problems that these models have already caused.”
SCIENCEEarth’s Inner Core: A Shifting, Spinning Mystery’s Latest Twist
Dennis Overbye | The New York Times“Imagine Earth’s inner core—the dense center of our planet—as a heavy, metal ballerina. This iron-rich dancer is capable of pirouetting at ever-changing speeds. That core may be on the cusp of a big shift. Seismologists reported Monday in the journal Nature Geoscience that after brief but peculiar pauses, the inner core changes how it spins—relative to the motion of Earth’s surface—perhaps once every few decades. And, right now, one such reversal may be underway.”
Image Credit: Robert Linder / Unsplash
Asteroid mining has long caught the imagination of space entrepreneurs, but conventional wisdom has always been that it’s little more than a pipe dream. That may be about to change after a startup announced plans to launch two missions this year designed to validate its space mining technology.
There are estimated to be trillions of dollars worth of precious metals locked up in asteroids strewn throughout the solar system. Given growing concerns about the scarcity of key materials required for batteries and other electronics, there’s been growing interest in attempts to extract these resources.
The enormous cost of space missions and the huge technical challenges involved in mining in space have led many to dismiss the idea as unworkable. The industry has already seen one boom and bust cycle after leading players like Deep Space Industries folded after investors lost their nerve.
But now, California-based startup AstroForge has taken concrete steps toward its goal of becoming the first company to mine an asteroid and bring the materials back to Earth. This year it will launch two missions, one designed to test out its in-space mineral extraction technology and another that will carry out a survey mission of a promising asteroid close to Earth.
“With a finite supply of precious metals on Earth, we have no other choice than to look to deep space to source cost-effective and sustainable materials,” CEO and co-founder Matt Gialich said in a statement.
The company, which raised $13 million in seed funding last April, is planning to target asteroids rich in platinum group metals in deep space. These materials are in major demand in many high-tech industries, but their reserves are limited and geographically concentrated. Extracting them can also be very environmentally damaging.
AstroForge is developing mineral refining technology that it hopes will allow it to extract precious metals from these asteroids and return them to Earth. A prototype will catch a lift into orbit on a spacecraft designed by OrbAstro and launched by a SpaceX Falcon 9 rocket in April. It will be pre-loaded with asteroid-like material, which it will then attempt to vaporize and sort into its different chemical constituents.
Then in October, the company will attempt an even more ambitious mission. A 220-pound spacecraft also designed by OrbAstro, called Brokkr-2, will attempt an 8-month journey to reach an asteroid orbiting the sun about 22 million miles from Earth. It will carry a host of instruments designed to assess the target asteroid in situ.
Both of these missions are precursors designed to test out systems that will be needed for AstroForge’s first proper asteroid mining mission, expected later this decade. The company plans to target asteroids between 66 to 4,920 feet in diameter and break them apart from a distance before collecting the remains.
Even if these missions are a success, there’s still a long road towards making space mining practical. According to research AstroForge recently conducted with the Colorado School of Mines, the bulk of metal-rich asteroids are found in the asteroid belt between Mars and Jupiter, which is currently a 14-year round trip.
Nonetheless, off-world mining does appear to be having somewhat of a renaissance, with dozens of space resources startups springing up in recent years. If AstroForge succeeds in proving out its technology this year, it could give this fledgling industry a major boost.
Image Credit: NASA
Even if you think you are good at analyzing faces, research shows many people cannot reliably distinguish between photos of real faces and images that have been computer-generated. This is particularly problematic now that computer systems can create realistic-looking photos of people who don’t exist.
A few years ago, a fake LinkedIn profile with a computer-generated profile picture made the news because it successfully connected with US officials and other influential individuals on the networking platform, for example. Counter-intelligence experts even say that spies routinely create phantom profiles with such pictures to home in on foreign targets over social media.
These deepfakes are becoming widespread in everyday culture which means people should be more aware of how they’re being used in marketing, advertising, and social media. The images are also being used for malicious purposes, such as political propaganda, espionage, and information warfare.
Making them involves something called a deep neural network, a computer system that mimics the way the brain learns. This is “trained” by exposing it to increasingly large data sets of real faces.
In fact, two deep neural networks are set against each other, competing to produce the most realistic images. As a result, the end products are dubbed GAN images, where GAN stands for “generative adversarial networks.” The process generates novel images that are statistically indistinguishable from the training images.
In a study published in iScience, my colleagues and I showed that a failure to distinguish these artificial faces from the real thing has implications for our online behavior. Our research suggests the fake images may erode our trust in others and profoundly change the way we communicate online.
We found that people perceived GAN faces to be even more real-looking than genuine photos of actual people’s faces. While it’s not yet clear why this is, this finding does highlight recent advances in the technology used to generate artificial images.
And we also found an interesting link to attractiveness: faces that were rated as less attractive were also rated as more real. Less attractive faces might be considered more typical, and the typical face may be used as a reference against which all faces are evaluated. Therefore, these GAN faces would look more real because they are more similar to mental templates that people have built from everyday life.
But seeing these artificial faces as authentic may also have consequences for the general levels of trust we extend to a circle of unfamiliar people—a concept known as “social trust.”
We often read too much into the faces we see, and the first impressions we form guide our social interactions. In a second experiment that formed part of our latest study, we saw that people were more likely to trust information conveyed by faces they had previously judged to be real, even if they were artificially generated.
It is not surprising that people put more trust in faces they believe to be real. But we found that trust was eroded once people were informed about the potential presence of artificial faces in online interactions. They then showed lower levels of trust, overall—independently of whether the faces were real or not.
This outcome could be regarded as useful in some ways, because it made people more suspicious in an environment where fake users may operate. From another perspective, however, it may gradually erode the very nature of how we communicate.
In general, we tend to operate on a default assumption that other people are basically truthful and trustworthy. The growth in fake profiles and other artificial online content raises the question of how much their presence and our knowledge about them can alter this “truth default” state, eventually eroding social trust.
Changing Our DefaultsThe transition to a world where what’s real is indistinguishable from what’s not could also shift the cultural landscape from being primarily truthful to being primarily artificial and deceptive.
If we are regularly questioning the truthfulness of what we experience online, it might require us to re-deploy our mental effort from the processing of the messages themselves to the processing of the messenger’s identity. In other words, the widespread use of highly realistic, yet artificial, online content could require us to think differently—in ways we hadn’t expected to.
In psychology, we use a term called “reality monitoring” for how we correctly identify whether something is coming from the external world or from within our brains. The advance of technologies that can produce fake, yet highly realistic, faces, images, and video calls means reality monitoring must be based on information other than our own judgments. It also calls for a broader discussion of whether humankind can still afford to default to truth.
It’s crucial for people to be more critical when evaluating digital faces. This can include using reverse image searches to check whether photos are genuine, being wary of social media profiles with little personal information or a large number of followers, and being aware of the potential for deepfake technology to be used for nefarious purposes.
The next frontier for this area should be improved algorithms for detecting fake digital faces. These could then be embedded in social media platforms to help us distinguish the real from the fake when it comes to new connections’ faces.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: The faces in this article’s banner image may look realistic, but they were generated by a computer. NVIDIA via thispersondoesnotexist.com
Ten years ago, a little-known bacterial defense mechanism skyrocketed to fame as a powerful genome editor. In the decade since, CRISPR-Cas9 has spun off multiple variants, expanding into a comprehensive toolbox that can edit the genetic code of life.
Far from an ivory tower pursuit, its practical uses in research, healthcare, and agriculture came fast and furious.
You’ve seen the headlines. The FDA approved its use in tackling the underlying genetic mutation for sickle cell disease. Some researchers edited immune cells to fight untreatable blood cancers in children. Others took pig-to-human organ transplants from dream to reality in an attempt to alleviate the shortage of donor organs. Recent work aims to help millions of people with high cholesterol—and potentially bring CRISPR-based gene therapy to the masses—by lowering their chances of heart disease with a single injection.
But to Dr. Jennifer Doudna, who won the Nobel Prize in 2020 for her role in developing CRISPR, we’re just scratching the surface of its potential. Together with graduate student Joy Wang, Doudna laid out a roadmap for the technology’s next decade in an article in Science.
If the 2010s were focused on establishing the CRISPR toolbox and proving its effectiveness, this decade is when the technology reaches its full potential. From CRISPR-based therapies and large-scale screens for disease diagnostics to engineering high-yield crops and nutritious foods, the technology “and its potential impact are still in their early stages,” the authors wrote.
A Decade of HighlightsWe’ve spilt plenty of ink on CRISPR advances, but it pays to revisit the past to predict the future—and potentially scout out problems along the way.
One early highlight was CRISPR’s incredible ability to rapidly engineer animal models of disease. Its original form easily snips away a targeted gene in a very early embryo, which when transplanted into a womb can generate genetically modified mice in just a month, compared to a year using previous methods. Additional CRISPR versions, such as base editing—swapping one genetic letter for another—and prime editing—which snips the DNA without cutting both strands—further boosted the toolkit’s flexibility at engineering genetically-altered organoids (think mini-brains) and animals. CRISPR rapidly established dozens of models for some of our most devasting and perplexing diseases, including various cancers, Alzheimer’s, and Duchenne muscular dystrophy—a degenerative disorder in which the muscle slowly wastes away. Dozens of CRISPR-based trials are now in the works.
CRISPR also accelerated genetic screening into the big data age. Rather than targeting one gene at a time, it’s now possible to silence, or activate, thousands of genes in parallel, forming a sort of Rosetta stone for translating genetic perturbations into biological changes. This is especially important for understanding genetic interactions, such as those in cancer or aging that we weren’t previously privy to, and gaining new ammunition for drug development.
But a crowning achievement for CRISPR was multiplexed editing. Like simultaneously tapping on multiple piano keys, this type of genetic engineering targets multiple specific DNA areas, rapidly changing a genome’s genetic makeup in one go.
The technology works in plants and animals. For eons, people have painstakingly bred crops with desirable features—be it color, size, taste, nutrition, or disease resilience. CRISPR can help select for multiple traits or even domesticate new crops in just one generation. CRISPR-generated hornless bulls, nutrient rich tomatoes, and hyper-muscular farm animals and fish are already reality. With the world population hitting 8 billion in 2022 and millions suffering from hunger, CRISPRed-crops may lend a lifeline—that is, if people are willing to accept the technology.
The Path ForwardWhere do we go from here?
To the authors, we need to further boost CRISPR’s effectiveness and build trust. This means going back to the basics to increase the tool’s editing accuracy and precision. Here, platforms to rapidly evolve Cas enzymes, the “scissor” component of the CRISPR machinery, are critical.
There have already been successes: one Cas version, for example, acts as a guardrail for the targeting component—the sgRNA “bloodhound.” In classic CRISPR, the sgRNA works alone, but in this updated version, it struggles to bind without Cas assistance. This trick helps tailor the edit to a specific DNA site and increases accuracy so the cut works as predicted.
Similar strategies can also boost precision with fewer side effects or insert new genes in cells such as neurons and others that no longer divide. While already possible with prime editing, its efficiency can be 30 times lower than classic CRISPR mechanisms.
“A main goal for prime editing in the next decade is improving efficiency without compromising editing product purity—an outcome that has the potential to turn prime editing into one of the most versatile tools for precision editing,” the authors said.
But perhaps more important is delivery, which remains a bottleneck especially for therapeutics. Currently, CRISPR is generally used on cells outside the body that are infused back—as in the case of CAR-T—or in some cases, tethered to a viral carrier or encapsulated in fatty bubbles and injected into the body. There have been successes: in 2021, the FDA approved the first CRISPR-based shot to tackled a genetic blood disease, transthyretin amyloidosis.
Yet both strategies are problematic: not many types of cells can survive the CAR-T treatment—dying when reintroduced into the body—and targeting specific tissues and organs remains mostly out of reach for injectable therapies.
A key advance for the next decade, the authors said, is to shuttle the CRISPR cargo into the targeted tissue without harm and release the gene editor at its intended spot. Each of these steps, though seemingly simple on paper, presents its own set of challenges that will require both bioengineering and innovation to overcome.
Finally, CRISPR can synergize with other technological advances, the authors said. For example, by tapping into cell imaging and machine learning, we could soon engineer even more efficient genome editors. Thanks to faster and cheaper DNA sequencing, we can then easily monitor gene-editing consequences. These data can then provide a kind of feedback mechanism with which to engineer even more powerful genome editors in a virtuous loop.
Real-World ImpactAlthough further expanding the CRISPR toolbox is on the agenda, the technology is sufficiently mature to impact the real world in its second decade, the authors said.
In the near future, we should see “an increased number of CRISPR-based treatments moving to later stages of clinical trials.” Looking further ahead, the technology, or its variants, could make pig-to-human organ xenotransplants routine, rather than experimental. Large-scale screens for genes that lead to aging or degenerative brain or heart diseases—our top killers today—could yield prophylactic CRISPR-based treatments. It’s no easy task: we need both knowledge of the genetics underlying multifaceted genetic diseases—that is, when multiple genes come into play—and a way to deliver the editing tools to their target. “But the potential benefits may drive innovation in these areas well beyond what is possible today,” the authors said.
Yet with greater power comes greater responsibility. CRISPR has advanced at breakneck speed, and regulatory agencies and the public are still struggling to catch up. Perhaps the most notorious example was that of the CRISPR babies, where experiments carried out against global ethical guidelines propelled an international consortium to lay down a red line for human germ-cell editing.
Similarly, genetically modified organisms (GMOs) remain a controversial topic. Although CRISPR is far more precise than previous genetic tools, it’ll be up to consumers to decide whether to welcome a new generation of human-evolved foods—both plant and animal.
These are important conversations that need global discourse as CRISPR enters its second decade. But to the authors, the future looks bright.
“Just as during the advent of CRISPR genome editing, a combination of scientific curiosity and the desire to benefit society will drive the next decade of innovation in CRISPR technology,” they said. “By continuing to explore the natural world, we will discover what cannot be imagined and put it to real-world use for the benefit of the planet.”
Image Credit: NIH
Boosting the role of renewables in our electricity supply will require a massive increase in grid-scale energy storage. But new research suggests that electric vehicle batteries could meet short-term storage demands by as soon as 2030.
While solar and wind are rapidly becoming the cheapest source of electricity in many parts of the world, their intermittency is a significant problem. One potential solution is to use batteries to store energy for times when the sun doesn’t shine and the wind doesn’t blow, but building enough capacity to serve entire power grids would be enormously costly.
That’s why people have suggested making use of the huge number of batteries being installed in the ever-growing global fleet of electric vehicles. The idea is that when they’re not on the road, utilities could use these batteries to store excess energy and draw from it when demand spikes.
While there have been some early pilots, so far it has been unclear whether the idea really has legs. Now, a new economic analysis led by researchers at Leiden University in the Netherlands suggests that electric vehicle batteries could play a major role in grid-scale storage in the relatively near future.
There are two main ways that these batteries could aid the renewables transition, according to the team’s study published in Nature Communications. Firstly, so-called vehicle-to-grid technology could make it possible to do smart vehicle charging, only charging cars when power demand is low. It could also make it possible for vehicle owners to temporarily store electricity for utilities for a price.
But old car batteries could also make a significant contribution. Their capacity declines over repeated charge and discharge cycles, and batteries typically become unsuitable for use in electric vehicles by the time they drop to 70 to 80 percent of their original capacity. That’s because they can no longer hold enough power to make up for their added weight. Weight isn’t a problem for grid-scale storage though, so these car batteries can be repurposed.
The researchers note that the lithium-ion batteries used in cars are probably only suitable for short-term storage of under four hours, but this accounts for most of the projected demand. So far though, there hasn’t been a comprehensive study of how large a contribution both current and retired electric vehicle batteries could play in the future of the grid.
To try and fill that gap, the researchers combined data on how many batteries are estimated to be produced over the coming years, how quickly batteries will degrade based on local conditions, and how electric vehicles are likely to be used in different countries—for instance, how many miles people drive in a day and how often they charge.
They found that the total available storage capacity from these two sources by 2050 was likely to be between 32 and 62 terawatt-hours. The authors note that this is significantly higher than the 3.4 to 19.2 terawatt-hours the world is predicted to need by 2050, according to the International Renewable Energy Agency and research group Storage Lab.
However, not every electric vehicle owner is likely to participate in vehicle-to-grid schemes and not all batteries will get repurposed at the end of their lives. So the researchers investigated how different participation rates would impact the ability of electric vehicle batteries to contribute to grid storage.
They found that to meet global demand by 2050, only between 12 and 43 percent of vehicle owners would need to take part in vehicle to grid schemes. If only half of secondhand batteries are used for grid storage, the required participation rates would drop to just 10 percent. In the most optimistic scenarios, electric vehicle batteries could meet demand by 2030.
Lots of factors will impact whether or not this could ever be achieved, including things like how quickly vehicle-to-grid infrastructure can be rolled out, how easy it is to convince vehicle owners to take part, and the economics of recycling car batteries at the end of their lives. The authors note that governments can and should play a role in incentivizing participation and mandating the reuse of old batteries.
But either way, the results suggest there may be a promising alternative to a costly and time-consuming rollout of dedicated grid storage. Electric vehicle owners may soon be doing their part for the environment twice over.
Image Credit: Shutterstock.com/Roman Zaiets
Google is one of the biggest companies on Earth. Google’s search engine is the front door to the internet. And according to recent reports, Google is scrambling.
Late last year, OpenAI, an artificial intelligence company at the forefront of the field, released ChatGPT. Alongside Elon Musk’s Twitter acquisition and fallout from FTX’s crypto implosion, breathless chatter about ChatGPT and generative AI has been ubiquitous.
The chatbot, which was born from an upgrade to OpenAI’s GPT-3 algorithm, is like a futuristic Q&A machine. Ask any question, and it responds in plain language. Sometimes it gets the facts straight. Sometimes not so much. Still, ChatGPT took the world by storm thanks to the fluidity of its prose, its simple interface, and a mainstream launch.
When a new technology hits public consciousness, people try to sort out its impact. Between debates about how bots like ChatGPT will impact everything from academics to journalism, not a few folks have suggestedChatGPT may end Google’s reign in search. Who wants to hunt down information fragmented across a list of web pages when you could get a coherent, seemingly authoritative, answer in an instant?
In December, The New York Times reported Google was taking the prospect seriously, with management declaring a “code red” internally. This week, as Google announced layoffs, CEO Sundar Pichai told employees the company will sharpen its focus on AI. The NYT also reported Google founders, Larry Page and Sergey Brin, are now involved in efforts to streamline development of AI products. The worry is that they’ve lost a step to the competition.
If true, it isn’t due to a lack of ability or vision. Google’s no slouch at AI.
The technology here—a flavor of deep learning model called a transformer—was developed at Google in 2017. The company already has its own versions of all the flashy generative AI models, from images (Imagen) to text (LaMDA). Indeed, in 2021, Google researchers published a paper pondering how large language models (like ChatGPT) might radically upend search in the future.
“What if we got rid of the notion of the index altogether and replaced it with a pre-trained model that efficiently and effectively encodes all of the information contained in the corpus?” Donald Metzler, a Google researcher, and coauthors wrote at the time. “What if the distinction between retrieval and ranking went away and instead there was a single response generation phase?” This should sound familiar.
Whereas smaller organizations opened access to their algorithms more aggressively, however, Google largely kept its work under wraps. Offering only small, tightly controlled demos to limited groups of people, it deemed the tech too risky and error-prone for wider release just yet. Damage to its brand and reputation was a chief concern.
Now, sweating it out under the bright lights of ChatGPT, the company is planning to release some 20 AI-powered products later this year, according to the NYT. These will encompass all the top generative AI applications, like image, text, and code generation—and they’ll test a ChatGPT-like bot in search.
But is the technology ready to go from splashy demo played around with by millions to a crucial tool trusted by billions? In their 2021 paper, the Google researchers suggested an ideal chatbot search assistant would be authoritative, transparent, unbiased, accessible, and contain diverse perspectives. Acing each of those categories is still a stretch for even the most advanced large language models.
Trust matters with search in particular. When it serves up a list of web pages today, Google can blame content creators for poor quality and vow to serve better results in the future. With an AI chatbot, it is the content creator.
As Fast Company’s Harry McCracken pointed out not long ago, if ChatGPT can’t get its facts straight, nothing else matters. “Whenever I chat with ChatGPT about any subject I know much about, such as the history of animation, I’m most struck by how deeply untrustworthy it is,” McCracken wrote. “If a rogue software engineer set out to poison our shared corpus of knowledge by generating convincing-sounding misinformation in bulk, the end result might look something like this.”
Google is clearly aware of the risk. And whatever implementation in search it unveils this year, it still aims to prioritize “getting the facts right, ensuring safety, and getting rid of misinformation.” How it will accomplish these goals is an open question. Just in terms of “ensuring safety,” for example, Google’s algorithms underperform OpenAI’s on metrics of toxicity, according to the NYT. But a Time investigation this week reported that OpenAI had to turn, at least in part, to human workers in Kenya, paid a pittance, to flag and scrub the most toxic data from ChatGPT.
Other questions, including about the copyright of works used to train generative algorithms, remain similarly unresolved. Two copyright lawsuits, one by Getty images and one by a group of artists, were filed earlier this week.
Still, the competitive landscape, it seems, is compelling Google, Microsoft—who has invested big in OpenAI and is already incorporating its algorithms into products—and others to go full steam ahead in an effort to minimize the risk of being left behind. We’ll have to wait and see what an implementation in search looks like. Maybe it’ll be in beta with a disclaimer for awhile, or maybe, as the year progresses, the tech will again surprise us with breakthroughs.
In either case, while generative AI will play a role in search, how much of a role and how soon is less settled. As to whether Google loses its perch? OpenAI’s CEO, Sam Altman, pushed back against the hype this week.
“I think whenever someone talks about a technology being the end of some other giant company, it’s usually wrong,” Altman said in response to a question about the likelihood ChatGPT dethrones Google. “I think people forget they get to make a countermove here, and they’re like pretty smart, pretty competent. I do think there’s a change for search that will probably come at some point—but not as dramatically as people think in the short term.”
Image Credit: D21_Gallery / Unsplash
ARTIFICIAL INTELLIGENCEWhat Happens When AI Has Read Everything?
Ross Andersen | The Atlantic“Artificial intelligence has in recent years proved itself to be a quick study, although it is being educated in a manner that would shame the most brutal headmaster. Locked into airtight Borgesian libraries for months with no bathroom breaks or sleep, AIs are told not to emerge until they’ve finished a self-paced speed course in human culture. On the syllabus: a decent fraction of all the surviving text that we have ever produced.”
GENE EDITINGNext Up for CRISPR: Gene Editing for the Masses?
Jessica Hamzelou | MIT Technology Review“We know the basics of healthy living by now. A balanced diet, regular exercise, and stress reduction can help us avoid heart disease—the world’s biggest killer. But what if you could take a vaccine, too? And not a typical vaccine—one shot that would alter your DNA to provide lifelong protection? That vision is not far off, researchers say. Advances in gene editing, and CRISPR technology in particular, may soon make it possible.”
ETHICSOpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic
Billy Perrigo | Time“ChatGPT’s creator, OpenAI, is now reportedly in talks with investors to raise funds at a $29 billion valuation, including a potential $10 billion investment by Microsoft. That would make OpenAI, which was founded in San Francisco in 2015 with the aim of building superintelligent machines, one of the world’s most valuable AI companies. But the success story is not one of Silicon Valley genius alone. In its quest to make ChatGPT less toxic, OpenAI used outsourced Kenyan laborers earning less than $2 per hour, a TIME investigation has found.”
ROBOTICSBoston Dynamics’ Atlas Robot Grows a Set of Hands, Attempts Construction Work
Ron Amadeo | Ars Technica“Atlas isn’t just clumsily picking things up and carrying them, though. It’s running, jumping, and spinning while carrying heavy objects. At one point it jumps and throws the heavy toolbox up to its construction partner, all without losing balance. It’s doing all this on rickety scaffolding and improvised plank walkways, too, so the ground is constantly moving under Atlas’ feet with every step. Picking up stuff is the start of teaching the robot to do actual work, and it looks right at home on a rough-and-tumble construction site.”
BIOTECHThese Scientists Used CRISPR to Put an Alligator Gene Into Catfish
Jessica Hamzelou | MIT Technology Review“Millions of fish are farmed in the US every year, but many of them die from infections. In theory, genetically engineering fish with genes that protect them from disease could reduce waste and help limit the environmental impact of fish farming. A team of scientists have attempted to do just that—by inserting an alligator gene into the genomes of catfish.”
3D PRINTINGCan 3D Printing Help Solve the Housing Crisis?
Rachel Monroe | The New Yorker“Until last year, Icon, one of the biggest and best-funded companies in the field, had printed fewer than two dozen houses, most of them essentially test cases. But, when I met Ballard, the company had recently announced a partnership with Lennar, the second-largest home-builder in the United States, to print a hundred houses in a development outside Austin. A lot was riding on the project, which would be a test of whether the technology was ready for the mainstream.”
FUTURE1923 Cartoon Eerily Predicted 2023’s AI Art Generators
Benj Edwards | Ars Technica“[The vintage cartoon] depicts a cartoonist standing by his drawing table and making plans for social events while an ‘idea dynamo’ generates ideas and a ‘cartoon dynamo’ renders the artwork. Interestingly, this separation of labor feels similar to our neural networks of today. In the actual 2023, the ‘idea dynamo’ would likely be a large language model like GPT-3 (albeit imperfectly), and the ‘cartoon dynamo’ is most similar to an image-synthesis model like Stable Diffusion.”
TECHOpenAI CEO Sam Altman on GPT-4: ‘People Are Begging to Be Disappointed and They Will Be’
James Vincent | The Verge“GPT-3 came out in 2020, and an improved version, GPT 3.5, was used to create ChatGPT. The launch of GPT-4 is much anticipated, with more excitable members of the AI community and Silicon Valley world already declaring it to be a huge leap forward. …’The GPT-4 rumor mill is a ridiculous thing. I don’t know where it all comes from,’ said the OpenAI CEO. ‘People are begging to be disappointed and they will be. The hype is just like… We don’t have an actual AGI and that’s sort of what’s expected of us.’i”
COMPUTINGAre We Living in a Computer Simulation, and Can We Hack It?
Dennis Overbye | The New York Times“If you could change the laws of nature, what would you change? Maybe it’s that pesky speed-of-light limit on cosmic travel—not to mention war, pestilence and the eventual asteroid that has Earth’s name on it. Maybe you would like the ability to go back in time— to tell your teenage self how to deal with your parents, or to buy Google stock. Couldn’t the universe use a few improvements?”
Image Credit: Victor Crespo / Unsplash
In 2020, California-based Good Meat became the first company in the world to start selling lab-grown meat. Its cultured chicken has been on the market in Singapore since then, and though it’s still awaiting FDA approval to sell its products in the US, this week the company reached another milestone when it received approval to sell serum-free meat in Singapore.
The approval was granted by the Singapore Food Agency, and means Good Meat is allowed to use synthetic processes to create its products.
Cultured meat is grown from animal cells and is biologically the same as meat that comes from an animal. The process starts with harvesting muscle cells from an animal, then feeding those cells a mixture of nutrients and naturally-occurring growth factors (or, as Good Meat’s process specifies, amino acids, fats, and vitamins) so that they multiply, differentiate, then grow to form muscle tissue, in much the same way muscle grows inside animals’ bodies.
Usually, getting animal cells to duplicate requires serum. One of the more common is fetal bovine serum, which is made from the blood of fetuses extracted from cows during slaughter. It sounds a bit brutal even for the non-squeamish carnivore. Figuring out how to replicate the serum’s effects with synthetic ingredients has been one of the biggest hurdles to making cultured meat viable.
“Our research and development team worked diligently to replace serum with other nutrients that provide the same functionality, and their hard work over several years paid off,” said Andrew Noyes, head of communications at Good Meat’s parent company, Eat Just. The approval should allow for greater scalability, lower manufacturing costs, and a more sustainable product.
The company is in the process of building a demonstration plant in Singapore that will house a 6,000-liter bioreactor, which it says will be the largest in the industry to date and will have the capacity to make tens of thousands of pounds of meat per year.
The serum-free approval “complements the company’s work in Singapore to build and operate its bioreactor facility, where over 50 research scientists and engineers will develop innovative capabilities in the cultivated meat space such as media optimization, process development, and texturization of cultivated meat products,” said Damian Chan, executive vice president of the Singapore Economic Development Board.
It won’t be the only plant of its type. Israeli company Believer Meats opened a facility to produce lab-grown meat at scale in Israel in 2021, and last month started construction of a 200,000-square-foot factory in Wilson, North Carolina.
This past November a third player in the industry, Upside Foods, became the first company to receive a No Questions Letter from the FDA, essentially an approval saying its lab-grown chicken is safe for consumers to eat (though two additional approvals are still needed before the company can actually start selling the product).
The timing of the cultured meat industry’s advancement is convenient, though not coincidental; more consumers are becoming conscious of factory farming’s negative environmental impact, and they’re looking for eco-friendly alternatives. Cultured meat will allow them to eat real meat (as opposed to plant-based “meat”) with a far smaller environmental impact and no animals harmed to boot.
It remains to be seen whether scaling production will go as smoothly as Good Meat and its competitors are hoping, as well as how long it will take for the products to reach price parity with regular meat. But if the industry’s recent streak of clearing hurdles continues, lab-grown meat may soon be found in restaurants and on grocery shelves.
Image Credit: Good Meat
Two major astronomy research programs, called EMU and PEGASUS, have joined forces to resolve one of the mysteries of our Milky Way: where are all the supernova remnants?
A supernova remnant is an expanding cloud of gas and dust marking the last phase in the life of a star, after it has exploded as a supernova. But the number of supernova remnants we have detected so far with radio telescopes is too low. Models predict five times as many, so where are the missing ones?
We have combined observations from two of Australia’s world-leading radio telescopes, the ASKAP radio telescope and the Parkes radio telescope, Murriyang, to answer this question.
The Gas Between the StarsThe new image reveals thin tendrils and clumpy clouds associated with hydrogen gas filling the space between the stars. We can see sites where new stars are forming, as well as supernova remnants.
In just this small patch, only about 1 percent of the whole Milky Way, we have discovered more than 20 new possible supernova remnants where only 7 were previously known.
These discoveries were led by PhD student Brianna Ball from Canada’s University of Alberta, working with her supervisor, Roland Kothes of the National Research Council of Canada, who prepared the image. These new discoveries suggest we are close to accounting for the missing remnants.
So why can we see them now when we couldn’t before?
The Power of Joining ForcesI lead the Evolutionary Map of the Universe or EMU program, an ambitious project with ASKAP to make the best radio atlas of the southern hemisphere.
EMU will measure about 40 million new distant galaxies and supermassive black holes to help us understand how galaxies have changed over the history of the universe.
Early EMU data have already led to the discovery of odd radio circles (or “ORCs”), and revealed rare oddities like the “Dancing Ghosts.”
For any telescope, the resolution of its images depends on the size of its aperture. Interferometers like ASKAP simulate the aperture of a much larger telescope. With 36 relatively small dishes (each 12m in diameter) but a 6km distance connecting the farthest of these, ASKAP mimics a single telescope with a 6km wide dish.
That gives ASKAP a good resolution, but comes at the expense of missing radio emission on the largest scales. In the comparison above, the ASKAP image alone appears too skeletal.
To recover that missing information, we turned to a companion project called PEGASUS, led by Ettore Carretti of Italy’s National Institute of Astrophysics.
PEGASUS uses the 64m diameter Parkes/Murriyang telescope (one of the largest single-dish radio telescopes in the world) to map the sky.
Even with such a large dish, Parkes has rather limited resolution. By combining the information from both Parkes and ASKAP, each fills in the gaps of the other to give us the best fidelity image of this region of our Milky Way galaxy. This combination reveals the radio emission on all scales to help uncover the missing supernova remnants.
Linking the datasets from EMU and PEGASUS will allow us to reveal more hidden gems. In the next few years we will have an unprecedented view of almost the entire Milky Way, about a hundred times larger than this initial image, but with the same level of detail and sensitivity.
We estimate there may be up to 1,500 or more new supernova remnants yet to discover. Solving the puzzle of these missing remnants will open new windows into the history of our Milky Way.
ASKAP and Parkes are owned and operated by CSIRO, Australia’s national science agency, as part of the Australia Telescope National Facility. CSIRO acknowledge the Wajarri Yamaji people as the Traditional Owners and native title holders of Inyarrimanha Ilgari Bundara, the CSIRO Murchison Radio-astronomy Observatory, where ASKAP is located, and the Wiradjuri people as the traditional owners of the Parkes Observatory.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA
Proteins are often called the building blocks of life.
While true, the analogy evokes images of Lego-like pieces snapping together to form intricate but rigid blocks that combine into muscles and other tissues. In reality, proteins are more like flexible tumbleweeds—highly sophisticated structures with “spikes” and branches protruding from a central frame—that morph and change with their environment.
This shapeshifting controls the biological processes of living things—for example, opening the protein tunnels dotted along neurons or driving cancerous growth. But it also makes understanding protein behavior and developing drugs that interact with proteins a challenge.
While recent AI breakthroughs in the prediction (and even generation) of protein structures are a huge advance 50 years in the making, they still only offer snapshots of proteins. To capture whole biological processes—and identify which lead to diseases—we need predictions of protein structures in multiple “poses” and, more importantly, how each of these poses changes a cell’s inner functions. And if we’re to rely on AI to solve the challenge, we need more data.
Thanks to a new protein atlas published this month in Nature, we now have a great start.
A collaboration between MIT, Harvard Medical School, Yale School of Medicine, and Weill Cornell Medical College, the study focused on a specific chemical change in proteins—called phosphorylation—that’s known to act as a protein on-off switch, and in many cases, lead to or inhibit cancer.
The atlas will help scientists dig into how signaling goes awry in tumors. But to Sean Humphrey and Elise Needham, doctors at the Royal Children’s Hospital and the University of Cambridge, respectively, who were not involved in the work, the atlas may also begin to help turn static AI predictions of protein shapes into more fluid predictions of how proteins behave in the body.
Let’s Talk About PTMs (Huh?)After they’re manufactured, the surfaces of proteins are “dotted” with small chemical groups—like adding toppings to an ice cream cone. These toppings either enhance or turn off the protein’s activity. In other cases, parts of the protein get chopped off to activate it. Protein tags in neurons drive brain development; other tags plant red flags on proteins ready for disposal.
All these tweaks are called post-translational modifications (PTMs).
PTMs essentially transform proteins into biological microprocessors. They’re an efficient way for the cell to regulate its inner workings without needing to alter its DNA or epigenetic makeup. PTMs often dramatically change the structure and function of proteins, and in some cases, they could contribute to Alzheimer’s, cancer, stroke, and diabetes.
For Elisa Fadda at Maynooth University in Ireland and Jon Agirre at the University of York, it’s high time we incorporated PTMs into AI protein predictors like AlphaFold. While AlphaFold is changing the way we do structural biology, they said, “the algorithm does not account for essential modifications that affect protein structure and function, which gives us only part of the picture.”
The King PTMSo, what kinds of PTMs should we first incorporate into an AI?
Let me introduce you to phosphorylation. This PTM adds a chemical group, phosphate, to specific locations on proteins. It’s a “regulatory mechanism that is fundamental to life,” said Humphrey and Needham.
The protein hotspots for phosphorylation are well-known: two amino acids, serine and threonine. Roughly 99 percent of all phosphorylation sites are due to the duo, and previous studies have identified roughly 100,000 potential spots. The problem is identifying what proteins—dubbed kinases, of which there are hundreds—add the chemical groups to which hotspots.
In the new study, the team first screened over 300 kinases that specifically grab onto over 100 targets. Each target is a short string of amino acids containing serine and threonine, the “bulls-eye” for phosphorylation, and surrounded with different amino acids. The goal was to see how effective each kinase is at its job at every target—almost like a kinase matchmaking game.
This allowed the team to find the most preferred motif—sequence of amino acids—for each kinase. Surprisingly, “almost two-thirds of phosphorylation sites could be assigned to one of a small handful of kinases,” said Humphrey and Needham.
A Rosetta StoneBased on their findings, the team grouped the kinases into 38 different motif-based classes, each with an appetite for a particular protein target. In theory, the kinases can catalyze over 90,000 known phosphorylation sites in proteins.
“This atlas of kinase motifs now lets us decode signaling networks,” said Yaffe.
In a proof-of-concept test, the team used the atlas to hunt down cellular signals that differ between healthy cells and those exposed to radiation. The test found 37 potential phosphorylation targets of a single kinase, most of which were previously unknown.
Ok, so what?
The study’s method can be used to track down other PTMs to begin building a comprehensive atlas of the cellular signals and networks that drive our basic biological functions.
The dataset, when fed into AlphaFold, RoseTTAFold, their variants, or other emerging protein structure prediction algorithms, could help them better predict how proteins dynamically change shape and interact in cells. This would be far more useful for drug discovery than today’s static protein snapshots. Scientist may also be able to use such tools to tackle the kinase “dark universe.” This subset of kinases, more than 100, have no discernible protein targets. In other words—we have no idea how these powerful proteins work inside the body.
“This possibility should motivate researchers to venture ‘into the dark’, to better characterize these elusive proteins,” said Humphrey and Needham.
The team acknowledges there’s a long road ahead, but they hope their atlas and methodology can influence others to build new databases. In the end, we hope “our comprehensive motif-based approach will be uniquely equipped to unravel the complex signaling that underlies human disease progressions, mechanisms of cancer drug resistance, dietary interventions and other important physiological processes,” they said.
Image Credit: DeepMind
From painkillers to antihistamines to caffeine and beyond, we’ve found many ways to get our bodies to tolerate uncomfortable circumstances, for better and for worse. Now DARPA wants to add another to the list: getting the human body to better tolerate extreme cold.
The idea doesn’t sound like a great one at first glance; our bodies aren’t made to live in the cold, nor even withstand it for more than a little while. Our teeth start to chatter, we shiver, and eventually lose feeling in extremities, all signals that we need to get ourselves warm, stat—otherwise we can get hypothermia, frostbite, or worse.
The Defense Advanced Research Projects Agency (DARPA) has a few different motives for this research, but the primary one shouldn’t be surprising (though it’s still a bit creepy, IMO): enabling soldiers to be comfortable in cold places for long periods of time. The technology, if successful, could also be used to help explorers or adventurers (at high altitudes where it’s cold or in places like Alaska or the Arctic, for example) better tolerate cold, or to treat hypothermia patients.
Last week, Rice University in Houston announced that one of its assistant professors of bioengineering, Jerzy Szablowski, received a Young Faculty Award from DARPA to research nongenetic drugs that can “temporarily enhance the human body’s resilience to extreme cold exposure.”
Thermogenesis is the use of energy to create heat, and our bodies have two different ways of doing this. One is shivering, which we’re all familiar with. The other, which Szablowski simply calls nonshivering thermogenesis, involves burning off brown adipose tissue (BAT), or brown fat.
This type of fat exists specifically to warm us up when we get cold; it stores energy and only activates in cold temperatures. Most of our body fat is white fat. It builds up when we ingest more calories than we burn, and stores those calories for when we don’t get enough energy from food. An unfortunate majority of American adults have the opposite problem: too much white fat, which increases the risk of conditions like heart disease and type two diabetes.
While white fat is made of fatty acids called lipids, brown fat is dense in mitochondria (the component of cells where energy production occurs). When we get cold our bodies start pumping out the hormone norepinephrine, which attaches to receptors on brown fat cells, signaling the mitochondria to create energy—and warming us up in the process.
Szablowski will be trying to find ways to boost the BAT response. “If you have a drug that makes brown fat more active, then instead of having to spend weeks and weeks adapting to cold, you can perform better within hours,” he said. He added that his research will focus on finding a site to intervene in the BAT response, “like a protein or a process in the cell that you can target with a drug.”
Is it possible to change the body’s normal BAT response without needing to burn through more brown fat, which healthy adults don’t have a ton of to spare? We’ll see. Though white fat and brown fat have different compositions, it’s possible that Szablowski’s research could lead to new ways to eliminate white fat and treat obesity as well.
Image Credit: StockSnap from Pixabay
Billions of dollars are pouring into longevity startups as a growing body of research shows that aging might not be as inevitable as we assumed. Now, a startup claims to have reached a major milestone by extending the lifespans of healthy mice using a promising approach called cellular reprogramming.
In 2017, scientists at the Salk Institute for Biological Studies in San Diego first showed that it was possible to rejuvenate the cells of mice by resetting their epigenetic markers, chemical modifications to the DNA that don’t alter the underlying genetic code but can regulate the activity of certain genes. These changes have long been suspected of playing a crucial role in the aging process.
The researchers discovered that the approach could increase the lifespan of the mice by as much as 30 percent and significantly rejuvenate some of their tissues, but the experiments were done on animals with the mouse-equivalent of progeria, a disease that causes accelerated aging in humans.
It was unclear whether this kind of life extension would translate to normal healthy mice, but now preliminary results from a longevity startup called Rejuvenate Bio suggest that it does. A non-peer-reviewed paper published to the preprint server bioRxiv claims that the approach can double the remaining lifespan of elderly mice.
“While aging cannot currently be prevented, its impact on life and healthspan can potentially be minimized by interventions that aim to return gene expression networks to optimal function,” Noah Davidsohn, chief scientific officer and co-founder of Rejuvenate Bio, said in a press release. “The study results suggest that partial reprogramming could be a potential treatment in the elderly for reversing age-associated diseases and could extend human lifespan.”
Cellular reprogramming builds on the Nobel Prize-winning work of Shinya Yamanaka, who showed that adult cells could be transformed back into stem cells by exposing them to a specific set of genome-regulating proteins known as transcription factors. The Salk team’s innovation was to reduce the exposure times to the so-called Yamanaka factors, which they found could reverse epigenetic changes to the cells without reverting them to stem cells.
While the approach led to clear increases in lifespan in prematurely aging mice, the fact that no one had been able to replicate the result in healthy mice since then raised doubts about the approach. “Different groups have tried this experiment, and the data have not been positive so far,” Alejandro Ocampo, from the University of Lausanne in Switzerland, who carried out the original Salk experiments, told MIT Technology Review.
But now, Rejuvenate Bio claims that when they exposed healthy mice near the end of their lives to a subset of the Yamanaka factors, they lived for another 18 weeks on average, compared to just 9 weeks for those that didn’t undergo cellular reprogramming.
The mice were already 124 weeks old at the time, so this only represents a 7 percent increase in lifespan. But the company says it’s still a significant demonstration of the potential life-extending powers of cellular reprogramming, and the treated mice also showed improvements in a range of health metrics.
Another reason why the research is interesting, though, is the method by which the Yamanaka factors were administered. Previous studies have generally relied on genetically modifying mice to produce the factors themselves, but this study delivered them to the animals’ cells using repurposed viruses, which is the approach used in clinically approved gene therapies.
The results have yet to be peer reviewed, so should be taken with a pinch of salt until other groups are able to replicate them, but there is growing evidence of the potential therapeutic benefits of cellular reprogramming. Recent research on mice has shown that it can boost liver regeneration and help restore sight in animals with glaucoma.
It’s likely to be a long road to human trials, as there are significant question marks about potential side effects, including concerns that the approach could increase the risk of cancer. But promisingly, recent research on mice from the Salk team has shown that long-term treatment with Yamanaka factors led to significant rejuvenating effects on the animals’ tissues without causing any cancers. The research also found that the longer the treatment time, the better the results.
Further work will need to be done to validate the research from Rejuvenate Bio, but the results suggest that age-reversing treatments may soon be within reach.
Image Credit: Alexa / Pixabay
Every time scientists present a groundbreaking biological innovation, it seems as though there is a crescendo of noise—articles beckoning for public discussion, social media posts sharing the public’s opinions, scientists urging for more public input about bioethical decisions. The noise grows and grows and then—silence.
In August 2022, two research groups published papers in Nature and Cell that demonstrated scientists’ newfound ability to create synthetic mouse embryos in the laboratory until 8.5 days post-fertilization—no egg cells, sperm cells, or wombs needed. The outcry was immediate: If this can be done with mice, are humans next?
Scientists were quick to ease the public’s worries: It’s not yet possible to create synthetic human embryos. Yet their response was concerning. Why did we need to wait until such a scientific advance occurred before we could discuss its implications? How can we have important discussions about bioethical issues—issues at the intersection of ethics and biological research—that already impact society?
Typically, when such challenging bioethical dilemmas arise, scientists and ethicists will discuss the potential implications on committees and in forums, and will often provide policy recommendations. But unfortunately, public input is not always sought—or is sought in a limited capacity. And whether their opinions make any difference to policy is an open question.
We should all have the right to not only partake in bioethical discussions—but to partake in them in an effective and impactful manner. Otherwise, we’ll go to sleep one day, wake up the next morning, and realize we live in a world that we had no hand in creating.
When it came to the mouse embryos, some scientists discussed the need for public input when making complex and controversial bioethical decisions, echoing a longstanding refrain. But creating avenues for public discussion and deliberation about bioethical issues can be difficult.
Designing public discussion opportunities is time consuming and requires the expertise of a wide variety of professionals. Meanwhile, barriers exist in the form of scientists and policymakers who believe that the public can’t meaningfully contribute to scientific discourse due to a lack of understanding.
Even if that were the case, it’s not a reason to exclude people who would be affected by such decisions. Institutions must extend the effort to both inform the public and allow them to express their opinion.
There are some initiatives that promote public deliberation, such as Harvard Medical School’s public bioethics forums, which bring together stakeholders to discuss important bioethical topics. Providing such spaces is an important first step, as it effectively opens a seat at the table. Healthy deliberation—one which allows people to hold conflicting viewpoints and actively discuss their beliefs rather than simply consume information—is critical for making bioethics a more inclusive and democratic space.
“We should all have the right to not only partake in bioethical discussions—but to partake in them in an effective and impactful manner.”
But public input doesn’t ultimately count for much if such discussions don’t exert any actual influence on policymaking. Despite their role in fostering educated discussions, initiatives such as Harvard’s do not allow citizens to contribute to new policy decisions.
Historically, there have been some attempts to do so. Since the 1970s, many countries, including the US, have implemented public deliberation as a part of bioethical decision-making, to varying degrees of success. In some instances, such as with the 1974 National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, public opinion was considered and some of the commission’s final reports were heavily influential in policy. But again, it’s questionable how much input the public truly had. Their input was sought solely through public hearings. Bioethicists and policymakers comprised the commission and created the final reports.
Fortunately, more recently, there have been public deliberation efforts that provide citizens with an opportunity to influence policymaking decisions. For instance, the Citizens’ Reference Panel on Health Technologies in Ontario, Canada made a small yet critical impact on governmental decision-making. This panel was created to allow Ontarians to inform how regulatory bodies assess five health technologies. The one technology the panel had the most profound effect on was screening methods for colorectal cancers and polyps. While widespread screening has many benefits, citizens expressed some concerns about the loss of patient autonomy when screening was performed automatically without patient input. This point was added to a final recommendation document created by the Ontario Health Technology Advisory Committee, and committee members have since said that the point would have gone unnoticed had it not been for the panel.
Another example comes from Buckinghamshire in England, where a citizens’ jury expressed their opinions about how to tackle back pain, a major health problem for the county’s citizens. In this context, a citizens’ jury is a two- to five-day event where a few dozen members of the general public come together to discuss an issue and ultimately produce a recommendation document. The Buckinghamshire Health Authority, or BHA, promised that they would take the jury’s recommendations into account, and they did. The BHA then formed a project team to implement these recommendations.
This begs the question: What makes certain public deliberation efforts successful and others not?
If success is defined as a near-direct impact on policy decisions, a common theme emerges: Citizens’ panels and juries that are connected to a governmental organization tend to be more impactful policy-wise, particularly in the short term.
In both previous examples, the government was involved to varying degrees, and—perhaps more importantly—the public’s recommendations were actually prioritized. As Susan Goold, an ethicist and professor at the University of Michigan, put it in an interview with Undark, policymakers should never say “see you later” after a deliberative session.
In Buckinghamshire, as part of an agreement with the King’s Fund—a health improvements charity that was supporting this public deliberation effort—the BHA was required to follow the panel’s recommendations. If they chose not to, they had to state specific reasons. This ensured accountability and the implementation of the recommendations.
Another critical aspect of successful public deliberation efforts is appropriate organization. Julia Abelson, lead of the Public Engagement in Health Policy Project and a professor at McMaster University, explained that there are examples of government-initiated public deliberation that have had little impact as well as efforts not directly linked to the government that were very impactful.
The differentiating factor is thoughtful planning and organization. For instance, it’s critical that, during the design phase of the process, organizers set clear goals and objectives they’d like to meet by the end of deliberation.
Additionally, organizers should carefully consider how information is presented to participants. How questions are framed, for example, can affect whether new ideas emerge from participants. Another important component organizers need to consider is how discussions are moderated. For instance, are the facilitators actively shaping the discussion or solely preventing one participant from dominating the conversation?
Though some research has been done on this topic, many questions remain. What researchers know is that all of the elements above must come together to create a successful citizens’ panel that can impact policy down the line.
There is no question that public input is immensely valuable whether we’re discussing gene editing or the creation of synthetic embryos. Thankfully, the increase in the number of deliberation efforts reflects that. However, public deliberation is a tool, and like all tools, it requires a guiding hand.
We must ensure that governments are involved in deliberation efforts when necessary and that citizens’ panels are designed thoughtfully. We must do this so one day, when we go to sleep and wake up the next morning, we’ll see the sun rising on a world we’ve built together.
This article was originally published on Undark. Read the original article.
Image Credit: Furiosa-L / Pixabay
ARTIFICIAL INTELLIGENCEMicrosoft Bets Big on the Creator of ChatGPT in Race to Dominate AI
Cade Metz and Karen Weise | The New York Times“Microsoft is in talks to invest another $10 billion in OpenAI as it seeks to push its technology even further, according to a person familiar with the matter. The potential $10 billion deal—which would mainly provide OpenAI with even larger amounts of computing power—has not been finalized and the funding amount could change. But the talks are indicative of the tech giant’s determination to be on the leading edge of what has become the hottest technology in the tech industry.”
BIOTECHThe Entrepreneur Dreaming of a Factory of Unlimited Organs
Antonio Regalado | MIT Technology Review“…if ‘unlimited organs’ really become available, it’s going to vastly increase the number of people who might be eligible, uncorking needs currently masked by strict transplant rules and procedures. …’We don’t really talk about it, but if there were unlimited organs, you could replace dialysis, replace heart assist devices, even replace medicines that don’t work that well,’ says [Robert Montgomery, the New York University surgeon who carried out the first transplant of a pig kidney]. ‘I think there are a million people with heart failure, and how many get a transplant? Only 3,500.’i”
SPACE Last Year Marked the End of an Era in Spaceflight—Here’s What We’re Watching Next
Eric Berger | Ars Technica“Consider the state of play in 2010: A handful of large government space agencies controlled spaceflight activities. NASA was still flying the venerable space shuttle with no clear plan for deep space exploration. The James Webb Space Telescope remained in development hell. Russia was the world’s dominant launch provider, putting as many rockets into space that year as the United States and China combined. At the time, China’s longest human spaceflight was four days. Much has changed in the last decade or so.”
BIOTECHFDA Will No Longer Require Animal Tests Before Human Trials for All Drugs
Lauren Leffer | Gizmodo“Instead of animal testing, new drugs can now move onto human trials following successful rounds of ‘non clinical tests,’ an umbrella term that includes animal tests but also allows for technological advances like computer simulations, organ chips, and 3D printed body parts to replace animals.”
FUTUREDARPA Wants to Find a Drug That Makes You Impervious to Cold
Ed Cara | Gizmodo“The Defense Advanced Research Projects Agency (DARPA) is looking for a new way to get nice and cozy: The agency is funding research into drugs that could protect people from extreme cold. Should these efforts bear fruit, the drugs could have a variety of uses, from treating hypothermia patients to helping people better explore the Arctic—and, what is surely DARPA’s main interest, creating soldiers who aren’t fazed by freezing conditions.”
ARTIFICIAL INTELLIGENCEIf ChatGPT Doesn’t Get a Better Grasp of Facts, Nothing Else Matters
Harry McCracken | Fast Company“…whenever I chat with ChatGPT about any subject I know much about, such as the history of animation, I’m most struck by how deeply untrustworthy it is. If a rogue software engineer set out to poison our shared corpus of knowledge by generating convincing-sounding misinformation in bulk, the end result might look something like this.”
TECHThe Slow Death of Surveillance Capitalism Has Begun
Morgan Meaker | Wired“Surveillance capitalism just got a kicking. In an ultimatum, the European Union has demanded that Meta reform its approach to personalized advertising—a seemingly unremarkable regulatory ruling that could have profound consequences for a company that has grown impressively rich by, as Mark Zuckerberg once put it, running ads.“
TRANSPORTATIONAirbus Is Testing Out Autonomous Flying Tech in Some of Its Planes
Andrew J. Hawkins | The Verge“Airbus is testing out a suite of new automated technology that it says has the potential to improve the safety and efficiency of flying. The automated technology, which has been branded as the company’s DragonFly project, includes ‘automated emergency diversion in cruise, automatic landing, and taxi assistance,’ Airbus says. The company is testing out the new features using an A350-1000 aircraft at the Toulouse-Blagnac Airport, which is a test site for Airbus.”
3D PRINTING3D-Printed Houses Are the Suburbs of the Future
Sam Lubell | Fast Company“[The 3D printing] evolution could finally give award-winning architects like EYRC and BIG—long frozen out of the formula-driven multi-billion-dollar mass homebuilding industry—a feasible way in; particularly if builders are looking to differentiate themselves through innovative tract layouts and home compositions. But while the few 3D printed tracts now going up show some promise, ratcheting up the greenery, sustainability, and design quality, their repetitive planning and architecture don’t stray far from the norm.”
FUTUREDon’t Ban ChatGPT in Schools. Teach With It.
Kevin Roose | The New York Times“There are legitimate questions about the ethics of AI-generated writing, and concerns about whether the answers ChatGPT gives are accurate. (Often, they’re not.) And I’m sympathetic to teachers who feel that they have enough to worry about, without adding AI-generated homework to the mix. But after talking with dozens of educators over the past few weeks, I’ve come around to the view that banning ChatGPT from the classroom is the wrong move.”
Image Credit: Planet Volumes / Unsplash
These days, we don’t have to wait long until the next breakthrough in artificial intelligence (AI) impresses everyone with capabilities that previously belonged only in science fiction.
In 2022, AI art generation tools such as Open AI’s DALL-E 2, Google’s Imagen, and Stable Diffusion took the internet by storm, with users generating high-quality images from text descriptions.
Unlike previous developments, these text-to-image tools quickly found their way from research labs to mainstream culture, leading to viral phenomena such as the “Magic Avatar” feature in the Lensa AI app, which creates stylized images of its users.
In December, a chatbot called ChatGPT stunned users with its writing skills, leading to predictions the technology will soon be able to pass professional exams. ChatGPT reportedly gained one million users in less than a week. Some school officials have already banned it for fear students would use it to write essays. Microsoft is reportedly planning to incorporate ChatGPT into its Bing web search and Office products later this year.
What does the unrelenting progress in AI mean for the near future? And is AI likely to threaten certain jobs in the following years?
Despite these impressive recent AI achievements, we need to recognize there are still significant limitations to what AI systems can do.
AI Excels at Pattern RecognitionRecent advances in AI rely predominantly on machine learning algorithms that discern complex patterns and relationships from vast amounts of data. This training is then used for tasks like prediction and data generation.
The development of current AI technology relies on optimizing predictive power, even if the goal is to generate new output.
For example, GPT-3, the language model behind ChatGPT, was trained to predict what follows a piece of text. GPT-3 then leverages this predictive ability to continue an input text given by the user.
“Generative AIs” such as ChatGPT and DALL-E 2 have sparked much debate about whether AI can be genuinely creative and even rival humans in this regard. However, human creativity draws not only on past data but also on experimentation and the full range of human experience.
Cause and EffectMany important problems require predicting the effects of our actions in complex, uncertain, and constantly changing environments. By doing this, we can choose the sequence of actions most likely to achieve our goals. But algorithms cannot learn about causes and effects from data alone. Purely data-driven machine learning can only find correlations.
To understand why this is a problem for AI, we can contrast the problems of diagnosing a medical condition versus choosing a treatment. Machine learning models are often helpful for finding abnormalities in medical images; this is a pattern recognition problem. We don’t need to worry about causality because abnormalities are already either present or not.
But choosing the best treatment for a diagnosis is a fundamentally different problem. Here, the goal is to influence the outcome, not just recognize a pattern. To determine the effectiveness of a treatment, medical researchers run randomized controlled trials. This way, they can try to control any factors that might affect the treatment.
The confusion between these two types of problems sometimes leads to suboptimal applications of machine learning in organizations.
While the success of recent work in AI demonstrates the value of data-driven models, many of the problems we would want computers to solve require an understanding of causation.
Current AI systems lack this ability, apart from specialized applications such as board games.
A Stable Diffusion artwork generated by the prompt ‘The limits of artificial intelligence’. Current AI technology tends to perform poorly in unexpected situations. In this case, the prompt is not well represented in Stable Diffusion’s training data. Image Credit: Stable DiffusionCommon Sense ReasoningLanguage models such GPT-3 and ChatGPT can successfully solve some tasks requiring common-sense reasoning.
However, the following interaction with ChatGPT, adapted from an experiment by Gary Marcus, suggests it is not entirely reliable in this respect.
I’m eight years old. When I was born, my biological mother was in Barcelona and my father was in Tokyo. Where was I born? Think through this step by step.
It is not mentioned where you were born. All we know is that your biological mother was in Barcelona and your father was in Tokyo at the time of your birth.
Whether AI systems such as ChatGPT can achieve common sense is a subject of lively debate among experts.
Sceptics such as Marcus point out we cannot trust language models to robustly display common sense since they neither have it built into them nor are directly optimized for it. Optimists argue that while current systems are imperfect, common sense may spontaneously emerge in sufficiently advanced language models.
Human ValuesWhenever groundbreaking AI systems are released, news articles and social media posts documenting racist, sexist, and other types of biased and harmful behaviors inevitably follow.
This flaw is inherent to current AI systems, which are bound to be a reflection of their data. Human values such as truth and fairness are not fundamentally built into the algorithms; that’s something researchers don’t yet know how to do.
While researchers are learning the lessons from past episodes and making progress in addressing bias, the field of AI still has a long way to go to robustly align AI systems with human values and preferences.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Mahdis Mousavi/Unsplash
AI is being used to generate everything from images to text to artificial proteins, and now another thing has been added to the list: speech. Last week researchers from Microsoft released a paper on a new AI called VALL-E that can accurately simulate anyone’s voice based on a sample just three seconds long. VALL-E isn’t the first speech simulator to be created, but it’s built in a different way than its predecessors—and could carry a greater risk for potential misuse.
Most existing text-to-speech models use waveforms (graphical representations of sound waves as they move through a medium over time) to create fake voices, tweaking characteristics like tone or pitch to approximate a given voice. VALL-E, though, takes a sample of someone’s voice and breaks it down into components called tokens, then uses those tokens to create new sounds based on the “rules” it already learned about this voice. If a voice is particularly deep, or a speaker pronounces their A’s in a nasal-y way, or they’re more monotone than average, these are all traits the AI would pick up on and be able to replicate.
The model is based on a technology called EnCodec by Meta, which was just released this part October. The tool uses a three-part system to compress audio to 10 times smaller than MP3s with no loss in quality; its creators meant for one of its uses to be improving the quality of voice and music on calls made over low-bandwidth connections.
To train VALL-E, its creators used an audio library called LibriLight, whose 60,000 hours of English speech is primarily made up of audiobook narration. The model yields its best results when the voice being synthesized is similar to one of the voices from the training library (of which there are over 7,000, so that shouldn’t be too tall of an order).
Besides recreating someone’s voice, VALL-E also simulates the audio environment from the three-second sample. A clip recorded over the phone would sound different than one made in person, and if you’re walking or driving while talking, the unique acoustics of those scenarios are taken into account.
Some of the samples sound fairly realistic, while others are still very obviously computer-generated. But there are noticeable differences between the voices; you can tell they’re based on people who have different speaking styles, pitches, and intonation patterns.
The team that created VALL-E knows it could very easily be used by bad actors; from faking sound bites of politicians or celebrities to using familiar voices to request money or information over the phone, there are countless ways to take advantage of the technology. They’ve wisely refrained from making VALL-E’s code publicly available, and included an ethics statement at the end of their paper (which won’t do much to deter anyone who wants to use the AI for nefarious purposes).
It’s likely just a matter of time before similar tools spring up and fall into the wrong hands. The researchers suggest the risks that models like VALL-E will present could be mitigated by building detection models to gauge whether audio clips are real or synthesized. If we need AI to protect us from AI, how do know if these technologies are having a net positive impact? Time will tell.
Image Credit: Shutterstock.com/Tancha
This past May the Biden administration allocated $3.16 billion in funding to start securing a battery supply chain for the US. Much of the focus has been on lithium ion batteries for electric vehicles, but soon a different kind of battery will join the fray. The Gates-backed battery startup Form Energy recently announced plans to build a $760 million factory for its iron air batteries.
These batteries are made for storing electricity produced by renewables like solar and wind, and a few key features differentiate them from lithium ion.
Each battery is as big as a washing machine and holds 50 iron-air cells, which are flat and about a square meter in size. The cells are surrounded with a water-based electrolyte similar to what’s used in AA batteries. To discharge, the battery breathes in oxygen from the air, converting the iron to iron oxide—commonly known as rust—and producing electricity in the process. Applying a current converts the rust back into iron, expels the oxygen, and charges the battery.
Iron air batteries have two big advantages over lithium ion. The first is cost. Iron is cheaper and far easier to procure than minerals like nickel, cobalt, or lithium. Form Energy says they can make their batteries at a cost of $20 per kilowatt hour, less than half the cost of lithium-ion batteries.
The second advantage is duration: iron air batteries can store power for 100 to 150 hours. This would come in particularly handy for storing energy generated by renewables like wind or solar, whose intermittency prevents them from being reliable baseload power sources.
The flip side of slow discharging, though, is slow charging; it takes iron air batteries longer to charge than their lithium ion counterparts. Form plans to create “powerblocks” made up of thousands of batteries, which will likely be used alongside conventional batteries; iron-air would power longer load demands and lithium would be there for demand spikes.
A final advantage of Form’s batteries is that they’re recyclable. The metals used to make them can be extracted at the end of a battery’s life and used elsewhere (this is precisely what a battery recycling plant under construction in South Carolina will do, except with conventional batteries).
Late last month, the company announced plans to partner with the state of West Virginia to build its first manufacturing plant in a former steel-producing town called Weirton. The facility will sit on 55 acres and employ around 750 people. Construction is expected to start this year, and the first batteries will come off the production line in 2024.
Form received $450 million in Series E funding last October, bringing its total funding to over $800 million. The company has a contract in place to install a 150-megawatt-hour battery in Minnesota and is looking to do a similar pilot in Georgia.
Whether iron air batteries can truly make a dent in the renewables storage dilemma remains to be seen, but signs point to Form Energy being set up for success.
Image Credit: Form Energy
There’s one deceptively simple early sign of Alzheimer’s not often talked about: a subtle change in speech patterns.
Increased hesitation. Grammatical mistakes. Forgetting the meaning of a word, or mispronouncing common words—or favorite phrases and idioms—that used to flow naturally.
Scientists have long thought to decode this linguistic degeneration as an early indicator of Alzheimer’s. One idea is to use natural language software as a “guide” of sorts that hunts down unusual use of language.
Sounds simple, right? Here’s the problem: everyone talks differently. It seems obvious, but it’s a giant headache for AI. Our speech patterns, cadence, tone, and word choice are all colored with shades of personal history and nuances that the average language AI struggles to decipher. A sentence that’s sarcastic for one person may be completely sincere for another. A recurrent grammatical error could be a personal habit from decades of misuse now hard to change—or a reflection of dementia.
So why not tap into the most creative AI language tools today?
In a study published in PLOS Digital Health, a team from Drexel University took a major step in bridging GPT-3’s creative force with neurological diagnosis. Using a publicly available dataset of speech transcripts from people with and without Alzheimer’s, the team retrained GPT-3 to pick out linguistic nuances that suggest dementia.
When fed with new data, the algorithm reliably detected Alzheimer’s patients from healthy ones and could predict the person’s cognitive testing score—all without any additional knowledge of the patients or their history.
“To our knowledge, this is the first application of GPT-3 to predicting dementia from speech,” the authors said. “The use of speech as a biomarker provides quick, cheap, accurate, and non-invasive diagnosis of AD and clinical screening.”
Early BirdDespite science’s best efforts, Alzheimer’s is incredibly hard to diagnose. The disorder, often with a genetic disposition, doesn’t have a unified theory or treatment. But what we know is that inside the brain, regions associated with memory start accumulating protein clumps that are toxic to neurons. This causes inflammation in the brain, which accelerates decline in memory, cognition, and mood, eventually eroding everything that makes you you.
The most insidious part of Alzheimer’s is that it’s hard to diagnose. For years, the only way to confirm the disorder was through an autopsy, looking for the telltale signs of protein clumps—beta-amyloid balls outside cells and strings of tau proteins inside. These days, brain scans can capture these proteins earlier. Yet scientists have long known that cognitive symptoms may creep up long before the protein clumps manifest.
Here’s the silver lining: even without a cure, diagnosing Alzheimer’s early can help patients and their loved ones make plans around support, mental health, and finding treatments to manage symptoms. With the FDA’s recent approval of Leqembi, a drug that moderately helps protect cognitive decline in people with early-stage Alzheimer’s, the race to catch the disease early is heating up.
Speak Your MindRather than focusing on brain scans or blood biomarkers, the Drexel team turned to something remarkably effortless: speech.
“We know from ongoing research that the cognitive effects of Alzheimer’s disease can manifest themselves in language production,” said study author Dr. Hualou Liang. “The most commonly used tests for early detection of Alzheimer’s look at acoustic features, such as pausing, articulation, and vocal quality, in addition to tests of cognition.”
The idea has long been pursued by cognitive neuroscientists and AI scientists. Natural Language Processing (NLP) has dominated the AI sphere in its ability to recognize everyday language. By feeding it recordings of a patient’s voice or their writings, neuroscientists could highlight particular vocal “tics” that a certain group of people may have—for example, those with Alzheimer’s.
It sounds great, but these are heavily-tailored studies. They rely on knowledge of specific problems rather than more universal Q-and-As. The resulting algorithms are hand-crafted, making them hard to scale to a broader population. It’s like going to a tailor for a perfectly fitted suit or dress, only to realize it doesn’t fit anyone else or even yourself after a few months.
That’s a problem for diagnoses. Alzheimer’s—or heck, any other neurological disorder—tends to progress. An algorithm trained in this way makes it “hard to generalize to other progression stages and disease types, which may correspond to different linguistic features,” the authors said.
In contrast, large language models (LLMs), which underlie GPT-3, are far more flexible to provide a “powerful and universal language understanding and generation,” the authors said.
One particular aspect caught their eye: embedding. Put simply, it means that the algorithm can learn from a hefty well of information and generate an “idea” of sorts for each “memory.” When used for text, the trick can uncover additional patterns and characteristics even beyond what most trained experts could detect, the authors said. In other words, a GPT-3-fueled program, based on text embedding, could potentially detect speech pattern differences that escape neurologists.
“GPT-3’s systemic approach to language analysis and production makes it a promising candidate for identifying the subtle speech characteristics that may predict the onset of dementia,” said study author Felix Agbavor. “Training GPT-3 with a massive dataset of interviews—some of which are with Alzheimer’s patients—would provide it with the information it needs to extract speech patterns that could then be applied to identify markers in future patients.”
A Creative SolutionThe team readily used GPT-3 for two critical measures of Alzheimer’s: discerning an Alzheimer’s patient from a healthy one and predicting a patient’s severity of dementia based on a benchmark for cognition dubbed the Mini-Mental State Exam (MMSE).
Similar to most deep learning models, GPT-3 is incredibly hungry for data. Here, the team fed it the ADReSSo Challenge (Alzheimer’s Dementia Recognition through Spontaneous Speech), which contains everyday speech from people with and without Alzheimer’s.
For the first challenge, the team pitted their GPT-3 programs against two that hunt down specific “tics” in language. Both models, Ada and Babbage (a nod to computing pioneers) far outperformed the conventional model based on acoustic features alone. The algorithms fared even better when predicting the accuracy of the dementia MMSE by speech features alone.
When pitted against other state-of-the-art Alzheimer’s detection models, the Babbage edition crushed the opponents for accuracy and level of recall.
“These results, all together, suggest that GPT-3-based text embedding is a promising approach for AD assessment and has the potential to improve early diagnosis of dementia,” the authors said.
With the hype of GPT-3 and AI in healthcare in general, it’s easy to lose sight of what really matters: the health and well-being of the patient. Alzheimer’s is a terrible disease, one that literally erodes the mind. An earlier diagnosis is information, and information is power—which can help inform life choices and assess treatment options.
“Our proof-of-concept shows that this could be a simple, accessible, and adequately sensitive tool for community-based testing,” said Liang. “This could be very useful for early screening and risk assessment before a clinical diagnosis.”
Image Credit: NIH
All life is made up of cells several magnitudes smaller than a grain of salt. Their seemingly simple-looking structures mask the intricate and complex molecular activity that enables them to carry out the functions that sustain life. Researchers are beginning to be able to visualize this activity to a level of detail they haven’t been able to before.
Biological structures can be visualized by either starting at the level of the whole organism and working down, or starting at the level of single atoms and working up. However, there has been a resolution gap between a cell’s smallest structures, such as the cytoskeleton that supports the cell’s shape, and its largest structures, such as the ribosomes that make proteins in cells.
By analogy of Google Maps, while scientists have been able to see entire cities and individual houses, they did not have the tools to see how the houses came together to make up neighborhoods. Seeing these neighborhood-level details is essential to being able to understand how individual components work together in the environment of a cell.
New tools are steadily bridging this gap. And ongoing development of one particular technique, cryo-electron tomography, or cryo-ET, has the potential to deepen how researchers study and understand how cells function in health and disease.
As the former editor-in-chief of Science magazine and as a researcher who has studied hard-to-visualize large protein structures for decades, I have witnessed astounding progress in the development of tools that can determine biological structures in detail. Just as it becomes easier to understand how complicated systems work when you know what they look like, understanding how biological structures fit together in a cell is key to understanding how organisms function.
A Brief History of MicroscopyIn the 17th century, light microscopy first revealed the existence of cells. In the 20th century, electron microscopy offered even greater detail, revealing the elaborate structures within cells, including organelles like the endoplasmic reticulum, a complex network of membranes that play key roles in protein synthesis and transport.
From the 1940s to 1960s, biochemists worked to separate cells into their molecular components and learn how to determine the 3D structures of proteins and other macromolecules at or near atomic resolution. This was first done using X-ray crystallography to visualize the structure of myoglobin, a protein that supplies oxygen to muscles.
Over the past decade, techniques based on nuclear magnetic resonance, which produces images based on how atoms interact in a magnetic field, and cryo-electron microscopy have rapidly increased the number and complexity of the structures scientists can visualize.
What Are Cryo-EM and Cryo-ET?Cryo-electron microscopy, or cryo-EM, uses a camera to detect how a beam of electrons is deflected as the electrons pass through a sample to visualize structures at the molecular level. Samples are rapidly frozen to protect them from radiation damage. Detailed models of the structure of interest are made by taking multiple images of individual molecules and averaging them into a 3D structure.
Cryo-ET shares similar components with cryo-EM but uses different methods. Because most cells are too thick to be imaged clearly, a region of interest in a cell is first thinned by using an ion beam. The sample is then tilted to take multiple pictures of it at different angles, analogous to a CT scan of a body part (although in this case the imaging system itself is tilted, rather than the patient). These images are then combined by a computer to produce a 3D image of a portion of the cell.
The resolution of this image is high enough that researchers (or computer programs) can identify the individual components of different structures in a cell. Researchers have used this approach, for example, to show how proteins move and are degraded inside an algal cell.
Many of the steps researchers once had to do manually to determine the structures of cells are becoming automated, allowing scientists to identify new structures at vastly higher speeds. For example, combining cryo-EM with artificial intelligence programs like AlphaFold can facilitate image interpretation by predicting protein structures that have not yet been characterized.
Understanding Cell Structure and FunctionAs imaging methods and workflows improve, researchers will be able to tackle some key questions in cell biology with different strategies.
The first step is to decide what cells and which regions within those cells to study. Another visualization technique called correlated light and electron microscopy, or CLEM, uses fluorescent tags to help locate regions where interesting processes are taking place in living cells.
Comparing the genetic difference between cells can provide additional insight. Scientists can look at cells that are unable to carry out particular functions and see how this is reflected in their structure. This approach can also help researchers study how cells interact with each other.
Cryo-ET is likely to remain a specialized tool for some time. But further technological developments and increasing accessibility will allow the scientific community to examine the link between cellular structure and function at previously inaccessible levels of detail. I anticipate seeing new theories on how we understand cells, moving from disorganized bags of molecules to intricately organized and dynamic systems.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Nanographics, CC BY-SA
Robots are already helping cook food, do construction work, clean homes, and more. In the future they’ll take over additional tasks—but which ones? At this year’s Consumer Electronics Show (CES) in Las Vegas, a plethora of robots with various purposes was on display. Some of them were silly, some ingenious, some a bit creepy. Not all of these will end up being widely used, but there’s certainly a variety of jobs robots could do for us in the not-too-distant future. Here are a few of them, more or less in descending order from “most likely to be of use or have a positive impact” to “least necessary/just for kicks.”
Harvest Our FoodImage Credit: AgristJapanese agritech startup Agrist’s simply-named “L” robot can identify and pick harvest-ready bell peppers with millimeter precision and through thickly-layered leaves. A robot like this could be not just handy, but necessary if the current agricultural worker shortage continues. L uses cameras and an AI algorithm to identify a pepper’s position, size, maturity, and clipping point. It moves along suspension wires that need to be pre-installed, and can then approach a plant, find a target pepper, clip it off, then fold to drop it into a collection box. L can also forecast harvest volume and collect data about crops, such as the number of days left to reach maturity. Agrist says L costs less than $10,000, as compared to an average $73,000 for conventional automatic harvesting robots. The robot could likely be trained to harvest a variety of fruits and vegetables.
Plant Our FoodImage Credit: John DeereIn keeping with the agricultural theme, John Deere brought a highly functional farming tool to the show. The company says its ExactShot robotic planter can reduce the amount of starter fertilizer farmers use by more than 60 percent. It uses sensors to place starter fertilizer straight onto individual seeds as they’re planted rather than blindly spraying fertilizer over the entire row of seeds. Across the US corn crop alone, the company says, ExactShot could save over 93 million gallons of starter fertilizer annually—which would also prevent excess fertilizer from causing weeds to grow or leeching into local waterways.
Take Care of UsImage Credit: Vanessa Bates RamirezAeo is a service robot made by Japanese company Aoelus Robotics. The company says its bot can be used for security, delivery, healthcare, and hospitality purposes. Aeo has two arms, one outfitted with grippers to pick up objects, open doors, or press buttons, and the other fitted with an L-shaped UV attachment to disinfect surfaces. Its 360° night-vision camera can monitor a home, office, or other space, and stream live video to your phone or laptop. Its Care function can detect when patients are in distress or at risk (details of how the robot does this are light). It’s relatively compact at 3.8 feet tall by 1.8 feet wide, and its arm can lift up to 8.8 pounds; so it won’t be helping any patients up if they fall, but it can bring them food, drinks, or other supplies. Aeo is already in use in airports, hotels, and hospitals in Taiwan, Hong Kong, and Japan.
Deliver Things to UsOttonomy wants to reduce the cost of deliveries by 50 percent with it Ottobot delivery robot. The tricked-out-box-on-wheels is about 4.5 feet tall, 4 feet long, and 2.5 feet wide, and weighs 200 pounds. It won’t win any races with a max speed of four miles per hour—that’s equivalent to the pace an average adult walks at—but depending where it’s coming from, its speed may not be all that important. The bot can do autonomous deliveries—where a door opens and a box is deposited on the ground—or attended deliveries, customer gets a text telling them the robot is there and a QR code to open the compartment. There’s a smaller compartment for things like wine bottles or other liquids, and a larger one that could hold groceries or food deliveries; the robot is customizable and modular, so customers can tailor its structure and compartments to their specific needs. It runs on a swappable battery and navigates autonomously through its environment.
Charge Our Electric CarsImage Credit: EvarEvar’s Parky robot was made to help electric vehicle owners get their cars’ batteries recharged faster and with less hassle. As EV adoption grows, tools like Parky could become helpful as drivers try to navigate an as-yet-slim charging infrastructure. Rather than having to park at a charging bay, drivers can park anywhere in a lot and have Parky come to them. The bot provides 15kW DC charging per hour, juicing vehicles up with about 50 miles of range. The catch is that drivers still have to find a spot next to an “EV robot connector” and plug in, so depending on supply-demand ratios, Parky may not make much of a difference in terms of convenience and speed; the robot makes the most sense for buildings that want to make their parking lots more EV-friendly without undertaking construction or redesign work or having to add electric capacity.
Make Us Bubble TeaImage Credit: Vanessa Bates RamirezRichtech Robotics’ Adam robot has two arms with grip handles that can be customized to make various drinks. During CES the bot was churning out bubble tea; customers could choose a flavor on a touch screen and the robot would mix the necessary ingredients, add ice and boba, seal the cup, then deposit it on the counter for the waiting customer. Adam can perform bartending or barista duties as well. One thing Adam may want to work on, though, is speed; I stood in line to get a robotic bubble tea for about five minutes, and after not moving an inch, I left the line before even selecting a flavor. I’m all for automating bubble tea and other drinks if it’ll make them better, faster, or cheaper, but between the long wait and not having gotten to try the final product, I can’t quite vouch for Adam yet.
Banner Image Credit: Vanessa Bates Ramirez
Space-based solar power could provide round-the-clock access to renewable energy, sidestepping one of the technology’s biggest limitations. Now the idea is going to get its first true test after a Falcon 9 rocket successfully launched experimental hardware designed to assess its feasibility.
The idea of stationing gigantic solar panels in orbit around Earth and beaming the power back has been around for decades. The possibility is attractive, because in space you’re no longer at the mercy of the weather or the planet’s cycles of day and night, and solar radiation levels are higher as sunlight has not had to pass through the atmosphere.
So far though, space-based solar power has remained in the realm of sci-fi due to the technical complexity and unforgiving economics of space technologies. But thanks to a $100 million donation in 2013, a multidisciplinary team from Caltech has been quietly working on it over the past decade, developing the various technologies required to make it a reality. And this past Tuesday, prototypes of some of the key subsystems required for a full-scale space-based solar power plant were delivered into orbit by SpaceX for testing.
Over the next few months, the team behind the Caltech Space Solar Power Project will test out the systems that will allow their flexible solar panels to unfurl in space and the technology designed to transmit power back to Earth. They will also assess how well different kinds of solar panel technologies hold up in the harsh environment of space.
“No matter what happens, this prototype is a major step forward,” Ali Hajimiri, one of the three Caltech professors leading the project, said in a statement. “It works here on Earth, and has passed the rigorous steps required of anything launched into space. There are still many risks, but having gone through the whole process has taught us valuable lessons.”
Building solar panels in space is a much more complicated business than doing so on Earth. The biggest challenge is getting them there in the first place, which is limited by the incredible cost of launching material into orbit. As a result, the team has had to focus on reducing the weight of their solar panels as much as possible without sacrificing their generating capacity.
Their solution combines ultra-thin flexible solar panels, an ingenious design that integrates power generation and transmission, and a novel modular architecture that makes it possible to combine many smaller, self-contained panels to create large arrays.
The basic unit of their design is a rectangular tile a few inches across whose surface is covered in mirror-like solar concentrators that direct sunlight to a strip of photovoltaic cells, where it’s converted into electricity. Beneath the surface is an integrated circuit that converts the power from the solar cells into microwaves, which are then transmitted out of the bottom of the tile by an array of ultra-thin and flexible patch antennas.
Image Credit: CaltechThis design generates significant weight savings, because it removes the need for bulky wiring to transport the generated electricity to a central transmitter. These tiles will then be arranged into strips and integrated into a novel folding structure that will be compact at launch and then unfurl once in space.
The result will be a self-contained spacecraft that is able to deploy itself, generate power, and transmit it back to Earth, but the vision involves combining many of these to create arrays able to produce comparable amounts of energy to a land-based system. That setup makes it easy to adjust the size and configuration of arrays, and also means that damage to individual modules won’t put the entire system out of action.
The experiments launched this week are designed to test several of the key underlying technologies behind this architecture. One called DOLCE (Deployable on-Orbit ultraLight Composite Experiment) will test out the unfurling mechanism by deploying a six-by-six-foot frame from a small trash-bin-sized canister.
Another called MAPLE (Microwave Array for Power-transfer Low-orbit Experiment) will test out an array of ultralight microwave transmitters designed to beam power over a distance in space. A final experiment called ALBA will put 32 different types of photovoltaic cells through their paces over several months to see which operates most effectively in the punishing environment of space.
Assuming all of the tests go according to plan, the researchers will have validated some of the key technologies required to make their vision a reality. But given the dropping price of solar power on Earth and the growing prevalence of energy storage technologies designed to deal with solar power’s intermittency, there are question marks over the economics and practicality of the idea.
The technology could play an important role in the longer run, though, John Timmer writes in Ars Technica. Most estimates suggest that we should be able to switch around 70 percent of our grid to renewable energy fairly easily, but the lack of reliability due to seasonal changes or rare weather events mean that going higher could be tough.
A source of renewable energy that’s available 24 hours a day, 7 days a week could help to plug the gap when conditions on Earth lead to a drop in generation. We are likely still decades away from needing that, but given how complex space-based solar power is, getting to work on the technology now seems like a smart bet.
Image Credit: Caltech
In the not-too-distant future, we’re going to need new ways to move. Cars are convenient and relatively affordable, but they’re also one of the world’s biggest sources of emissions, not to mention the traffic, noise, and accidents that come with them. So how might we get from point A to point B in more efficient, Earth-friendly ways?
There are plenty of options on display at this year’s Consumer Electronics Show (CES) in Las Vegas. From autonomous shuttles to personal aircraft to three-wheeled “autocycles,” it seems there’ll be no shortage of alternatives to the combustion-engine cars that fill our roads today. Which of these ends up being widely adopted and which fade into irrelevance will be revealed with time, but until then, it’s fun to contemplate a future where they all exist simultaneously.
The Squad Solar CarImage Credit: Squad MobilityDutch startup Squad Mobility is showcasing its Squad car, which it describes as a compact city car that charges itself on solar energy through a solar panel on its roof. It’s not the first of its kind; there are at least three other companies working on solar-powered cars, including Aptera Motors, Sono Motors, and Lightyear. These cars run on batteries and can be plugged into an outlet to charge, but they can also charge with sunlight. But while the Aptera, Sion, and Lightyear 0 are truly cars, the Squad is closer to a golf cart due to its lower speed (28 miles per hour max). You don’t even need a driver’s license to operate it. Charged just on sun for a day, the car can drive up to 12 miles, and doesn’t require full sun to charge. The company says the car will be available in 2024 starting at $6,250.
The GreenStreet AutocycleImage Credit: GreenStreetThe makers of this three-wheeled cross between a car and a motorcycle say it combines “the excitement of a tourer with the practicality of a daily commuter.” Classified as a motorcycle by the US Department of Transportation, the Autocycle is all-electric, can travel 250 miles on a single charge, and go from 0-60 in 6 seconds. With a maximum speed of 80 miles per hour, it’s slower than both cars and motorcycles, but that’s probably good since it has an open cab. The first prototype is on display at the show, and the company’s aiming to bring the Autocycle to market in 2024 (at a price as yet unspecified). The odds of ever seeing these on roads seem pretty low, but the concept is fun.
A (Partially) 3D Printed Electric CarImage Credit: APMAAn entire 3D printed car may exist in the future, but for now it’s just one component that’s being made this way, and a crucial one: the chassis. Made by Canadian company Xaba, the 3D printed chassis is part of a zero-emissions concept car called Project Arrow. Xaba’s aiming to make manufacturing more sustainable using AI-driven control systems and automation. 3D printing auto parts could not only lower their cost, it could make it easier to incorporate sustainable or recycled materials.
Holon Autonomous People-MoverImage Credit: HolonIf we’re being practical, the future of mobility should involve more sharing. The Holon looks like a nicer, more comfortable version of a city bus, with big windows and seats that face each other—and notably, no driver’s seat. Its maximum speed is 60 kilometers per hour (37 miles per hour) and can fit up to 15 passengers. Due to its slow speed and small size compared to a city bus, the Holon would most likely be used as a shuttle in places like college campuses or airports. It was designed by Italian car company Pininfarina, and Israeli autonomous driving tech company Mobileye is developing its self-driving system. Holon is scheduled to start production in the US in 2025.
Icoma Tatamel EbikeImage Credit: IcomaThis electric bike from Japanese company Icoma is still in the concept stage, but it’s a pretty cool concept, partly because it can function as a portable power station for your home. When you’re using it as a bike (which would be the case probably 99 percent of the time), its 600-watt motor can go up to 25 miles per hour for about 18 miles on a single charge. Once the battery’s spent it takes three hours to recharge, and it folds up to be even smaller than it already is. Its one big drawback, besides the fact that it’ll probably be pretty pricey, is its 110-pound weight; you won’t be storing this thing anywhere that requires going up or down a few stairs. The company wants to launch the Tatamel in the US this spring.
The Aska A5 Flying CarFormally called an electric drive and fly vertical takeoff and landing (eVTOL) vehicle, the Aska A5 could also be called an aircraft you can drive on roads or a car that can take to the skies. Unlike most of the mobility solutions on display at CES, it’s not all-electric; rather, its power system has both lithium-ion battery packs and a gasoline engine that acts as an onboard range extender. These give the vehicle a 250-mile flight range carrying its maximum of four passengers. It can take off vertically like a helicopter, or from a runway like planes, and its wings fold in when it’s not flying. In drive mode, all four of its wheels are placed outside the fuselage for better traction and aerodynamics. The company is developing a ride-sharing service using the vehicles, and is targeting availability in major cities by 2026. That sounds like a long shot, but maybe one day this Jetsons-like dream will become reality.
Check back throughout the week, because CES is just getting started, and this list could very well grow to include even more wheeled and winged futuristic modes of transportation.
Banner Image Credit: Aska
Black holes form natural time machines that allow travel to both the past and the future. But don’t expect to be heading back to visit the dinosaurs anytime soon.
At present, we don’t have spacecraft that could get us anywhere near a black hole. But even leaving that small detail aside, attempting to travel into the past using a black hole might be the last thing you ever do.
What Are Black Holes?A black hole is an extremely massive object that is typically formed when a dying star collapses in on itself.
Like planets and stars, black holes have gravitational fields around them. A gravitational field is what keeps us stuck to Earth, and what keeps Earth revolving around the sun.
As a rule of thumb, the more massive an object is, the stronger its gravitational field.
Earth’s gravitational field makes it extremely difficult to get to space. That’s why we build rockets: we have to travel very fast to break out of Earth’s gravity.
The gravitational field of a black hole is so strong that even light can’t escape it. That’s impressive, since light is the fastest thing known to science!
Incidentally, that’s why black holes are black: we can’t bounce light off a black hole the way we might bounce a torch light off a tree in the dark.
Stretching SpaceAlbert Einstein’s general theory of relativity tells us matter and energy have a curious effect on the universe. Matter and energy bend and stretch space. The more massive an object is, the more space is stretched and bent around it.
A massive object creates a kind of valley in space. When objects come near, they fall into the valley. That’s why, when you get close enough to any massive object, including a black hole, you fall towards it. It’s also why light can’t escape a black hole: the sides of the valley are so steep that light isn’t going fast enough to climb out.
The valley created by a black hole gets steeper and steeper as you approach it from a distance. The point at which it gets so steep that light can’t escape is called the event horizon. Event horizons aren’t just interesting for would-be time travelers: they’re also interesting for philosophers, because they have implications for how we understand the nature of time.
Stretching TimeWhen space is stretched, so is time. A clock that is near a massive object will tick slower than one that is near a much less massive object.
A clock near a black hole will tick very slowly compared to one on Earth. One year near a black hole could mean 80 years on Earth, as you may have seen illustrated in the movie Interstellar.
In this way, black holes can be used to travel to the future. If you want to jump into the future of Earth, simply fly near a black hole and then return to Earth. If you get close enough to the center of the black hole, your clock will tick slower, but you should still be able to escape so long as you don’t cross the event horizon.
Loops in TimeWhat about the past? This is where things get truly interesting. A black hole bends time so much that it can wrap back on itself.
Imagine taking a sheet of paper and joining the two ends to form a loop. That’s what a black hole seems to do to time. This creates a natural time machine. If you could somehow get onto the loop, which physicists call a closed timelike curve, you would find yourself on a trajectory through space that starts in the future and ends in the past.
Inside the loop, you would also find that cause and effect get hard to untangle. Things that are in the past cause things to happen in the future, which in turn cause things to happen in the past.
The CatchSo, you’ve found a black hole and you want to use your trusty spaceship to go back and visit the dinosaurs. Good luck.
There are three problems. First, you can only travel into the black hole’s past. That means that if the black hole was created after the dinosaurs died out, then you won’t be able to go back far enough.
Second, you’d probably have to cross the event horizon to get into the loop. This means that to get out of the loop at a particular time in the past, you’d need to exit the event horizon. That means travelling faster than light, which we’re pretty sure is impossible.
Third, and probably worst of all, you and your ship would undergo “spaghettification.” Sounds delicious, right?
Sadly, it’s not. As you crossed the event horizon you would be stretched flat, like a noodle. In fact, you’d probably be stretched so thin that you’d just be a string of atoms spiraling into the void.
So, while it’s fun to think about the time-warping properties of black holes, for the foreseeable future that visit to the dinosaurs will have to stay in the realm of fantasy.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: NASA / Goddard Space Flight Center / Jeremy Schnittman
Remember when predicting protein shapes using AI was the breakthrough of the year?
That’s old news. Having solved nearly all protein structures known to biology, AI is now turning to a new challenge: designing proteins from scratch.
Far from an academic pursuit, the endeavor is a potential game-changer for drug discovery. Having the ability to draw up protein drugs for any given target inside the body—such as those triggering cancer growth and spread—could launch a new universe of medicines to tackle our worst medical foes.
It’s no wonder multiple AI powerhouses are answering the challenge. What’s surprising is that they converged on a similar approach. This year DeepMind, Meta, and Dr. David Baker’s team at the University of Washington all took inspiration from an unlikely source: DALL-E and GPT-3.
These generative algorithms have taken the world by storm. When given just a few simple prompts in everyday English, the programs can produce mind-bending images, paragraphs of creative writing, or film scenes, and even remix the latest fashion designs. The same underlying technology recently took a stab at writing computer code, besting nearly half of human competitors in a highly challenging programming task.
What does any of that have to do with proteins?
Here’s the thing: proteins are essentially strings of “letters” molded into secondary structures—think sentences—and then 3D “paragraphs.” If AI can generate gorgeous images and clean writing, why not co-opt the technology to rewrite the code of life?
Here Come the ChampionsProtein is the key to life. It builds our bodies. It runs our metabolisms. It underlies intricate brain functions. It’s also the basis for a wealth of new drugs that could treat some of our most insurmountable health problems to date—and create new sources of biofuels, lab-grown meats, or even entirely novel lifeforms through synthetic biology.
While “protein” often evokes pictures of chicken breasts, these molecules are more similar to an intricate Lego puzzle. Building a protein starts with a string of amino acids—think a myriad of Christmas lights on a string— which then fold into 3D structures (like rumpling them up for storage).
DeepMind and Baker both made waves when they each developed algorithms to predict the structure of any protein based on their amino acid sequence. It was no simple endeavor; the predictions were mapped at the atomic level.
Designing new proteins raises the complexity to another level. This year Baker’s lab took a stab at it, with one effort using good old screening techniques and another relying on deep learning hallucinations. Both algorithms are extremely powerful for demystifying natural proteins and generating new ones, but they were hard to scale up.
But wait. Designing a protein is a bit like writing an essay. If GPT-3 and ChatGPT can write sophisticated dialogue using natural language, the same technology could in theory also rejigger the language of proteins—amino acids—to form functional proteins entirely unknown to nature.
AI Creativity Meets BiologyOne of the first signs that the trick could work came from Meta.
In a recent preprint paper, they tapped into the AI architecture underlying DALL-E and ChatGPT, a type of machine learning called large language models (LLMs), to predict protein structure. Instead of feeding the models exuberant amounts of text or images, the team instead trained them on amino acid sequences of known proteins. Using the model, Meta’s AI predicted over 600 million protein structures by reading their amino acid “letters” alone—including esoteric ones from microorganisms in the soil, ocean water, and our bodies that we know little about.
More impressively, the AI, called ESMFold, eventually learned to “autocomplete” protein sequences even when some amino acid letters were obscured. Although not as accurate as DeepMind’s AlphaFold, it ran roughly 60 times faster, making it easier to scale up to larger databases.
Baker’s lab took the protein “autocomplete” function to a new level in a preprint published earlier this month. If AI can already fill in the blanks when it comes to predicting protein structures, a similar principle could potentially also generate proteins from a prompt—in this case, its potential biological function.
The key came down to diffusion models, a type of machine learning algorithm that powers DALL-E. Put simply, these neural networks are especially good at adding and then removing noise from any given data—be it images, texts, or protein sequences. During training, they first destroy training data by adding noise. The model then learns to recover the original data by reversing the process through a step called denoising. It’s a bit like dismantling a laptop or other electronic and putting it back together to see how different components work.
Because diffusion models usually start with scrambled data (say, all the pixels of an image are rearranged into noise) and eventually learn to reconstruct the original image, it’s especially effective at generating new images—or proteins—from seemingly random samples.
Baker’s lab tapped into the approach with a bit of fine-tuning of their signature RoseTTAFold structure prediction network. Previously, a version of the software generated protein scaffolds—the backbone of a protein—in just a single step. But proteins aren’t uniform blobs: each has multiple hotspots that allow them to physically tag onto each other, which triggers various biological processes. When RoseTTAFold faced tough problems—such as designing protein hotspots with minimal knowledge—it struggled.
The team’s solution was to integrate RoseTTAFold with a diffusion model, with the former helping with the denoising step. The resulting algorithm, RoseTTAFold Diffusion (RF Diffusion), is a love-child between protein structure prediction and creative generation. The AI designed a wide range of elaborate proteins with little resemblance to any known protein structures, constrained by pre-defined but biologically relevant limits.
Designing proteins is just the first step. The next is translating these digital designs into actual proteins and seeing how they work in cells. In one test, the team took 44 candidates with antibacterial and antiviral potential and made the proteins inside the trusty E. Coli bacteria. Over 80 percent of the AI designer proteins folded into their predicted final form. This isquite the feat, as several sub-units had to come together in specific numbers and orientations.
The proteins also grabbed onto their intended targets. One example had a protein structure binding to SARS-CoV-2, the virus that causes Covid-19. The AI design specifically honed in on the virus’s spike protein, the target for Covid-19 vaccines.
In another example, the AI designed a protein that binds to a hormone to regulate calcium levels in the blood. The resulting candidate readily grabbed onto the target—so much so that it needed just a tiny amount. Speaking to MIT Technology Review, Baker said the AI seemed to pull protein drug solutions “out of thin air.”
“These works reveal just how powerful diffusion models can be for protein design,” said study author Dr. Joseph Watson.
Do AIs Dream of Molecular Sheep?Baker’s lab isn’t the only one chasing AI-based protein drugs.
Generate Biomedicines, a startup based in Massachusetts, also has its eyes on diffusion models for generating proteins. Dubbed Chroma, their software works similarly to RF Diffusion, including the generated proteins adhering to biophysical constraints. According to the company, Chroma can generate large proteins—over 4,000 amino acid residues—in just a few minutes on a GPU (graphics processing unit).
While just ramping up, it’s clear that the race for on-demand protein drug design is on. “It’s extremely exciting,” said David Juergens, author of the RF Diffusion study, “and it’s really just the beginning.”
Image Credit: Ian Haydon / Institute for Protein Design / University of Washington
Charles Darwin believed evolution created “endless forms most beautiful.” It’s a nice sentiment but it doesn’t explain why evolution keeps making crabs.
Scientists have long wondered whether there are limits to what evolution can do or if Darwin had the right idea. The truth may lie somewhere between the two.
While there doesn’t seem to be a ceiling on the number of species that might evolve, there may be restraints on how many fundamental forms those species can evolve into. The evolution of crab-like creatures may be one of the best examples of this, since they have evolved not just once but at least five times.
Crabs belong to a group of crustaceans called decapods—literally “ten footed”, since they have five pairs of walking legs. Some decapods, like lobsters and shrimp, have a thick, muscular abdomen, which is the bulk of the animal that we eat. With a quick flick of their abdomen lobsters can shoot off backwards and escape predators.
Crabs, by contrast, have a compressed abdomen, tucked away under a flattened but widened thorax and shell. This allows them to scuttle into rock crevices for protection. Evolution repeatedly hit upon this solution because it works well under similar sets of circumstances.
Five Groups of “Crabs”King crabs evolved from lobster-like ancestors within the Anomura. Image Credit: CSIRO, CC BYThe largest crab group are the Brachyura (true crabs) including the edible crab and Atlantic blue crab. They had an ancestor that was also crab shaped. Some species have evolved “backwards” and straightened out their abdomens again. The other large group are the Anomura (false crabs), with an ancestor that looked more like a lobster.
However, at least four groups of Anomura—sponge crabs, porcelain crabs, king crabs, and the Australian hairy stone crab—have independently evolved into a crab-like form in much the same way as the true crabs. Like the true crabs, their compact bodies are more defensive, and can move sideways faster.
This means “crabs” aren’t a real biological group. They are a collection of branches in the decapod tree that evolved to look the same.
Hairy stone crab (Lomis hirta). Image Credit: Tim Binns / Wikimedia Commons, CC BY-SABut crabs aren’t the exception.
Something similar happened in the evolution of birds from feathered dinosaurs. Feathers may have first evolved for insulation, to attract mates, for protecting eggs and possibly also as “nets” for catching prey. Millions of years later, feathers elongated and streamlined for flying.
Palaeontologists disagree about the details, but all modern birds (Neoaves) evolved from ground-dwelling ancestors just after the mass extinction that wiped out the other dinosaurs. However, feathered wings and flight also evolved earlier in other groups of dinosaurs, including troodontids and dromaeosaurs. Some of these, like Microraptor, had four wings.
Microraptors had two pairs of wings. Image Credit: Fred Wierum / Wikimedia CommonsRe-Running the Tape of LifeUnfortunately we can’t run evolutionary experiments to see if the same things keep happening because that would take hundreds of millions of years. But the history of life has already done something similar to that for us, when closely related lineages evolve and diversify on different continents. In many cases, these ancestral lines repeatedly came up with the same or almost identical solutions to problems.
One of the best examples is our own group, the mammals.
There are two major groups of living mammals. The placentals (including us) and the marsupials (pouched mammals who give birth to tiny young). Both groups evolved from the same common ancestor over 100 million years ago, the marsupials largely in Australasia and the Americas and the placentals elsewhere.
This isolation led to two almost independent runs of the “experiment” to see what could be done with the mammal body plan. There are marsupial and placental versions of moles, mice, anteaters, gliders, and cats. There was even a marsupial wolf (the thylacine, extinct in 1936), whose skull and teeth match those of the placental wolf in astonishing detail.
Skulls of the marsupial thylacine (left) and placental wolf (right) show striking convergence, despite evolving apart on different continents.It’s not only body forms that evolve independently, but also organs and other structures. Humans have complex camera eyes with a lens, iris and retina. Squid, and octopuses, which are molluscs and more closely related to snails and clams, also evolved camera eyes with the same components.
Eyes more generally may have evolved independently up to 40 times in different groups of animals. Even box jellyfish, which don’t have a brain, have eyes with lenses at the bases of their four tentacles.
The more we look, the more we find. Structures such as jaws, teeth, ears, fins, legs and wings all keep evolving independently across the animal tree of life.
More recently, scientists discovered convergence also happens at the molecular level. The opsin molecules in eyes that convert photons of light into chemical energy and enable humans to see have a tight resemblance to those in box jellyfish, and evolved that way in parallel. Even more bizarrely, animals as different as whales and bats have striking convergence in the genes that enable them to echolocate.
Are Humans Really Unique?Many of the things we like to think make humans special have been reinvented by evolution elsewhere. Corvids like crows and ravens have problem-solving intelligence and, along with owls, can use simple tools.
Whales and dolphins have complex social structures, and their big brains allowed them to develop language. Dolphins use tools like sponges to cover their noses while they forage across stony sea bottoms. Octopuses also use tools and learn from watching what happens to other octopuses.
Octopus marginatus hiding between two shells from East Timor. Image Credit: Nick Hobgood, CC BYIf things keep evolving in similar ways here on Earth, there’s a possibility they might also follow a related course if life has evolved elsewhere in the universe. It might mean extraterrestrial beings look less alien and more familiar than we expect.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: vastateparkstaff / Wikimedia Commons
Every Saturday we post a selection of articles from the week. These stories might include an eye-catching bit of news or a deep dive into a bigger theme. With the end of the year here, we dug through every one of those posts again to surface 25 stories that managed to stay fresh amid another wild year of science and tech news.
Some big trends stood out amid the chatter. This was the year of generative AI. Algorithms producing words and images aren’t new, but with the likes of DALL-E 2, Midjourney, Stable Diffusion, and ChatGPT, they became a mainstream hit in 2022. And even as dodgy claims of machine sentience were firmly refuted, applications for “synthetic creativity,” as Kevin Kelly called it, widened beyond the literary and visual to include AI hallucinations of drug molecules and math proofs. With next-generation algorithms already in the works, expect more in 2023.
Even as AI hit new highs, pandemic-fueled trends whipsawed the other way.
Companies, from Amazon to Meta, announced layoffs as tech stocks dived. Big visions, including self-driving cars and the metaverse, seemed to gain little ground as notable projects, like Argo AI, pulled the plug. Maybe the hottest such trend, cryptocurrency, was deep into an epic slump when FTX, one of the world’s biggest crypto exchanges, imploded overnight. And of course, under new owner Elon Musk, drama at Twitter continues to haunt headlines.
Still, even as tech continues wading through the current correction, longer trends in science and technology are unlikely to slow down. “While the economy reliably fluctuates between boom and bust, and valuations rise and fall on Wall Street’s whims, the technology itself goes in only one direction,” Steven Levy recently wrote for Wired. “Connection speeds get faster, chips get more capacity, and rocket ships get more reliably reusable.”
Beyond the dominant themes, we found a few standalone gems too, like a deep exploration into the threat posed by enormous solar storms, a fascinating study into doppelgängers, a CEO who takes his “flying car” to work, and a road trip to the edge of the universe.
Without further ado: Here’s this year’s list. Enjoy! See you in 2023.
Picture Limitless Creativity at Your Fingertips
Kevin Kelly | Wired“For the first time in history, humans can conjure up everyday acts of creativity on demand, in real time, at scale, for cheap. Synthetic creativity is a commodity now. Ancient philosophers will turn in their graves, but it turns out that to make creativity—to generate something new—all you need is the right code. We can insert it into tiny devices that are presently inert, or we can apply creativity to large statistical models, or embed creativity in drug discovery routines. What else can we use synthetic creativity for?”
An End to Doomerism
Hannah Ritchie | Big Think“The issue is that people mistake optimism for ‘blind optimism’—the blinkered faith that things will always get better. Problems will fix themselves. If we just hope things turn out well, they will. Blind optimism really is dumb. And it’s not just stupid, it’s dangerous. If we sit back and do nothing, we will not make progress. That’s not the kind of optimism that I’m talking about. Optimism is seeing problems as challenges that are solvable; it’s having the confidence that there are things that we can do to make a difference.”
Paradise at the Crypto Arcade: Inside the Web3 Revolution
Gilad Edelman | Wired
“…to a core of true believers, Web3 stands apart from the garish excesses and brazen misbehavior of the flashing-neon crypto casino. If cryptocurrency was originally about decentralizing money, Web3 is about decentralizing…everything. Its mission is almost achingly idealistic: to free humanity not only from Big Tech domination but also from exploitative capitalism itself—and to do it purely through code.”
Yann LeCun Has a Bold New Vision for the Future of AI
Melissa Heikkiläarchive page and Will Douglas Heaven | MIT Technology Review
“In a draft document shared with MIT Technology Review, LeCun sketches out an approach that he thinks will one day give machines the common sense they need to navigate the world. For LeCun, the proposals could be the first steps on a path to building machines with the ability to reason and plan like humans—what many call artificial general intelligence, or AGI.”
The State of the Transistor in 3 Charts
Samuel K. Moore and David Schneider | IEEE Spectrum“In 1947, there was only one transistor. According to TechInsight’s forecast, the semiconductor industry is on track to produce almost 2 billion trillion (10^21) devices this year. That’s more transistors than were cumulatively made in all the years prior to 2017.”
The Metaverse Is Inevitable, Regardless of What Happens to Meta
Louis Rosenberg | BigThink“The metaverse is about transforming how we humans experience the digital world. So far, digital content has been accessed primarily through flat media viewed in the third-person. In the metaverse, our digital lives increasingly will involve immersive media that appears all around us and is experienced in the first-person. Regardless of Meta’s fate, the metaverse is inevitable because the human organism evolved to understand our world through first-person experiences in spatial environments.”
CRISPR, 10 Years On: Learning to Rewrite the Code of Life
Carl Zimmer | The New York Times
“i‘I remember thinking very clearly, when we publish this paper, it’s like firing the starting gun at a race,’ [Jennifer Doudna] said. In just a decade, CRISPR has become one of the most celebrated inventions in modern biology. It is swiftly changing how medical researchers study diseases: Cancer biologists are using the method to discover hidden vulnerabilities of tumor cells. Doctors are using CRISPR to edit genes that cause hereditary diseases. ‘The era of human gene editing isn’t coming,’ said David Liu, a biologist at Harvard University. ‘It’s here.’i“
Will Transformers Take Over Artificial Intelligence?
Stephen Ornes | Quanta
“Just 10 years ago, disparate subfields of AI had little to say to each other. But the arrival of transformers suggests the possibility of a convergence. ‘I think the transformer is so popular because it implies the potential to become universal,’ said the computer scientist Atlas Wang of the University of Texas, Austin. ‘We have good reason to want to try transformers for the entire spectrum’ of AI tasks.’i”
Google’s ‘Sentient’ Chatbot Is Our Self-Deceiving Future
Ian Bogost | The Atlantic
“…a Google engineer became convinced that a software program was sentient after asking the program, which was designed to respond credibly to input, whether it was sentient. A recursive just-so story. I’m not going to entertain the possibility that LaMDA is sentient. (It isn’t.) More important, and more interesting, is what it means that someone with such a deep understanding of the system would go so far off the rails in its defense, and that, in the resulting media frenzy, so many would entertain the prospect that Lemoine is right.”
What If We Didn’t Have to Test New Drugs on Animals?
Emily Sohn | Neo.Life“In one place [the bipartisan FDA Modernization Act 2.0] changes the word ‘preclinical’ to ‘nonclinical,’ and in another it replaces the word ‘animal’ with the more anodyne ‘nonclinical tests or studies.’ That may not sound like a lot. Enshrined into law, it would eliminate an 85-year-old requirement that pharmaceutical companies must test drugs on animals before starting clinical trials in people and would usher in a new era of cell-based or computer-based testing instead.”
Twitter’s Potential Collapse Could Wipe Out Vast Records of Recent Human History
Chris Stokel-Walker | MIT Technology Review“Almost from the time the first tweet was posted in 2006, Twitter has played an important role in world events. The platform has been used to record everything from the Arab Spring to the ongoing war in Ukraine. It’s also captured our public conversations for years. But experts are worried that if Elon Musk tanks the company, these rich seams of media and conversation could be lost forever. Given his admission to employees in a November 10 call that Twitter could face bankruptcy, it’s a real and present risk.”
*This Is Life in the Metaverse
Kashmir Hill | The New York Times*“My goal was to visit at every hour of the day and night, all 24 of them at least once, to learn the ebbs and flows of Horizon and to meet the metaverse’s earliest adopters. I gave up television, books and a lot of sleep over the past few months to spend dozens of hours as an animated, floating, legless version of myself. I wanted to understand who was currently there and why, and whether the rest of us would ever want to join them.”
Here Comes the Sun—to End Civilization
Matt Ribel | Wired
“When another big [coronal mass ejection] heads our way, as it could at any time, existing imaging technology will offer one or two days’ notice. But we won’t understand the true threat level until the cloud reaches the Deep Space Climate Observatory, a satellite about a million miles from Earth. It has instruments that analyze the speed and polarity of incoming solar particles. If a cloud’s magnetic orientation is dangerous, this $340 million piece of equipment will buy humanity—with its 7.2 billion cell phones, 1.5 billion automobiles, and 28,000 commercial aircraft—at most one hour of warning before impact.”
Can Computers Learn Common Sense?
Matthew Hutson | The New Yorker
“Oren Etzioni, the CEO of the Allen Institute for Artificial Intelligence, in Seattle, told me that common sense is ‘the dark matter’ of AI.’ It ‘shapes so much of what we do and what we need to do, and yet it’s ineffable,’ he added. …If computer scientists could give their AI systems common sense, many thorny problems would be solved. …Such systems would be able to function in the world because they possess the kind of knowledge we take for granted.”
Quantum Computing Has a Hype Problem
Sankar Das Sarma | MIT Technology Review
“It took the aviation industry more than 60 years to go from the Wright brothers to jumbo jets carrying hundreds of passengers thousands of miles. The immediate question is where quantum computing development, as it stands today, should be placed on that timeline. Is it with the Wright brothers in 1903? The first jet planes around 1940? Or maybe we’re still way back in the early 16th century, with Leonardo da Vinci’s flying machine? I do not know. Neither does anybody else.”
Why Twitter Is More Powerful Than the Printing Press
Jessica E. Lessin | The Information
“…those who dismiss [Elon] Musk’s takeover of Twitter as just a modern example of a rich mogul buying printing presses or television stations fall into a dangerous trap. They forget that the internet is unlike any communication technology that has come before it; they underestimate the power of the technology to scale and to control the public conversation.”
MoMA’s Newest Artist Is an AI Trained on 180,000 Works, From Warhol to Pac-Man
Jesus Diaz | Fast Company“The colossal installation—a stunning 24- by 24-foot digital display that fills the entire MoMA lobby—renders an infinite animated flow of images, each of them dreamed up as you watch by an AI model fed by the museum’s entire collection of artwork. This flow is controlled by what happens around it, making the piece feel like it’s alive.”
How to Build a Wormhole in Just 3 (Nearly Impossible) Steps
Paul Sutter | Ars Technica
“You’ve got yourself a fancy new spaceship and you want to start on a five-year tour of the galaxy. But there’s a problem: Space is big. Really big. And even at the fastest speeds imaginable, it takes eons of crawling across the interstellar voids to get anywhere interesting. The solution? It’s time to build a wormhole. …It’s a staple of science-fiction, and it’s rooted in science-fact. How difficult could it be? Here’s a hint: incredibly difficult.”
Can We Prove the World Isn’t a Simulation?
David Chalmers | Nautilus
“You might think you have definitive evidence that you’re not [in a simulation]. I think that’s impossible, because any such evidence could be simulated. Maybe you think the glorious forest around you proves that your world isn’t a simulation. But in principle, the forest could be simulated down to every last detail, and every last bit of light that reaches your eyes from the forest could be simulated, too. Your brain will react exactly as it would in the nonsimulated, ordinary world, so a simulated forest will look exactly like an ordinary one.”
The Hibernator’s Guide to the Galaxy
Brendan I. Koerner | Wired“Scientists are on the verge of figuring out how to put humans in a state of suspended animation. It could be the key to colonizing Mars. …In recent years, these researchers have been piecing together the molecular changes that occur when certain species ratchet down their metabolism. And since so many hibernators are our close genomic cousins, there is good reason to believe that we can tweak our brains and bodies to mimic what they do.”
Please Ignore My Last 577 Tweets
Jacob Stern | The Atlantic
“If you had told me last Wednesday afternoon, when my Twitter account had a grand total of three tweets and 200-something followers, that roughly 24 hours later the account would have tweeted 577 times and boosted its follower count to 42,000, I would not have believed you. And if you had further told me that this unfathomable ascent was all part of a massive scam to con would-be Moonbird buyers out of tens of thousands of dollars in cryptocurrency, I would have asked you what a Moonbird is. And yet here we are.”
Your Doppelgänger Is Out There and You Probably Share DNA With Them
Kate Golembiewski | The New York Times“Because the doppelgängers’ appearances are more attributable to shared genes than shared life experiences, that means that, to some extent, their similarities are just the luck of the draw, spurred on by population growth. There are, after all, only so many ways to build a face. ‘Now there are so many people in the world that the system is repeating itself,’ Dr. Esteller said. It’s not unreasonable to assume that you, too, might have a look-alike out there.”
Can Planting a Trillion New Trees Save the World?
Zach St. George | The New York Times
“The idea that planting trees can effectively and simultaneously cure a host of the world’s most pressing maladies has become increasingly popular in recent years, bolstered by a series of widely cited scientific studies and by the inspiring and marketable goal, memorably proposed by a charismatic 13-year-old, of planting one trillion trees. …Nearly everyone agrees that planting trees can be a useful, wholesome activity. The problem is that, in practice, planting trees is more complicated than it sounds.”
Jetson CEO Takes His eVTOL on a Commute to Work
Loz Blain | New Atlas
“Walk out into your back yard, jump into a next-generation electric VTOL flying machine, lift off and soar your way to the office helipad: that’s the dream of personal eVTOL ownership, and Jetson co-founder Tomasz Patan has lived it, in a new video.”
How Long Is the Drive to the Edge of the Universe?
Randall Munroe | The New York Times“The edge of the observableuniverse is about 270,000,000,000,000,000,000,000 miles away. If you drive at a steady 65 miles per hour, it will take you 480,000,000,000,000,000—that’s 4.8 × 10¹⁷—years to get there, or 35 million times the current age of the universe. …Be sure to pack extra snacks.”
Image Credit: André Lopes / Unsplash
Ever since deep learning burst into the mainstream in 2012, the hype around AI research has often outpaced its reality. Over the past year though, a series of breakthroughs and major milestones suggest the technology may finally be living up to its promise.
Despite the obvious potential of deep learning, over the past decade the regular warnings about the dangers of runaway superintelligence and the prospect of technological unemployment were tempered by the fact that most AI systems were preoccupied with identifying images of cats or providing questionable translations from English to Chinese.
In the last year, however, there has been an undeniable step change in the capabilities of AI systems, in fields as varied as the creative industries, fundamental science, and computer programming. What’s more, these AI systems and their outputs are become increasingly visible and accessible to ordinary people.
Nowhere have the advances been more obvious than in the burgeoning field of generative AI, a catch-all term for a host of models muscling in on creative tasks.
This has been primarily thanks to a kind of model called a transformer, which was actually first unveiled by Google in 2017. Indeed, many of the AI systems that have made headlines this year are updates of models that their developers have been working on for some time, but the results they have produced in 2022 have blown previous iterations out of the water.
Most prominent among these is ChatGPT, an AI chatbot based on the latest version of OpenAI’s GPT-3 large language model. Released to the public at the end of November, the service has been wowing people with its uncanny ability to engage in natural-sounding conversations, answer complicated technical questions, and even produce convincing prose and poetry.
Earlier in the year, another OpenAI model called DALL-E 2 took the internet by storm with its ability to generate hyper-realistic images in response to prompts as bizarre as “a raccoon playing tennis at Wimbledon in the 1990s” and “Spider-Man from ancient Rome.” Meta took things a step further in September with a system that could produce short video clips from text prompts, and Google researchers have even managed to create an AI that can generate music in the style of an audio clip it is played.
The implications of this explosion in AI creativity and fluency are hard to measure right now, but they have already spurred predictions that it could replace traditional search engines, kill the college essay, and lead to the death of art.
This is as much due to the improving capabilities of these models as their increasing accessibility, with services like ChatGPT, DALL-E 2, and text-to-image generator Midjourney open to everyone for free (for now, at least). Going even further, the independent AI lab Stable Diffusion has even open-sourced their text-to-image AI, allowing anyone with a modestly powerful computer to run it themselves.
AI has also made progress in more prosaic tasks over the last year. In January, Deepmind unveiled AlphaCode, an AI-powered code generator that the company said could match the average programmer in coding competitions. In a similar vein, GitHub Co-pilot, an AI coding tool developed by GitHub and OpenAI, moved from a prototype to a commercial subscription service.
Another major bright spot for the field has been AI’s increasingly prominent role in fundamental science. In July, DeepMind announced that its groundbreaking AlphaFold AI had predicted the structure of almost every protein known to science, setting up a potential revolution in both the life sciences and drug discovery. The company also announced in February that it had trained its AI to control the roiling plasmas found inside experimental fusion reactors.
And while AI seems to be increasingly moving away from the kind of toy problems the field was preoccupied with over the past decade, it has also made major progress in one of the mainstays of AI research: games.
In November, Meta showed off an AI that ranked in the top 10 percent of players in the board game Diplomacy, which requires a challenging combination of strategy and natural language negotiation with other players. The same month, a team at Nvidia trained an AI to play the complex 3D videogame Minecraft using only high-level natural language instructions. And in December, DeepMind cracked the devilishly complicated game Stratego, which involves long-term planning, bluffing, and a healthy dose of uncertainty.
It’s not all been plain sailing, though. Despite the superficially impressive nature of the output of generative AI like ChatGPT, many have been quick to point out that they are highly convincing bullshit generators. They are trained on enormous amounts of text of variable quality from the internet. And ultimately all they do is guess what text is most likely to come after a prompt, with no capacity to judge the truthfulness of their output. This has raised concerns that the internet may soon be flooded with huge amounts of convincing-looking nonsense.
This was brought to light with the release of Meta’s Galactica AI, which was supposed to summarize academic papers, solve math problems, and write computer code for scientists to help speed up their research. The problem was that it would produce convincing-sounding material that was completely wrong or highly biased, and the service was pulled in just three days.
Bias is a significant problem for this new breed of AI, which is trained on vast tracts of material from the internet rather than the more carefully-curated datasets previous models were fed. Similar problems have surfaced with ChatGPT, which despite filters put in place by OpenAI can be tricked into saying that only white and Asian men make good scientists. And popular AI image generation app Lensa has been called out for sexualizing women’s portraits, particularly those of Asian descent.
Other areas of AI have also had a less-than-stellar year. One of the most touted real-world use cases, self-driving cars, has seen significant setbacks, with the closure of Ford and Volkswagen-backed Argo, Tesla fending off claims of fraud over its failure to deliver “full self-driving,” and a growing chorus of voices claiming the industry is stuck in a rut.
Despite the apparent progress that’s been made, there are also those, such as Gary Marcus, who say that deep learning is reaching its limits, as it’s not capable of truly understanding any of the material it’s being trained on and is instead simply learning to make statistical connections that can produce convincing but often flawed results.
But for those behind some of this year’s most impressive results, 2022 is simply a taste of what’s to come. Many predict that the next big breakthroughs will come from multi-modal models that combine increasingly powerful capabilities in everything from text to imagery and audio. Whether the field can keep up the momentum in 2023 remains to be seen, but either way this year is likely to go down as a watershed moment in AI research.
Image Credit: DeepMind / Unsplash
To see what the future might look like it is often helpful to study our history. This is what I will do in this article. I retrace the brief history of computers and artificial intelligence to see what we can expect for the future.
How Did We Get Here?How rapidly the world has changed becomes clear by how even quite recent computer technology feels ancient to us today. Mobile phones in the ‘90s were big bricks with tiny green displays. Two decades before that the main storage for computers was punch cards.
In a short period computers evolved so quickly and became such an integral part of our daily lives that it is easy to forget how recent this technology is. The first digital computers were only invented about eight decades ago, as the timeline shows.
Since the early days of this history, some computer scientists have strived to make machines as intelligent as humans. The next timeline shows some of the notable artificial intelligence systems and describes what they were capable of.
The first system I mention is the Theseus. It was built by Claude Shannon in 1950 and was a remote-controlled mouse that was able to find its way out of a labyrinth and could remember its course.1 In seven decades the abilities of artificial intelligence have come a long way.
Language and Image Recognition Capabilities of AI Systems Are Now Comparable to Those of HumansThe language and image recognition capabilities of AI systems have developed very rapidly.
The chart shows how we got here by zooming into the last two decades of AI development. The plotted data stems from a number of tests in which human and AI performance were evaluated in five different domains, from handwriting recognition to language understanding.
Within each of the five domains the initial performance of the AI system is set to -100, and human performance in these tests is used as a baseline that is set to zero. This means that when the model’s performance crosses the zero line is when the AI system scored more points in the relevant test than the humans who did in the same test.2
Just 10 years ago, no machine could reliably provide language or image recognition at a human level. But, as the chart shows, AI systems have become steadily more capable and are now beating humans in tests in all these domains.
Outside of these standardized tests the performance of these AIs is mixed. In some real-world cases these systems are still performing much worse than humans. On the other hand, some implementations of such AI systems are already so cheap that they are available on the phone in your pocket: image recognition categorizes your photos and speech recognition transcribes what you dictate.
From Image Recognition to Image GenerationThe previous chart showed the rapid advances in the perceptive abilities of artificial intelligence. AI systems have also become much more capable of generating images.
This series of nine images shows the development over the last nine years. None of the people in these images exist; all of them were generated by an AI system.
The series begins with an image from 2014 in the top left, a primitive image of a pixelated face in black and white. As the first image in the second row shows, just three years later AI systems were already able to generate images that were hard to differentiate from a photograph.
In recent years, the capability of AI systems has become much more impressive still. While the early systems focused on generating images of faces, these newer models broadened their capabilities to text-to-image generation based on almost any prompt. The image in the bottom right shows that even the most challenging prompts—such as “A Pomeranian is sitting on the King’s throne wearing a crown. Two tiger soldiers are standing next to the throne”—are turned into photorealistic images within seconds.4
Language Recognition and Production Is Developing FastJust as striking as the advances of image-generating AIs is the rapid development of systems that parse and respond to human language.
Shown in the image are examples from an AI system developed by Google called PaLM. In these six examples, the system was asked to explain six different jokes. I find the explanation in the bottom right particularly remarkable: the AI explains an anti-joke that is specifically meant to confuse the listener.
AIs that produce language have entered our world in many ways over the last few years. Emails get auto-completed, massive amounts of online texts get translated, videos get automatically transcribed, school children use language models to do their homework, reports get auto-generated, and media outlets publish AI-generated journalism.
AI systems are not yet able to produce long, coherent texts. In the future, we will see whether the recent developments will slow down—or even end—or whether we will one day read a bestselling novel written by an AI.
Where We Are Now: AI Is HereThese rapid advances in AI capabilities have made it possible to use machines in a wide range of new domains:
When you book a flight, it is often an artificial intelligence, and no longer a human, that decides what you pay. When you get to the airport, it is an AI system that monitors what you do at the airport. And once you are on the plane, an AI system assists the pilot in flying you to your destination.
AI systems also increasingly determine whether you get a loan, are eligible for welfare, or get hired for a particular job. Increasingly they help determine who gets released from jail.
Several governments are purchasing autonomous weapons systems for warfare, and some are using AI systems for surveillance and oppression.
AI systems help to program the software you use and translate the texts you read. Virtual assistants, operated by speech recognition, have entered many households over the last decade. Now self-driving cars are becoming a reality.
In the last few years, AI systems helped to make progress on some of the hardest problems in science.
Large AIs called recommender systems determine what you see on social media, which products are shown to you in online shops, and what gets recommended to you on YouTube. Increasingly they are not just recommending the media we consume, but based on their capacity to generate images and texts, they are also creating the media we consume.
Artificial intelligence is no longer a technology of the future; AI is here, and much of what is reality now would have looked like sci-fi just recently. It is a technology that already impacts all of us, and the list above includes just a few of its many applications.
The wide range of listed applications makes clear that this is a very general technology that can be used by people for some extremely good goals—and some extraordinarily bad ones, too. For such ‘dual use technologies’, it is important that all of us develop an understanding of what is happening and how we want the technology to be used.
Just two decades ago the world was very different. What might AI technology be capable of in the future?
What Is Next?The AI systems that we just considered are the result of decades of steady advances in AI technology.
The big chart below brings this history over the last eight decades into perspective. It is based on the dataset produced by Jaime Sevilla and colleagues.7
Each small circle in this chart represents one AI system. The circle’s position on the horizontal axis indicates when the AI system was built, and its position on the vertical axis shows the amount of computation that was used to train the particular AI system.
Training computation is measured in floating point operations, or FLOP for short. One FLOP is equivalent to one addition, subtraction, multiplication, or division of two decimal numbers.
All AI systems that rely on machine learning need to be trained, and in these systems training computation is one of the three fundamental factors that are driving the capabilities of the system. The other two factors are the algorithms and the input data used for the training. The visualization shows that as training computation has increased, AI systems have become more and more powerful.
The timeline goes back to the 1940s, the very beginning of electronic computers. The first shown AI system is ‘Theseus’, Claude Shannon’s robotic mouse from 1950 that I mentioned at the beginning. Towards the other end of the timeline you find AI systems like DALL-E and PaLM, whose abilities to produce photorealistic images and interpret and generate language we have just seen. They are among the AI systems that used the largest amount of training computation to date.
The training computation is plotted on a logarithmic scale, so that from each grid-line to the next it shows a 100-fold increase. This long-run perspective shows a continuous increase. For the first six decades, training computation increased in line with Moore’s Law, doubling roughly every 20 months. Since about 2010 this exponential growth has sped up further, to a doubling time of just about 6 months. That is an astonishingly fast rate of growth.8
The fast doubling times have accrued to large increases. PaLM’s training computation was 2.5 billion petaFLOP, more than 5 million times larger than that of AlexNet, the AI with the largest training computation just 10 years earlier.9
Scale-up was already exponential and has sped up substantially over the past decade. What can we learn from this historical development for the future of AI?
Studying the Long-Run Trends to Predict the Future of AIAI researchers study these long-term trends to see what is possible in the future.11
Perhaps the most widely discussed study of this kind was published by AI researcher Ajeya Cotra. She studied the increase in training computation to ask at what point in time the computation to train an AI system could match that of the human brain. The idea is that at this point the AI system would match the capabilities of a human brain. In her latest update, Cotra estimated a 50% probability that such “transformative AI” will be developed by the year 2040, less than two decades from now.12
In a related article, I discuss what transformative AI would mean for the world. In short, the idea is that such an AI system would be powerful enough to bring the world into a ‘qualitatively different future’. It could lead to a change at the scale of the two earlier major transformations in human history, the agricultural and industrial revolutions. It would certainly represent the most important global change in our lifetimes.
Cotra’s work is particularly relevant in this context as she based her forecast on the kind of historical long-run trend of training computation that we just studied. But it is worth noting that other forecasters who rely on different considerations arrive at broadly similar conclusions. As I show in my article on AI timelines, many AI experts believe that there is a real chance that human-level artificial intelligence will be developed within the next decades, and some believe that it will exist much sooner.
Building a Public Resource to Enable the Necessary Public ConversationComputers and artificial intelligence have changed our world immensely, but we are still at the early stages of this history. Because this technology feels so familiar, it is easy to forget that all of these technologies that we interact with are very recent innovations, and that most profound changes are yet to come.
Artificial intelligence has already changed what we see, what we know, and what we do. And this is despite the fact that this technology has had only a brief history.
There are no signs that these trends are hitting any limits anytime soon. To the contrary, particularly over the course of the last decade, the fundamental trends have accelerated: investments in AI technology have rapidly increased, and the doubling time of training computation has shortened to just six months.
All major technological innovations lead to a range of positive and negative consequences. This is already true of artificial intelligence. As this technology becomes more and more powerful, we should expect its impact to become greater still.
Because of the importance of AI, we should all be able to form an opinion on where this technology is heading and to understand how this development is changing our world. For this purpose, we are building a repository of AI-related metrics, which you can find on OurWorldinData.org/artificial-intelligence.
We are still in the early stages of this history and much of what will become possible is yet to come. A technological development as powerful as this should be at the center of our attention. Little might be as important for how the future of our world—and the future of our lives—will play out.
Acknowledgements: I would like to thank my colleagues Natasha Ahuja, Daniel Bachler, Julia Broden, Charlie Giattino, Bastian Herre, Edouard Mathieu, and Ike Saunders for their helpful comments to drafts of this essay and their contributions in preparing the visualizations.
This article was originally published on Our World in Data and has been republished here under a Creative Commons license. Read the original article.
Image Credit: DeepMind / Unsplash
It’s that time of year again! As 2022 comes to a close, I’ve been reflecting on the biotech and life science stories from the year that are living rent-free in my head. Here are the ones at the top of the list.
Brain Implants Had a Great RunIn a first, a paralyzed man simultaneously operated two robotic arms with his mind, allowing him to feed himself for the first time in years. (And it was cake!) Implants helped a man with locked-in syndrome—with a sharp mind but paralyzed body—translate his thoughts into sentences, opening a gateway to finally communicate with his loved ones. Memory prosthetics—a blue-sky idea to boost memory with an implant—scored their first success in people. A spinal cord stimulator, based on a new algorithm that mimics the natural electrical pulses the brain uses to control lower body movement, helped completely paralyzed people stand and walk with assistance in just one day. Within a few months, they cruised city streets on Segway-like wheels, swam, and kayaked, using an off-the-shelf tablet to control their movements.
We Know Much More About Aging—Partly Thanks to PuppiesThe Dog Aging Project is following 60,000 dogs as they age at home, tracking genetics, metabolism, and microbiome factors that accelerate (or derail) healthy aging. The goal is to gain insight into their aging process, which could also inform our own. In a first, a controlled human trial strengthened previous findings in flies, worms, and mice that cutting calories increases healthy lifespan (not the best news for indulging in that year-end feast).
Partial cellular reprogramming, a technology that dramatically reduces a cell’s age while still allowing it to retain its identity—as opposed to transforming into stem cells—garnered attention from Silicon Valley giants as a longevity breakthrough. “Aging clocks,” based on epigenetics, tag-teamed with AI to gauge a person’s biological age and associated health status with hopes of finding age-related diseases and nipping them in the bud.
Blockbuster, Jaw-Dropping, Mega-Scale StudiesIn a mind-blowing feat, one team used CRISPR to comprehensively map nearly every gene—the genotype—to its function. In another, genetic sequencing of over 12,000 tumors built a database of DNA mutations that lead to multiple types of cancers. The brain also got the big-data treatment, with a study using nearly 125,000 scans across the lifespan to chart an atlas of changes as we age.
The size of these studies isn’t the point—rather, the databases provide unprecedented views of human biology, including multiple genetic backgrounds and ethnicities. All open-access, the resources are a wealth of information for scientists to mine for individual projects. For example, can a familiar trend of different cancers be explained by specific genetic mutations? Or can we detect factors contributing to early signs of dementia or Alzheimer’s by referencing the longevity brain atlas?
Xenotransplantation Makes It to HumansThere’s a drastic shortage of available transplant organs—so much so that there’s an ethically-fraught but regulated market for organ donation.
One solution is to commandeer organs from another species—specifically, pigs. Their organs are similar in size and functionality as ours, making them potentially valuable donors. While plagued by violent immune rejections, the blue-sky idea has increasingly gained steam in the past few years.
Back in 2017, several teams found that kidneys from pigs, genetically engineered to dampen their ability to trigger immune responses in the host, sustained a rhesus macaque monkey for more than 400 days before the organ was rejected. In 2021 and early this year, kidneys from heavily genetically-edited pigs were transplanted into brain-dead humans, maintaining function for at least 50 days.
Part of the success comes from CRISPR. Pig organs can carry a family of viruses called PERVs, or porcine endogenous retrovirus, inside their genome. While benign to pigs, they’re deadly to humans. Using CRISPR, in 2015 scientists edited over 60 porcine genes responsible for immune rejection and viral infections, allowing a transplanted pig heart to survive for over half a year inside a baboon.
Earlier this year, pig-to-human heart transplant went from moonshot to reality. With a failing heart and no chance at a human heart transplant, a patient was granted compassionate use by the FDA. The pig heart, with 10 genetic edits, was designed to limit immune responses. Two weeks after implant, all looked well; the patient survived on the pig heart, and his immune system seemingly accepted the heart.
But roughly a month later, he took a turn for the worst. Two months post-transplant, he died. The transplanted heart carried a porcine virus, a common—but treatable—infection if caught early to potentially keep the heart going.
It’s a devastating loss, especially for the volunteer David Bennett Sr.’s family. The perilous trial has bioethicists wondering where to draw the line for pig-to-human transplants. Yet Bennett’s bravery provided invaluable insight into xenotransplantation in humans—such as better ways to tackle porcine viruses and keep the host’s immune system in check. With multiple companies racing to bring xenotransplants safely to market, we could see more trials in the coming year.
New Tech for Predicting and Combating Viral InfectionsYes, I know. We’re all sick from and tired of Covid-19. But with the flu, RSV, and new Covid variants surging in a triple threat, there’s an increasing need to keep ahead as these viruses evolve. One idea is to profile antibodies triggered by a virus, say, SARS-CoV-2, which causes Covid, as a therapeutic. Yet with the Omicron variant and its subvariants rampaging across the globe, antibodies rapidly lose their efficiency.
Viral mutations—be they SARS-CoV-2, HIV, or flu viruses—occur rapidly. Creating therapies based on antibodies is basically a game of whack-a-mole. Rather than chasing the next variant, what about finding antibodies that are resilient to viral changes?
Multiple teams can now analyze—or even predict—new viral variant soups. One study tracked how the Covid-19 virus evades thousands of antibodies, pinpointing mutations that help them dodge immunity. A similar approach predicted variants based on antibodies from people earlier diagnosed with Covid-19: for example, the team realized that even people who had previously recovered from an early Omicron variant could be bulldozed by the BA.5 variant that’s now responsible for most infections.
While just half a step in front of viral mutations, the team, led by Dr. Richard Cao, is on the bleeding edge of generating data for new variants and sharing it with the world. The data could help find antibody treatments that become more resilient to viral mutations. Covid-19 is just the start; a universal flu vaccine or one for HIV—both notorious for their ability to mutate—could also benefit from the insights.
Awesomely Weird StudiesA unique type of material exploiting metasurfaces opened the door to human telepathy. Freeze-dried skin cells from mice birthed healthy pups. Hundreds of thousands of isolated neurons in a dish learned to play Pong in minutes.
In synthetic biology, an artificial womb kept mouse embryos alive for over eight days, unleashing the next chapter of artificial reproduction. A little mouse called Xiao Zhu pushed the field’s boundaries, alive and thriving while harboring an artificially fused chromosome, one number short of a mouse’s usual chromosomal count.
2022 was a great year of scientific exploration. These are just the stories that resonated with me. I’m eager to see what the new year has to offer, and share with you intriguing, profound, and utterly weird new research that goes where no one has gone before.
Image Credit: Maximiliano D’Angelo and Martin Hetzer, Salk Institute
As the sense of urgency around climate change intensifies, most of the focus is on shifting energy production away from fossil fuels and electrifying transport, from cars to buses to planes. Transportation and electricity production are the top two culprits when it comes to emitting CO2 (but also two of the most necessary tools for our day-to-day lives). Third on the list and an equally complex beast is industry, and a big part of industry is concrete.
It’s been said that concrete is the most widely-used substance on Earth after water. It’s all around us, but we never really think about it. Modern society is built on it; it’s in our roads, schools, homes, offices, and more; we can’t live without it. Yet we’re going to have to start trying.
The manufacture of cement, concrete’s key ingredient, accounts for a whopping eight percent of the world’s emissions. We’re not going to stop building things; on the contrary, we’re in the midst of a major housing crisis that’s going to require a lot more building of things (and doing so cheaply). So how do we build strong, durable structures without continuing to harm the planet? What could reliably and affordably take the place, going forward, of the concrete that blankets our cities?
A startup called CarbiCrete has been developing one promising solution: carbon-negative concrete.
CarbiCrete was founded by Dr. Mehrdad Mahoutian and Chris Stern, both alumni of Montreal’s McGill University; Mahoutian started developing the company’s tech as a PhD student. Earlier this year the company secured $17.3 million (23.5 million CAD) in Series A funding.
Status-Quo ConcreteThe key ingredient in concrete is cement, a complex compound made from calcium, silicon, aluminum, iron, and other ingredients These are heated to extremely high temperatures (2,700 degrees Fahrenheit!), causing a chemical reaction where some elements burn off and the remaining ones end up as a powder. There’s a double-whammy of emissions from this process: first, coal or natural gas are burned to create the energy and heat needed to reach such high temperatures; and second, the chemical reaction of the cement compounds emits CO2.
The cement powder gets mixed with aggregate materials like sand and gravel, and when water is added another chemical reaction takes places that causes the whole mixture to harden, reaching its full strength in a little under a month.
Earth-Friendly ConcreteCarbiCrete is doing things differently in a couple ways. For starters, they’ve cut out cement altogether and replaced it with steel slag. Slag is the waste that comes from the metal-making process; once iron is extracted from iron ore to make steel, slag is what’s left over. It’s not uncommon to use slag as an aggregate in construction, most often for paving roads.
One of CarbiCrete’s CMUs. Image Credit: CarbiCreteThey mix the slag with aggregate and water, then pour the mixture into forms to make CMUs (concrete masonry units, the concrete blocks used for construction). The last step is to cure the blocks so that they harden and reach full strength. This happens in an absorption chamber into which CO2 is injected, causing yet another chemical reaction; the company’s website explains, “During the carbonation process, the CO2 is permanently captured and converted into stable calcium carbonates, filling the voids of the matrix to form a dense structure and giving the concrete its strength.” Full strength is reached in 24 hours.
What makes CarbiCrete carbon-negative instead of carbon-neutral is that the company uses CO2 gas sourced from industrial vents in its absorption chambers. So they’re not creating CO2 up front, and they’re sequestering some that’s been removed from the atmosphere.
The company says its CMUs have mechanical and durability properties equivalent to or better than cement-based CMUs, including higher compressive strength by up to 30 percent, and better freeze/thaw resistance.
Scaling UpOne potential drawback, though, is that since the CO2 absorption is a critical part of the process and must be done in a special chamber, CarbiCrete can only be used in pre-cast form; it can’t be put in a mixer truck and poured on-site at a construction location. Rather than selling CMUs, CarbiCrete licenses out its technology to concrete manufacturers, who can implement the company’s technology in precast facilities. Depending on the size of the absorption chamber, the tech could be used to make blocks, panels, beams, or really any other pre-cast product.
CarbiCrete claims that if a typical CMU-producing plant adopts its technology, the environmental impact can be significant, with 20,000 tons of CO2 abated and removed, 4,400 cubic meters of water saved, and 33,000 tons in landfill avoidance annually.
There’s no doubt CarbiCrete’s product seems like the way to go. But in addition to having to be pre-cast, it could be difficult to scale the product’s final curing process to reach the volume necessary to make a dent in traditional concrete use.
Hopefully the company has more innovations up its sleeve that can address its current limitations. Investors seem to think so; last month CarbiCrete secured a new $5 million (USD) from BDC Capital’s newly-launched Climate Tech Fund II, which the founders say they’ll use for working capital, product development, and building out business development and marketing operations.
We’re still a ways away from converting to truly sustainable building technology, but carbon-negative concrete, even on a small scale, is a step in the right direction.
Image Credit: Dan Meyers on Unsplash
Cryptocurrencies have had a calamitous year, littered with hacks, bankruptcies, and precipitously declining prices. What went wrong—and are there any bright spots to look forward to in 2023?
Crypto markets hit all-time highs in November 2021, with Bitcoin’s price peaking at $68,000, driven by excitement around NFTs, play-to-earn gaming, decentralized finance (DeFi), and the amorphous concept of Web3, a fuzzy vision of a decentralized internet running on blockchains.
While the crypto takeover of prestigious Super Bowl ad slots in early 2022 suggested the industry was on the cusp of mainstream acceptance and sustained growth, some were already pointing to warning signs that the industry’s rise might not be as inevitable as others were making it out to be.
As inflation surged at the start of the year and the Federal Reserve began hiking interest rates, proponents claimed Bitcoin could be a reliable hedge against rising prices. Goldman Sachs even labeled it “digital gold” in January, predicting it could displace the traditional investor safe haven.
But the thesis didn’t pan out, and by April, it became clear that leading cryptocurrencies were sinking alongside stocks, while gold actually went up in value. By early May, Bitcoin had lost more than half its value since its all-time high the year before.
Then in the second week of May, the industry’s first major collapse sparked a death spiral crypto has yet to recover from. The stablecoin Terra, whose price was supposed to be firmly pegged to the dollar, started dropping in value. By the end of the week, it was worth just 10 cents, and its sister coin Luna became essentially worthless.
The failure wiped roughly $45 billion off the crypto market in a matter of days. The blame lay mainly with the risky approach the founders of Terra took to maintaining its peg to the dollar. While most stablecoins back their tokens with cash reserves, Terra was relying on an arcane system of algorithms and game theory that was supposed to play off investor behavior to ensure it always traded at almost exactly one dollar.
Many had criticized the plan as unworkable in the long run, and they were proven right. People were incentivized to hold Terra by a savings scheme called Anchor that offered 20 percent returns, but people started pulling out after the organization decided to switch to a variable rate. This was followed by investors selling large amounts of Terra, which caused the house of cards to collapse.
The Terra collapse had a cascading effect on the wider crypto market. In June, the world’s largest crypto hedge fund Three Arrows Capital (3AC) announced it had taken heavy losses due to Luna’s descent. By the end of the month, it defaulted on a $670 million loan from crypto broker Voyager Digital and both companies filed for bankruptcy the following month.
Poor risk management practices and the incestuous nature of crypto trading—nearly every major crypto lender had made loans to 3AC—meant the failure of this single entity sent ripples through the entire crypto industry. The summer saw a series of crises, with crypto exchanges and lenders freezing withdrawals and companies filing for bankruptcy, most notably major crypto lender Celsius Network.
In the background, an ever-growing list of hacks on some of the industry’s biggest names were further denting investor confidence. In October, consultancy Chainalysis pointed out there had already been more than 125 hacks in 2022, racking up losses of as much as $3 billion and putting the year well on course to be the worst for crypto hacks to date.
The coup de grâce came in November when leading exchange FTX plunged from a valuation of around $32 billion to bankruptcy in just a few days. It turned out that an affiliated trading firm founded by FTX CEO Sam Bankman-Fried, had effectively been using FTX customer deposits as collateral to invest in various crypto projects. When this came to light, people rushed to withdraw their funds, leading to a run on the exchange that quickly sapped its reserves.
The failure of such a massive player in the crypto ecosystem pushed prices even lower and is driving continued concerns about “contagion” as a growing number of companies disclose their exposure to FTX. By the end of the month crypto lender BlockFi, which had been in discussions with FTX about a possible acquisition, also folded. All this has left cryptocurrencies in a tailspin at the end of 2022, with some predicting that there is further pain to come.
But amongst the wreckage of the industry, there are still a handful of bright spots.
In September, the number two cryptocurrency Ethereum carried out an ambitious update known as the Merge. The currency’s blockchain had previously relied on a security protocol called proof-of-work. Under proof-of-work people compete to solve complex mathematical puzzles in order to win the right to verify transactions in exchange for a cryptocurrency reward. The Merge switched Ethereum to an approach called proof-of-stake, in which people put up chunks of crypto as collateral in exchange for the right to verify.
The previous approach required so-called “miners” to run thousands of high-end computer processors, burning huge amounts of energy to confirm transactions. This has led to concerns around the environmental impact of cryptocurrencies, but proof-of-stake could provide a solution.
The approach is still largely unproven, leading many to highlight the potential risks of the Merge. But so far the upgrade has gone smoothly, and preliminary analysis suggests energy usage is down significantly, perhaps pointing towards a greener future for cryptocurrencies. Future changes may also allow Ethereum to run more transactions at a higher rate and lower cost. More updates are set to roll out over the next few years, beginning with the division of the Ethereum blockchain into a series of smaller databases, a process known as “sharding,” in 2023.
Among all the doom and gloom, some are also saying that this year’s crypto crash was a much needed corrective to all the hype that had built up around the industry, and could go a long way to weeding out speculators and charlatans. It’s also increased calls for regulation of the sector, which in the long run could help it become more sustainable.
Ultimately, despite the depth of the crisis, many in traditional finance think cryptocurrencies are likely to rebound in 2023, although it may be a slow and gradual recovery. Tellingly, they are predicting that projects, like Ethereum, that can be used to support practical real-world applications, rather than just financial speculation, will be the drivers of growth in crypto’s next phase.
Image Credit: Shubham Dhage / Unsplash
With the end of 2022 approaching fast, we took a look back at the stories that struck a chord with readers this year. Below, you’ll find Singularity Hub’s 10 most-read articles of 2022.
They’re a diverse bunch: from a quantum computer demonstrating quantum advantage—the ever-fluid benchmark for tasks only a quantum machine can handle—to a house 3D printed from recycled plastic water bottles. We wrote about a transistor gate the size of a carbon atom, a DeepMind AI that can write computer code like your average programmer, a gambit to drill 12 miles into Earth’s crust and liberate the nearly limitless energy below, and a spinal chord implant that helped paralyzed people walk again.
There was plenty of nourishment for curious minds this year, and alongside another frenzied news cycle, science and technology continued to move ahead. As always, thanks for reading.
Quantum Chip Takes Microseconds to Do a Task a Supercomputer Would Spend 9,000 Years On
By Shelly Fan
“Are quantum computers overhyped? A new study in Nature says no. A cleverly-designed quantum device developed by Xanadu, a company based in Toronto, Canada, obliterated conventional computers on a benchmark task that would otherwise take over 9,000 years. For the quantum chip Borealis, answers came within 36 microseconds. Xanadu’s accomplishment is the latest to demonstrate the power of quantum computing over conventional computers—a seemingly simple idea dubbed quantum advantage.”
Our Conscious Experience of the World Is But a Memory, Says New Theory
By Shelly Fan
“[According to the theory, consciousness] helps us remember the events of our lives—the whens, wheres, whats, and whos—which in turn can help us creatively and flexibly recombine them to predict or imagine alternative possibilities. It gets more mind-bending. Rather than perceiving the world in real time, we’re actually experiencing a memory of that perception. That is, our unconscious minds filter and process the world under the hood, and often make split-second decisions. When we become aware of those perceptions and decisions—that is, once they’ve risen to the level of consciousness—we’re actually experiencing ‘memories of those unconscious decisions and actions,’ the authors explained. In other words, it’s mainly the unconscious mind at the wheel.”
The World’s Biggest Cultured Meat Factory Is Under Construction in the US
By Vanessa Bates Ramirez
“Despite the fact that consumers have never tasted it and it’s only legal in Singapore, cultured meat is on a roll. Its production cost is dropping, multiple companies have entered the space, and the FDA recently granted its first approval to one of them. Last week the industry hit another milestone as Israeli company [Believer Meats] broke ground on what it says will be the biggest cultured meat plant in the world…with a production capacity of 10,000 metric tons.”
This Engineered ‘Superplant’ Cleans Indoor Air Like 30 Regular Plants
By Vanessa Bates Ramirez
“Most air purifiers are designed to remove particulate matter, like dust, dirt, smoke, or airborne bacteria. But Neo P1 was made to combat a type of pollution called volatile organic compounds. These are found in all sorts of household items, from furniture and cleaning products to paint, upholstery, and flooring. The chemicals in these items that are most harmful to human health—which are also the ones the plant was engineered to neutralize—are formaldehyde, benzene, toluene, and xylene. They can contribute to lung problems like cancer and COPD, as well as heart disease and other health issues.”
*A Spinal Cord Implant Allowed Paralyzed People to Walk in Just One Day
By Shelly Fan
“Michel Roccati never thought he’d walk again, much less swim, cycle, or paddle a kayak. A terrifying motorcycle collision in 2017 damaged his spinal cord, leaving him completely paralyzed from the waist down. Yet on a cold, snowy day last December in Lausanne, Switzerland, he took his first step outside—with the help of a walker—since his accident. His aid? A new spinal cord implant that bridges signals from the brain to his lower muscles, hopping over damaged portions to restore movement. All it took was one* day of stimulation. ‘The first few steps were incredible—a dream come true!’ he said.”
*Moore’s Law: Scientists Just Made a Graphene Transistor Gate the Width of an Atom
By Jason Dorrier
“There’s been no greater act of magic in technology than the sleight of hand performed by Moore’s Law. Electronic components that once fit in your palm have long gone atomic, vanishing from our world to take up residence in the quantum realm. But we’re now brushing the bitter limits of this trend. In a paper published in Nature* this week, scientists at Tsinghua University in Shanghai wrote that they’ve built a graphene transistor gate with a length of 0.34 nanometers—or roughly the size of a single carbon atom.”
*DeepMind’s AlphaCode Conquers Coding, Performing as Well as Humans
By Shelly Fan*
“The secret to good programming might be to ignore everything we know about writing code. At least for AI. It seems preposterous, but DeepMind’s new coding AI just trounced roughly 50 percent of human coders in a highly competitive programming competition. On the surface the tasks sound relatively simple: each coder is presented with a problem in everyday language, and the contestants need to write a program to solve the task as fast as possible—and hopefully, free of errors. But it’s a behemoth challenge for AI coders. The agents need to first understand the task—something that comes naturally to humans—and then generate code for tricky problems that challenge even the best human programmers.”
*Startup Will Drill 12 Miles Into Earth’s Crust to Tap the Boundless Energy Below
By Jason Dorrier*
“What if there were a nearly limitless source of energy available anywhere on the planet? What if the only thing preventing us from tapping said energy source was technology? And what if that tech drew on the expertise of a century-old, trillion-dollar industry, and could readily slot into much of the infrastructure already built for that industry? The answer to these questions is and has always been directly beneath our feet. The core of our planet is hotter than the surface of the sun—all we have to do is drill deep enough to liberate some of its heat. At least, that’s the dream Quaise Energy is pitching, and the startup, spun out of MIT in 2018, recently secured $40 million in new funding to go after it.”
*First Controlled Human Trial Shows Cutting Calories Improves Health, Longevity
By Shelly Fan*
“Bring up caloric restriction, or ‘CR,’ in humans at any longevity forum, and you’ll trigger a furious debate between die-hard proponents and passionate dissenters. The reason why is also simple: we only have theories, but lack sufficient data in humans. …Enter CALERIE. The Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy trial is the first controlled study of CR in the average Joe. Headed by scientists at Yale University and Pennington Biomedical Research, the trial found that cutting calories by a mere 14 percent for 2 years—about one less muffin per day—conferred multiple health benefits known to combat aging.”
*These Sleek Houses Are 3D Printed From Recycled Plastic. Prices Start at $26,900
By Vanessa Bates Ramirez*
“The US has a housing shortage problem. We also have a plastic waste problem. What if we could solve both these problems simultaneously with an unexpected two birds/one stone innovation? If you’ve ever longed to live in a small house made out of 100,000 recycled plastic water bottles, your lucky day is just around the corner.”
Image Credit: Erick Chévez / Unsplash
In 2017, the Scottish philosopher William MacAskill coined the name “longtermism” to describe the idea “that positively affecting the long-run future is a key moral priority of our time.” The label took off among like-minded philosophers and members of the “effective altruism” movement, which sets out to use evidence and reason to determine how individuals can best help the world.
This year, the notion has leapt from philosophical discussions to headlines. In August, MacAskill published a book on his ideas, accompanied by a barrage of media coverage and endorsements from the likes of Elon Musk. November saw more media attention as a company set up by Sam Bankman-Fried, a prominent financial backer of the movement, collapsed in spectacular fashion.
Critics say longtermism relies on making impossible predictions about the future, gets caught up in speculation about robot apocalypses and asteroid strikes, depends on wrongheaded moral views, and ultimately fails to give present needs the attention they deserve.
But it would be a mistake to simply dismiss longtermism. It raises thorny philosophical problems—and even if we disagree with some of the answers, we can’t ignore the questions.
Why all the Fuss?It’s hardly novel to note that modern society has a huge impact on the prospects of future generations. Environmentalists and peace activists have been making this point for a long time—and emphasizing the importance of wielding our power responsibly.
In particular, “intergenerational justice” has become a familiar phrase, most often with reference to climate change.
Seen in this light, longtermism may look like simple common sense. So why the buzz and rapid uptake of this term? Does the novelty lie simply in bold speculation about the future of technology—such as biotechnology and artificial intelligence—and its implications for humanity’s future?
For example, MacAskill acknowledges we are not doing enough about the threat of climate change, but points out other potential future sources of human misery or extinction that could be even worse. What about a tyrannical regime enabled by AI from which there is no escape? Or an engineered biological pathogen that wipes out the human species?
These are conceivable scenarios, but there is a real danger in getting carried away with sci-fi thrills. To the extent that longtermism chases headlines through rash predictions about unfamiliar future threats, the movement is wide open for criticism.
Moreover, the predictions that really matter are about whether and how we can change the probability of any given future threat. What sort of actions would best protect humankind?
Longtermism, like effective altruism more broadly, has been criticized for a bias towards philanthropic direct action—targeted, outcome-oriented projects—to save humanity from specific ills. It is quite plausible that less direct strategies, such as building solidarity and strengthening shared institutions, would be better ways to equip the world to respond to future challenges, however surprising they turn out to be.
Optimizing the FutureThere are in any case interesting and probing insights to be found in longtermism. Its novelty arguably lies not in the way it might guide our particular choices, but in how it provokes us to reckon with the reasoning behind our choices.
A core principle of effective altruism is that, regardless of how large an effort we make towards promoting the “general good”—or benefiting others from an impartial point of view —we should try to optimize: we should try to do as much good as possible with our effort. By this test, most of us may be less altruistic than we thought.
For example, say you volunteer for a local charity supporting homeless people, and you think you are doing this for the “general good.” If you would better achieve that end, however, by joining a different campaign, you are either making a strategic mistake or else your motivations are more nuanced. For better or worse, perhaps you are less impartial, and more committed to special relationships with particular local people, than you thought.
In this context, impartiality means regarding all people’s wellbeing as equally worthy of promotion. Effective altruism was initially preoccupied with what this demands in the spatial sense: equal concern for people’s wellbeing wherever they are in the world.
Longtermism extends this thinking to what impartiality demands in the temporal sense: equal concern for people’s wellbeing wherever they are in time. If we care about the wellbeing of unborn people in the distant future, we can’t outright dismiss potential far-off threats to humanity—especially since there may be truly staggering numbers of future people.
How Should We Think About Future Generations and Risky Ethical Choices?An explicit focus on the wellbeing of future people unearths difficult questions that tend to get glossed over in traditional discussions of altruism and intergenerational justice.
For instance: is a world history containing more lives of positive wellbeing, all else being equal, better? If the answer is yes, it clearly raises the stakes of preventing human extinction.
A number of philosophers insist the answer is no—more positive lives is not better. Some suggest that, once we realize this, we see that longtermism is overblown or else uninteresting.
But the implications of this moral stance are less simple and intuitive than its proponents might wish. And premature human extinction is not the only concern of longtermism.
Speculation about the future also provokes reflection on how an altruist should respond to uncertainty.
For instance, is doing something with a one percent chance of helping a trillion people in the future better than doing something that is certain to help a billion people today? (The “expectation value” of the number of people helped by the speculative action is one percent of a trillion, or 10 billion—so it might outweigh the billion people to be helped today).
For many people, this may seem like gambling with people’s lives, and not a great idea. But what about gambles with more favorable odds, and which involve only contemporaneous people?
There are important philosophical questions here about apt risk aversion when lives are at stake. And, going back a step, there are philosophical questions about the authority of any prediction: how certain can we be about whether a possible catastrophe will eventuate, given various actions we might take?
Making Philosophy Everybody’s BusinessAs we have seen, longtermist reasoning can lead to counter-intuitive places. Some critics respond by eschewing rational choice and “optimization” altogether. But where would that leave us?
The wiser response is to reflect on the combination of moral and empirical assumptions underpinning how we see a given choice. And to consider how changes to these assumptions would change the optimal choice.
Philosophers are used to dealing in extreme hypothetical scenarios. Our reactions to these can illuminate commitments that are ordinarily obscured.
The longtermism movement makes this kind of philosophical reflection everybody’s business, by tabling extreme future threats as real possibilities.
But there remains a big jump between what is possible (and provokes clearer thinking) and what is in the end pertinent to our actual choices. Even whether we should further investigate any such jump is a complex, partly empirical question.
Humanity already faces many threats that we understand quite well, like climate change and massive loss of biodiversity. And, in responding to those threats, time is not on our side.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Drew Beamer / Unsplash
As the world attempts to transition away from fossil fuels over the next several decades, critical minerals will likely be among the world’s most sought-after commodities. The US isn’t in a great position with respect to resources like cobalt, lithium, or graphite, all of which are needed for electric vehicles. China, meanwhile, controls 65 percent of the supply chains for battery-ready lithium chemicals and has 20 times more battery manufacturing capacity than the US.
That will take a while to change in any significant way, but an announcement last week from battery manufacturer Redwood Materials is a small step towards evening the scales. The company will be building what it calls a “battery materials campus” near Charleston, South Carolina that will eventually be able to power more than a million electric vehicles per year.
Redwood runs a combination recycling/manufacturing operation: the company takes in batteries (from cars, laptops, phones, tablets, and other electronics) that are at the end of their useful life, then breaks them down and extracts metals like nickel, copper, cobalt, and lithium. They then rebuild those metals into cathode and anode products, which are the fundamental components of electric vehicle batteries (and account for most of their cost).
Anode and cathode components aren’t produced anywhere in the US at present (or anywhere in North America, for that matter). According to Redwood, companies that make battery cells have to source them through a 50,000-mile global supply chain—and that’s not cheap. As a result, American battery manufacturers will spend more than $150 billion overseas on anode and cathode components by 2030.
The Inflation Reduction Act (IRA) President Biden signed into law this past August aims to change that. The act includes provisions to aid the onshoring of critical minerals mining, processing, and related manufacturing. Mining companies that produce aluminum, lithium, or graphite will qualify for a tax credit equivalent to 10 percent of the cost of production for that mineral, and consumers who buy electric vehicles get tax credits if a certain proportion of the minerals in the cars were extracted or processed in the US or free trade partner countries.
The IRA was preceded by an announcement last May of $3.16 billion in government funding for domestic battery manufacturing and supply chains for battery materials as part of the Bipartisan Infrastructure Law passed in November 2021.
There’s no shortage of incentives, then, for companies like Redwood to get cranking on those batteries. The South Carolina plant will be built on 600 acres, cost $3.5 billion, and create about 1,500 jobs. The company says the operation won’t use any fossil fuel, sourcing only clean energy, and its plant design and manufacturing process will yield an 80 percent reduction in the CO2 emissions from producing these components (as compared to the current Asia-based supply chain).
The facility is expected to supply battery materials to Ford, SK On, Toyota, Volvo, and Envision AESC plants in nearby states. Redwood plans to break ground on the project in the first quarter of 2023, have its first recycling process running by the end of next year, and eventually produce 100 GWh of cathode and anode components annually.
Image Credit: Redwood Materials
Big companies like Amazon and Walmart have recently been testing drone delivery, flying customers’ orders to them through traffic-less skies. Given the volume of packages now circulating through the mail system every day (and the Amazon delivery vehicles constantly clogging up urban streets), it’s important that companies find better ways to get goods to buyers. But while ferrying paper towels or shampoo (or the innumerable other items people buy online) into buyers’ waiting hands via drone is convenient (and, let’s be honest, pretty cool), the speed and urgency of the airborne technology is somewhat lost on these mundane tasks.
There are other uses for delivery drones that can truly make a difference between life and death. As a paper in Science Robotics explains, transporting human organs is at the top of that list.
Last year, a team of Canadian researchers transported a donor lung via remote-piloted drone in downtown Toronto. The flight happened in September of 2021, but their paper was just published today (peer review can be a slow process). The lung went from Toronto Western Hospital to Toronto General Hospital; the two facilities are just under two kilometers (1.24 miles) apart. That’s not far, but as you know if you’ve ever tried to hurriedly drive through a dense city center—especially at busy times of day like rush hour—driving a short distance can take longer than walking. There are buses, pedestrians, delivery vehicles, cyclists, and other frustrated drivers to contend with.
Even an emergency vehicle with sirens blaring may not cover a short distance fast enough; when it comes to transplanting an organ, time is of the essence. “The moment an organ is removed from the human body, it begins to rapidly deteriorate,” the authors wrote. “Failure to deliver and transplant an organ in a timely manner can result in a missed opportunity to save a life.”
The delivery drone with the lung transport box, pictured during a test flight. Image Credit: University Health Network/Unither Bioelectronics IncThe drone they used was an M600 Pro, made by a Chinese company called DJI (this model is no longer in production). The team modified it to remove its original landing gear and payload rack so that the team could install a specially-designed lung transport box. They also modified the drone’s electronic systems for better connectivity to make sure it wouldn’t get thrown off track by interfering signals, and added safety features like a parachute recovery system, cameras, lights, and GPS trackers. Including the lung transport box and the lungs themselves, the drone was designed to weigh a maximum of 25 kg (55 pounds).
Flight time to go from one hospital to the other was about five minutes. Upon arrival, the lung was transplanted into a 63-year-old patient with idiopathic pulmonary fibrosis (a chronic condition where the lungs become scarred and it’s hard to breathe). The patient survived the transplant and recovered normally. Though everything went smoothly, this wasn’t left to chance; the team performed more than 400 test flights of the route starting in 2019.
It’s not the first time drones are used to transport organs or other crucial medical supplies, though. Donor kidneys were flown across Baltimore via drone in 2019 and over the Las Vegas desert in 2020. Various organs were recently moved between Texas and Oklahoma in an unmanned Cessna (though these were donated for research rather than being implanted in recipients). Matternet is using drones to deliver diagnostic samples in Switzerland, and in the UK a 165-mile drone superhighway is being built, and Britain’s National Health Service is trialing drone delivery for medication.
The researchers from the Toronto flight believe their trial is just the beginning of a method that will become commonplace. Regulations and infrastructure aren’t where they need to be yet, so it could be a while, but there have been enough successful proofs of concept by now that we know drone transport is a viable option. “Even for short trips between nearby hospitals, drones offer a reliable transportation method that overcomes typical city congestion,” the authors wrote. “Thus, it is likely that all donor organs will be delivered by drone in the future, irrespective of distance from the transplant hospital.”
Image Credit: University Health Network/Unither Bioelectronics Inc
Exercise more. That’s usually my (and many other peoples’) top New Year’s resolution.
But it’s drizzly with bone-chilling winds howling outside. And I’m wrapped in a fuzzy blanket on the couch with a cup of hot cocoa and the latest Netflix show. My resolve quickly dwindles.
According to a new study in Nature, I could get a motivational boost from a surprising source: my gut microbes. In a tour-de-force study, a team from the University of Pennsylvania found that changing the millions of gut bugs in your microbiome can peel you off the couch and motivate exercise—at least, if you’re a mouse.
Alone, the results sound like pseudoscientific nonsense. But the study dug deep: the team honed in on how and why gut microbes encourage mice to run and keep running. The crux is a chemical produced by the microbiome that sends a signal from the gut to the brain, triggering a deluge of dopamine to be released into the ventral striatum—the brain’s “motivation center”—in turn sparking a desire to work out.
I’ve said this often: mice are not men. But the study propels the relatively new field of gut-brain interaction into new territory. Can the gut directly influence the brain’s motivations and desires? By hunting down the molecules in the gut that spur the brain to want to be physically active, the study gave us a first answer: yes.
“If these findings are relevant to humans, they raise the question of whether targeting gut bacteria could improve the mental processes associated with the decision to exercise across individuals, whether elite athlete or not,” said neuroscientists Drs. Gulistan Agirman and Elaine Y. Hsiao at the University of California, Los Angeles, who weren’t involved in the study.
The Exercise DilemmaWe all know working out is good for us. Thousands of studies have shown that regular exercise helps with everything from weight control to decreasing the risk of heart disease and upping mental health and mood, and even battling aging and dementia.
So why is it that despite knowing the benefits, it’s still so hard to get motivated?
Mindset—that is, your psychology—was originally considered the main culprit, explained Agirman and Hsiao. But the new study suggests that the gut microbiome could also give you a hefty motivational boost.
The gut-brain connection is one of the most influential discoveries of the past decade. The brain doesn’t exist in a vacuum. Rather, molecules and hormones from the body can significantly impact its function. Chemicals released from the liver, for example, bolster memory function in aging mice after exercise, birthing more new neurons in the dentate gyrus—the “nursery” in the hippocampus, a region critical for memory.
A major source of these systemic molecules is the gut microbiome. Its symbiotic microbes thrive inside our guts, helping digest nutrients and support metabolism. A decade ago, neuroscientists surprisingly found that they also impact the brain. Wiping the bacteria out with antibiotics, for example, increases depressive symptoms in mice. Subsequent studies found that certain microbes excrete chemicals as they digest food, which activates the vagus nerve, a main signaling highway that goes from gut to brain.
They also help the body respond to exercise. Specific bacterial groups in the gut have emerged “as key regulators of exercise performance,” said Agirman and Hsiao. Usually this occurs through microbe-excreted chemicals to generate energy, or those that help eliminate molecules that lead to physical exhaustion, such as lactate. The new study wondered: can the gut microbiome directly shape our desire to exercise by impacting brain function?
Honing InMice generally love to run. But like humans, depending on their genetics and physiology, they have different propensities—some like to run fast, others long, and some not at all.
To understand why, the team started with nearly 200 mice specifically bred to encourage a diverse genetic background and gathered their bodily data. These included genetic sequencing, metabolic profiling, and sequencing the RNA in their stool—an established method to gauge a gut microbiome profile.
Overall, the team collected over 10,500 data points for each mouse and roughly two million in total.
The mice next ran on a treadmill or a running wheel. The latter is a treat, as (anyone with a hamster or other rodent pet knows) they will happily hop on and run considerable distances every night—some more than nine miles per day.
But there were also couch potatoes. These fluffballs were happy to chill out, barely touching the wheel during a two-day test period.
Surprisingly, the mice’s genetic signatures had very little impact on their motivation to run. Widening their hunt, the team turned to machine learning to analyze molecules in their blood, their metabolism, and their gut microbiomes to see if individual differences matched up with running performance.
The answer raised eyebrows: the only factor that predicted a mouse’s willingness to run was its gut bacteria. It suggested that “gut bacteria drive exercise performance,” said Agirman and Hsiao.
But correlation isn’t causation. In the next tests, the team wiped out the microbiome of one group of athletic mice using antibiotics, turning them into couch potatoes. In contrast, mice raised inside a germ-free bubble—who naturally lack gut bacteria—transformed into marathon runners when transplanted with gut bugs from their naturally vigorous peers.
A Brainy LinkWhy does the gut microbiome have anything to do with motivation?
The answer seems to be dopamine. Often dubbed the “pleasure chemical,” dopamine has various roles in the brain, including flagging errors that don’t fit predictions and directing smooth movements. But its best-known role is to combine movement and reward, which happens in a deep brain nugget called the ventral striatum, a part of the brain’s “reward center.”
Digging into the mice’s microbiome data, the team found that athletic mice had a population of gut bugs particularly good at secreting fatty acid amides (FAA). Acting as “keys,” these chemicals then activated a receptor “lock”—the CB1 receptor that dots the outside of a specific type of sensory neuron inside the gut (yes, the gut has neurons, and yes, the CB1 receptor is also the target of marijuana’s main chemical components). These specialized neurons then send electrical signals directly through the spinal cord into the brain’s striatum, flooding it with a hit of dopamine.
In contrast, mice without gut bacteria didn’t have this dopamine spike. A bit more sleuthing found that their brains had a high level of an enzyme that rapidly chews up dopamine, essentially killing off their “runner’s high.” However, giving them a dose of FAA as a dietary supplement or transferring gut bacteria that produces FAA into their guts upped their running games.
The authors “have demonstrated that the circuits involved in the motivation needed to sustain physical activity in mice are modulated by gut microbes,” said Agirman and Hsiao.
New Year’s ResolutionTo be clear, these results are in mice. We don’t know if they hold up in humans. But they do offer new clues to long-lingering questions, such as why runner’s high feels great even when you’re in physical pain. I wouldn’t be surprised if the gut bug chemicals are bottled up into pre-workout motivation elixirs—though again, buyer beware!
Zooming out, the study adds to a growing pantheon of evidence that our microbiomes directly impact the brain’s function, especially for mood and motivation. But our gut does not control our desires.
“Although tempting to consider the human implications of this research, gauging the practical relevance of these findings will require extensive further assessment,” said Agirman and Hsiao. “A variety of other factors influence motivational states in people, requiring a range of strategies to strengthen motivational and reward circuits in unfavorable environments.”
Image Credit: Wokandapix from Pixabay
A rapid transition to renewable power is essential to avoid the worst effects of climate change, but governments have been lukewarm in their commitment. Energy security concerns spurred by Russia’s invasion of Ukraine seem to be sharpening minds, though, according to a new report.
In its latest assessment of the state of renewable power, the International Energy Agency (IEA) says that the global energy crisis the conflict has caused is driving a significant acceleration in the roll-out of green energy projects as governments try to reduce their reliance on imported fossil fuels.
The upshot is that global capacity is expected to grow by as much as 2,400 gigawatts (GW) between now and 2027. That’s equal to China’s total power capacity today, and more renewable power than the world has installed in the previous 20 years.
It’s also about 30 percent higher than the agency was predicting last year, making this the largest-ever upward revision of its renewable energy forecasts. The report predicts that renewables will make up 90 percent of all new power projects over the next half-decade, and by 2025 solar is likely to overtake coal as the world’s single biggest source of power.
“Renewables were already expanding quickly, but the global energy crisis has kicked them into an extraordinary new phase of even faster growth as countries seek to capitalize on their energy security benefits,” IEA executive director Fatih Birol said in a statement. “This is a clear example of how the current energy crisis can be a historic turning point towards a cleaner and more secure energy system.”
Nowhere has the energy crisis spurred a bigger reaction than in Europe. Much of the continent has long been reliant on Russian fossil fuels, with the EU importing nearly half its natural gas from the country. Given the growing rifts with its neighbor, the bloc is keen to rectify this situation.
In May, the European Commission released its REPowerEU plan in response to the Russian invasion, which outlines how the bloc plans to reduce its energy use, boost renewables, and diversify the sources of its fossil fuel supplies. This includes commitments to end reliance on Russian fossil fuels by 2027 and boost renewables’ share of the energy mix to 45 percent.
Combined with existing climate ambitions, the IEA report predicts that this will see Europe add twice as much renewable energy capacity by 2027 as it did in the previous five years. This will be led by Germany and Spain, which have recently enacted a host of renewables-friendly policies designed to spur growth.
The acceleration isn’t driven by Europe alone, though. China’s 14th five-year plan, which was officially endorsed in March 2021, will see the country contribute about half of all new renewable power capacity over the next five years. And at the same time, the Inflation Reduction Act passed by the Biden administration earlier this year is likely to drive a significant increase in renewable capacity in the US.
Most of this expansion is going to come from solar and wind energy, according to the report, which are now the cheapest options for new electricity generation in most countries. But there is also likely to be significant growth in both hydrogen and biofuels, which may be better suited to helping decarbonize industry and transport.
The agency predicts the amount of renewable energy dedicated to producing green hydrogen will rise a 100-fold increase to 50GW, thanks to new policies and targets introduced in more than 25 countries.
Biofuel demand is expected to jump by 27 percent over the next 5 years, with up to a third of that due to be produced from waste rather than purpose-grown fuel crops. That is likely to exhaust supplies of easily-available waste, though, so the amount of vegetable oil dedicated to producing biofuel is likely to jump from 17 percent to 23 percent, raising questions about the fuel’s sustainability.
Despite all this progress, though, our chances of avoiding the worst impacts of climate change still aren’t good. The report outlines an accelerated case in which renewables growth is 25 percent higher than the baseline they identify, which could be possible if the world’s governments make major regulatory reforms and ensure better financing options for renewables projects.
But even in this optimistic scenario, which could lead to zero net emissions by 2050, the report says we would only have an even chance of limiting warming to 1.5 °C. So while the news is positive, we’re still a long way from being able to rest on our laurels.
Image Credit: seagul from Pixabay
Artificial intelligence that surpasses our own intelligence sounds like the stuff from science fiction books or films. What do experts in the field of AI research think about such scenarios? Do they dismiss these ideas as fantasy, or are they taking such prospects seriously?
A human-level AI would be a machine, or a network of machines, capable of carrying out the same range of tasks that we humans are capable of. It would be a machine that is “able to learn to do anything that a human can do,” as Norvig and Russell put it in their textbook on AI.1
It would be able to choose actions that allow the machine to achieve its goals and then carry out those actions. It would be able to do the work of a translator, a doctor, an illustrator, a teacher, a therapist, a driver, or the work of an investor.
In recent years, several research teams contacted AI experts and asked them about their expectations for the future of machine intelligence. Such expert surveys are one of the pieces of information that we can rely on to form an idea of what the future of AI might look like.
The chart shows the answers of 352 experts. This is from the most recent study by Katja Grace and her colleagues, conducted in the summer of 2022.2
Experts were asked when they believe there is a 50% chance that human-level AI exists.3 Human-level AI was defined as unaided machines being able to accomplish every task better and more cheaply than human workers. More information about the study can be found in the fold-out box at the end of the text on this page.4
Each vertical line in this chart represents the answer of one expert. The fact that there are such large differences in answers makes it clear that experts do not agree on how long it will take until such a system might be developed. A few believe that this level of technology will never be developed. Some think that it’s possible, but it will take a long time. And many believe that it will be developed within the next few decades.
As highlighted in the annotations, half of the experts gave a date before 2061, and 90% gave a date within the next 100 years.
Other surveys of AI experts come to similar conclusions. In the following visualization, I have added the timelines from two earlier surveys conducted in 2018 and 2019. It is helpful to look at different surveys, as they differ in how they asked the question and how they defined human-level AI. You can find more details about these studies at the end of this text.
In all three surveys, we see a large disagreement between experts and they also express large uncertainties about their own individual forecasts.5
What Should We Make of the Timelines of AI Experts?Expert surveys are one piece of information to consider when we think about the future of AI, but we should not overstate the results of these surveys. Experts in a particular technology are not necessarily experts in making predictions about the future of that technology.
Experts in many fields do not have a good track record in making forecasts about their own field, as researchers including Barbara Mellers, Phil Tetlock, and others have shown.6 The history of flight includes a striking example of such failure. Wilbur Wright is quoted as saying, “I confess that in 1901, I said to my brother Orville that man would not fly for 50 years.” Two years later, ‘man’ was not only flying, but it was these very men who achieved the feat.7
Additionally these studies often find large ‘framing effects’, two logically identical questions get answered in very different ways depending on how exactly the questions are worded.8
What I do take away from these surveys however, is that the majority of AI experts take the prospect of very powerful AI technology seriously. It is not the case that AI researchers dismiss extremely powerful AI as mere fantasy.
The huge majority thinks that in the coming decades there is an even chance that we will see AI technology which will have a transformative impact on our world. While some have long timelines, many think it is possible that we have very little time before these technologies arrive. Across the three surveys more than half think that there is a 50% chance that a human-level AI would be developed before some point in the 2060s, a time well within the lifetime of today’s young people.
The Forecast of the Metaculus CommunityIn the big visualization on AI timelines below, I have included the forecast by the Metaculus forecaster community.
The forecasters on the online platform Metaculus.com are not experts in AI but people who dedicate their energy to making good forecasts. Research on forecasting has documented that groups of people can assign surprisingly accurate probabilities to future events when given the right incentives and good feedback.9 To receive this feedback, the online community at Metaculus tracks how well they perform in their forecasts.
What does this group of forecasters expect for the future of AI?
At the time of writing, in November 2022, the forecasters believe that there is a 50/50-chance for an ‘Artificial General Intelligence’ to be ‘devised, tested, and publicly announced’ by the year 2040, less than 20 years from now.
On their page about this specific question, you can find the precise definition of the AI system in question, how the timeline of their forecasts has changed, and the arguments of individual forecasters for how they arrived at their predictions.10
The timelines of the Metaculus community have become much shorter recently. The expected timelines have shortened by about a decade in the spring of 2022, when several impressive AI breakthroughs happened faster than many had anticipated.11
The Forecast by Ajeya CotraThe last shown forecast stems from the research by Ajeya Cotra, who works for the nonprofit Open Philanthropy.12 In 2020 she published a detailed and influential study asking when the world will see transformative AI. Her timeline is not based on surveys, but on the study of long-term trends in the computation used to train AI systems. I present and discuss the long-run trends in training computation in this companion article.
Cotra estimated that there is a 50% chance that a transformative AI system will become possible and affordable by the year 2050. This is her central estimate in her “median scenario.” Cotra emphasizes that there are substantial uncertainties around this median scenario, and also explored two other, more extreme, scenarios. The timelines for these two scenarios—her “most aggressive plausible” scenario and her “most conservative plausible” scenario—are also shown in the visualization. The span from 2040 to 2090 in Cotra’s “plausible” forecasts highlights that she believes that the uncertainty is large.
The visualization also shows that Cotra updated her forecast two years after its initial publication. In 2022 Cotra published an update in which she shortened her median timeline by a full ten years.13
It is important to note that the definitions of the AI systems in question differ very much across these various studies. For example, the system that Cotra speaks about would have a much more transformative impact on the world than the system that the Metaculus forecasters focus on. More details can be found in the appendix and within the respective studies.
What Can We Learn From the Forecasts?The visualization shows the forecasts of 1128 people—812 individual AI experts, the aggregated estimates of 315 forecasters from the Metaculus platform, and the findings of the detailed study by Ajeya Cotra.
There are two big takeaways from these forecasts on AI timelines:
- There is no consensus, and the uncertainty is high. There is huge disagreement between experts about when human-level AI will be developed. Some believe that it is decades away, while others think it is probable that such systems will be developed within the next few years or months. There is not just disagreement between experts; individual experts also emphasize the large uncertainty around their own individual estimate. As always when the uncertainty is high, it is important to stress that it cuts both ways. It might be very long until we see human-level AI, but it also means that we might have little time to prepare.
- At the same time, there is large agreement in the overall picture. The timelines of many experts are shorter than a century, and many have timelines that are substantially shorter than that. The majority of those who study this question believe that there is a 50% chance that transformative AI systems will be developed within the next 50 years. In this case it would plausibly be the biggest transformation in the lifetime of our children, or even in our own lifetime.
The public discourse and the decision-making at major institutions have not caught up with these prospects. In discussions on the future of our world—from the future of our climate, to the future of our economies, to the future of our political institutions—the prospect of transformative AI is rarely central to the conversation. Often it is not mentioned at all, not even in a footnote.
We seem to be in a situation where most people hardly think about the future of artificial intelligence, while the few who dedicate their attention to it find it plausible that one of the biggest transformations in humanity’s history is likely to happen within our lifetimes.
Acknowledgements: I would like to thank my colleagues Natasha Ahuja, Daniel Bachler, Bastian Herre, Edouard Mathieu, Esteban Ortiz-Ospina and Hannah Ritchie for their helpful comments to drafts of this essay.
And I would like to thank my colleague Charlie Giattino who calculated the timelines for individual experts based on the data from the three survey studies and supported the work on this essay. Charlie is also one of the authors of the cited study by Zhang et al. on timelines of AI experts.
Image Credit: DeepMind / Unsplash
ARTIFICIAL INTELLIGENCEGenerative AI Is Changing Everything. But What’s Left When the Hype Is Gone?
Will Douglas Heaven | MIT Technology Review“The exciting truth is, we don’t really know. For while creative industries—from entertainment media to fashion, architecture, marketing, and more—will feel the impact first, this tech will give creative superpowers to everybody. In the longer term, it could be used to generate designs for almost anything, from new types of drugs to clothes and buildings. The generative revolution has begun.”
BIOTECHNew ‘Cellular Glue’ Concept Could Heal Wounds, Regrow Nerves
Monisha Ravisetti | CNET“Researchers from the University of California, San Francisco announced a fascinating innovation on Monday. They call it ‘cellular glue’ and say it could one day open doors to massive medical achievements, like building organs in a lab for transplantation and reconstructing nerves that’ve been damaged beyond the reach of standard surgical repair.”
ARTIFICIAL INTELLIGENCEThe Viral AI Avatar App Lensa Undressed Me—Without My Consent
Melissa Heikkilä | MIT Technology Review“Lensa generates its avatars using Stable Diffusion, an open-source AI model that generates images based on text prompts. Stable Diffusion is built using LAION-5B, a massive open-source data set that has been compiled by scraping images off the internet. And because the internet is overflowing with images of naked or barely dressed women, and pictures reflecting sexist, racist stereotypes, the data set is also skewed toward these kinds of images.”
SPACEScientists May Have Found the First Water Worlds
John Timmer | Ars Technica“…continued observation has produced data that indicates the planets are much less dense than we originally thought. And the only realistic way to get the sort of densities they now seem to have is for a substantial amount of their volume to be occupied by water or a similar fluid. We do have bodies like this in our Solar System—most notably the moon Europa, which has a rocky core surrounded by a watery shell capped by ice. But these new planets are much closer to their host star, which means their surfaces are probably a blurry boundary between a vast ocean and a steam-filled atmosphere.”
TRANSPORTATIONThe Tech Is Finally Good Enough for an Airship Revival
Michael Koziol | IEEE Spectrum“At Moffett Field in Mountain View, Calif., Lighter Than Air (LTA) Research is floating a new approach to a technology that saw its rise and fall a century ago: airships. Although airships have long since been supplanted by planes, LTA, which was founded in 2015 by CEO Alan Weston, believes that through a combination of new materials, better construction techniques, and technological advancements, airships are poised to—not reclaim the skies, certainly—but find a new niche.”
ENERGYThe Real Fusion Energy Breakthrough Is Still Decades Away
Gregory Barber | Wired“Today, the NIF researchers said they got as much energy out as their laser fired at the experiment—a massive, long-awaited achievement. But the problem is that the energy in those lasers represents a tiny fraction of the total power involved in firing up the lasers. By that measure, NIF is getting way less than it’s putting in. ‘That type of breakeven is way, way, way, way down the road,’ Cappelli says. ‘That’s decades down the road. Maybe even a half-century down the road.’i”
TECHAI-Generated Fake Faces Have Become a Hallmark of Online Influence Operations
Shannon Bond | NPR“Facebook parent company Meta says more than two-thirds of the influence operations it found and took down this year used profile pictures that were generated by a computer. As the artificial intelligence behind these fakes has become more widely available and better at creating life-like faces, bad actors are adapting them for their attempts to manipulate social media networks.”
ETHICSWhat Does It Mean to Align AI With Human Values?
Melanie Mitchell | Quanta“We humans are prone to giving machines ambiguous or mistaken instructions, and we want them to do what we mean, not necessarily what we say. …To solve this problem, [AI researchers] believe, we must find ways to align AI systems with human preferences, goals and values. …But without a better understanding of what intelligence is and how separable it is from other aspects of our lives, we cannot even define the problem, much less find a solution. Properly defining and solving the alignment problem won’t be easy; it will require us to develop a broad, scientifically based theory of intelligence.”
Image Credit: Clark Van Der Beken / Unsplash
American scientists have announced what they have called a major breakthrough in a long-elusive goal of creating energy from nuclear fusion.
The US Department of Energy said on Dec. 13, 2022, that for the first time—and after several decades of trying—scientists have managed to get more energy out of the process than they had to put in.
But just how significant is the development? And how far off is the long-sought dream of fusion providing abundant, clean energy? Carolyn Kuranz, an associate professor of nuclear engineering at the University of Michigan who has worked at the facility that just broke the fusion record, helps explain this new result.
What Happened in the Fusion Chamber?Fusion is a nuclear reaction that combines two atoms to create one or more new atoms with slightly less total mass. The difference in mass is released as energy, as described by Einstein’s famous equation, E = mc2 , where energy equals mass times the speed of light squared. Since the speed of light is enormous, converting just a tiny amount of mass into energy—like what happens in fusion—produces a similarly enormous amount of energy.
Researchers at the US government’s National Ignition Facility in California have demonstrated, for the first time, what is known as “fusion ignition.” Ignition is when a fusion reaction produces more energy than is being put into the reaction from an outside source and becomes self-sustaining.
The fuel is held in a tiny canister designed to keep the reaction as free from contaminants as possible. Credit: U.S. Department of Energy/Lawrence Livermore National LaboratoryThe technique used at the National Ignition Facility involved shooting 192 lasers at a 0.04 inch (1 mm) pellet of fuel made of deuterium and tritium—two versions of the element hydrogen with extra neutrons—placed in a gold canister. When the lasers hit the canister, they produce X-rays that heat and compress the fuel pellet to about 20 times the density of lead and to more than 5 million degrees Fahrenheit (3 million Celsius)—about 100 times hotter than the surface of the sun. If you can maintain these conditions for a long enough time, the fuel will fuse and release energy.
The fuel and canister get vaporized within a few billionths of a second during the experiment. Researchers then hope their equipment survived the heat and accurately measured the energy released by the fusion reaction.
So What Did They Accomplish?To assess the success of a fusion experiment, physicists look at the ratio between the energy released from the process of fusion and the amount of energy within the lasers. This ratio is called gain.
Anything above a gain of one means that the fusion process released more energy than the lasers delivered.
On Dec. 5, 2022, the National Ignition Facility shot a pellet of fuel with two million joules of laser energy—about the amount of power it takes to run a hair dryer for 15 minutes—all contained within a few billionths of a second. This triggered a fusion reaction that released three million joules. That is a gain of about 1.5, smashing the previous record of a gain of 0.7 achieved by the facility in August 2021.
How Big a Deal Is This Result?Fusion energy has been the “holy grail” of energy production for nearly half a century. While a gain of 1.5 is, I believe, a truly historic scientific breakthrough, there is still a long way to go before fusion is a viable energy source.
While the laser energy of 2 million joules was less than the fusion yield of 3 million joules, it took the facility nearly 300 million joules to produce the lasers used in this experiment. This result has shown that fusion ignition is possible, but it will take a lot of work to improve the efficiency to the point where fusion can provide a net positive energy return when taking into consideration the entire end-to-end system, not just a single interaction between the lasers and the fuel.
Machinery used to create the powerful lasers, like these pre-amplifiers, currently requires a lot more energy than the lasers themselves produce. Credit: Lawrence Livermore National Laboratory, CC BY-SAWhat Needs to Be Improved?There are a number of pieces of the fusion puzzle that scientists have been steadily improving for decades to produce this result, and further work can make this process more efficient.
First, lasers were only invented in 1960. When the US government completed construction of the National Ignition Facility in 2009, it was the most powerful laser facility in the world, able to deliver one million joules of energy to a target. The two million joules it produces today is 50 times more energetic than the next most powerful laser on Earth. More powerful lasers and less energy-intensive ways to produce those powerful lasers could greatly improve the overall efficiency of the system.
Fusion conditions are very challenging to sustain, and any small imperfection in the capsule or fuel can increase the energy requirement and decrease efficiency. Scientists have made a lot of progress to more efficiently transfer energy from the laser to the canister and the X-ray radiation from the canister to the fuel capsule, but currently only about 10 to 30 percent of the total laser energy is transferred to the canister and to the fuel.
Finally, while one part of the fuel, deuterium, is naturally abundant in sea water, tritium is much rarer. Fusion itself actually produces tritium, so researchers are hoping to develop ways of harvesting this tritium directly. In the meantime, there are other methods available to produce the needed fuel.
These and other scientific, technological, and engineering hurdles will need to be overcome before fusion will produce electricity for your home. Work will also need to be done to bring the cost of a fusion power plant well down from the $3.5 billion of the National Ignition Facility. These steps will require significant investment from both the federal government and private industry.
It’s worth noting that there is a global race around fusion, with many other labs around the world pursuing different techniques. But with the new result from the National Ignition Facility, the world has, for the first time, seen evidence that the dream of fusion is achievable.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: U.S. Department of Energy/Lawrence Livermore National Laboratory
3D printing is taking off as a viable construction technology. The first several homes and buildings were all made of some sort of cement mixture (the exact composition of which varies by company), but now the list of materials that can be used as printer “ink” is growing. There’s clay, recycled plastic, regolith from the moon (this one hasn’t actually been used yet, but NASA’s working on it), and most recently, wood.
How do you 3D print a house out of wood if wood tends to come in plank or beam form (for building purposes, that is)? Well, this wood has been ground up into sawdust then mixed with reinforcing and binding agents to form a composite material that’s squishy enough to push through the hose of a 3D printer.
The material was used to build a 600-square-foot prototype house that’s now sitting on the University of Maine’s Orono campus. Called BioHome3D, the project was spearheaded by the University of Maine’s Advanced Structures and Composites Center (ASCC), funded by the US Department of Energy’s Hub and Spoke program, and included MaineHousing and the Maine Technology Institute as partners.
BioHome3D’s living room space. Image Credit: MJ Gautrau/ASCCUnlike many 3D printed homes, which are printed on-site at their final location, BioHome3D was printed off-site in four separate modules, then moved to the university campus and assembled in half a day. Another feature that sets it apart from it predecessors is that all of it was 3D printed (well, except the windows and door).
“Unlike the existing technologies, the entire BioHome3D was printed, including the floors, walls, and roof,” said Habib Dagher, executive director of ASCC. “The biomaterials used are 100 percent recyclable, so our great-grandchildren can fully recycle BioHome3D.”
While the thought of a house being recyclable is reassuring from an eco-friendliness perspective, it’s less reassuring to wonder about its strength and durability. But Dagher and his team at ASCC have been researching engineered biomaterials for 20 years, and their work wasn’t for naught.
“The house that we built meets all building requirements, whether it’s structural, or fire or toxicity,” Dagher said. “These materials are new, but…we’ve learned a lot about what they can do and can’t do.”
They’ll learn even more over the next few years; the prototype house is decked out with sensors to monitor thermal, environmental, and structural data. How might the house hold up through a cold, snowy Maine winter? Conversely, how transferable is it to other climates? The ASCC team says they’ve sent samples of the printing material as far as Brazil, where its resistance to humidity will be tested.
In terms of the house itself, BioHome3D is much like any other small home or apartment; it has an open-concept kitchen, living, and dining area with grooved wooden walls, a bedroom that can double as an office, and a tiled bathroom. Once its modules were put together, it only took two hours and one electrician to get the power up and running.
BioHome3D’s bedroom space. Image Credit: MJ Gautrau/ASCCDagher and his team chose sawdust as the basis for their printing material because it’s organic, but more importantly, because their state has a surplus of it. Maine is heavily wooded and historically had a lot of sawmills and paper mills, but the paper mills have come upon hard times as paper documents transition to digital-only, and cheaper paper from other countries undercuts the remaining market.
The sawmills are still around, but they can no longer pass on as much of their residual sawdust and other byproducts to be made into paper. So why not make it into houses instead?
“In our region, there’s an estimated 1,000 tons of biomass residuals every year that’s being generated right now,” Dagher said. “We asked ourselves, could we print a home with that material?”
His team hopes to eventually build a manufacturing plant to produce many more BioHome3Ds, with the goal of churning out a whole house in just two days. They won’t limit themselves to their home state, either. “There’s a lot of potential not only to solve a crisis in Maine, but to assist in a solution to the housing crisis nationally as well,” Dagher said.
Image Credit: MJ Gautrau/University of Maine Advanced Structures and Composites Center
Despite the fact that consumers have never tasted it and it’s only legal in Singapore, cultured meat is on a roll. Its production cost is dropping, multiple companies have entered the space, and the FDA recently granted its first approval to one of them. Last week the industry hit another milestone as an Israeli company broke ground on what it says will be the biggest cultured meat plant in the world.
The company was founded under the name Future Meat Technologies in 2018, but rebranded to Believer Meats last month. In 2021 they opened a facility to produce lab-grown meat at scale in Israel, and were aiming to secure FDA approval and start offering their products in US restaurants by the end of this year. That doesn’t seem to have happened, as the first FDA approval went to competitor Upside Foods.
But true to its name, Believer Meats hasn’t been deterred by this slower-than-anticipated series of events. Last week the company started construction of a 200,000-square-foot factory in Wilson, North Carolina, about 45 miles due east of Raleigh. In a press release the company stated, somewhat perplexingly, that it chose this location partly because of its “success in integrating technology-driven solutions to improve the lives of residents.”
With a production capacity of 10,000 metric tons, Believer says the facility will be the biggest of its type in the world. They’re putting $123 million into the plant, and say it will create more than 100 new jobs over the next three years. This huge investment seems like a bit of a leap of faith considering the company doesn’t have regulatory approval to produce and sell cultivated meat anywhere, including in the US; but co-founder Yaakov Nahmias says they’ve been working with the FDA towards gaining approval for years.
And Believer isn’t the only company taking such a leap of faith. Its competitor Good Meat is finalizing the location of a similarly large cultured meat plant in the US (which it also claims will be the biggest in the world), and aiming to start production by late 2024.
Despite the lagging approvals, cultured meat is starting to look more promising as a viable alternative to factory farming. That’s not to say the former will replace the latter anytime this decade (nor probably the next), but it’s getting off the ground.
Another encouraging aspect of the technology is the variety of meats will seemingly be available once this stuff hits the market. The industry started with ground beef (food critics sampled the world’s first lab-grown burger in 2013, and it came at an estimated cost of $330,000), and has since expanded to chicken, pork, salmon, and steak.
Given the way cultured meat is made, there’s no reason why the above list can’t expand even more. Cells are extracted from an animal’s tissue (in a process that doesn’t harm the animal at all) and mixed with a cocktail of nutrients, oxygen, and moisture. Inside large bioreactors, the mixture is kept at the same temperature cells would be at in an animal’s body. The cells divide, multiply, and mature, with any waste products being removed to keep the environment pure.
You can just as easily take cells from a pig as from a cow, chicken, turkey, lamb, or fish (etc, etc). But growing the cells is the (relatively) easy part; the bioreactors don’t pop out ready-to-eat chicken breasts or racks of lamb. Replicating meat’s structure—that is, the tendons, muscle, fat, bone, and connective tissue that comes with it—is a complex process, and crucial to giving whole cuts of meat their distinctive texture and flavor. After the cells are “harvested” from the bioreactors they grew in, they need to be refined and shaped into a final product, which could involve extrusion cooking, molding, and even 3D printing.
Believer Meats hasn’t disclosed an anticipated completion date for the Wilson facility. But Nahmias, for his part, is optimistic. “While cultivated meat has its skeptics, we believe in demystifying the technology and critically demonstrating that there is a better way to produce meat through open science and innovation,” he said. “As the demand for meat continues to grow in coming decades, the current conventional meat industry won’t be able to meet the supply needed. That’s why we believe cultivated meat is needed to secure healthy, sustainable, and affordable nutrition for coming generations.”
Image Credit: Believer Meats
The secret to good programming might be to ignore everything we know about writing code. At least for AI.
It seems preposterous, but DeepMind’s new coding AI just trounced roughly 50 percent of human coders in a highly competitive programming competition. On the surface the tasks sound relatively simple: each coder is presented with a problem in everyday language, and the contestants need to write a program to solve the task as fast as possible—and hopefully, free of errors.
But it’s a behemoth challenge for AI coders. The agents need to first understand the task—something that comes naturally to humans—and then generate code for tricky problems that challenge even the best human programmers.
AI programmers are nothing new. Back in 2021, the non-profit research lab OpenAI released Codex, a program proficient in over a dozen programming languages and tuned in to natural, everyday language. What sets DeepMind’s AI release—dubbed AlphaCode—apart is in part what it doesn’t need.
Unlike previous AI coders, AlphaCode is relatively naïve. It doesn’t have any built-in knowledge about computer code syntax or structure. Rather, it learns somewhat similarly to toddlers grasping their first language. AlphaCode takes a “data-only” approach. It learns by observing buckets of existing code and is eventually able to flexibly deconstruct and combine “words” and “phrases”—in this case, snippets of code—to solve new problems.
When challenged with the CodeContest—the battle rap torment of competitive programming—the AI solved about 30 percent of the problems, while beating half the human competition. The success rate may seem measly, but these are incredibly complex problems. OpenAI’s Codex, for example, managed single-digit success when faced with similar benchmarks.
“It’s very impressive, the performance they’re able to achieve on some pretty challenging problems,” said Dr. Armando Solar-Lezama at MIT, who was not involved in the research.
The problems AlphaCode tackled are far from everyday applications—think of it more as a sophisticated math tournament in school. It’s also unlikely the AI will take over programming completely, as its code is riddled with errors. But it could take over mundane tasks or offer out-of-the-box solutions that evade human programmers.
Perhaps more importantly, AlphaCode paves the road for a novel way to design AI coders: forget past experience and just listen to the data.
“It may seem surprising that this procedure has any chance of creating correct code,” said Dr. J. Zico Kolter at Carnegie Mellon University and the Bosch Center for AI in Pittsburgh, who was not involved in the research. But what AlphaCode shows is when “given the proper data and model complexity, coherent structure can emerge,” even if it’s debatable whether the AI truly “understands” the task at hand.
Language to CodeAlphaCode is just the latest attempt at harnessing AI to generate better programs.
Coding is a bit like writing a cookbook. Each task requires multiple tiers of accuracy: one is the overall structure of the program, akin to an overview of the recipe. Another is detailing each procedure in extremely clear language and syntax, like describing each step of what to do, how much of each ingredient needs to go in, at what temperature and with what tools.
Each of these parameters—say, cacao to make hot chocolate—are called “variables” in a computer program. Put simply, a program needs to define the variables—let’s say “c” for cacao. It then mixes “c” with other variables, such as those for milk and sugar, to solve the final problem: making a nice steaming mug of hot chocolate.
The hard part is translating all of that to an AI, especially when typing in a seemingly simple request: make me a hot chocolate.
Back in 2021, Codex made its first foray into AI code writing. The team’s idea was to rely on GPT-3, a program that’s taken the world by storm with its prowess at interpreting and imitating human language. It’s since grown into ChatGPT, a fun and not-so-evil chatbot that engages in surprisingly intricate and delightful conversations.
So what’s the point? As with languages, coding is all about a system of variables, syntax, and structure. If existing algorithms work for natural language, why not use a similar strategy for writing code?
AI Coding AIAlphaCode took that approach.
The AI is built on a machine learning model called “large language model,” which underlies GPT-3. The critical aspect here is lots of data. GPT-3, for example, was fed billions of words from online resources like digital books and Wikipedia articles to begin “interpreting” human language. Codex was trained on over 100 gigabytes of data scraped from Github, a popular online software library, but still failed when faced with tricky problems.
AlphaCode inherits Codex’s “heart” in that it also operates similarly to a large language model. But two aspects set it apart, explained Kolter.
The first is training data. In addition to training AlphaCode on Github code, the DeepMind team built a custom dataset from CodeContests from two previous datasets, with over 13,500 challenges. Each came with an explanation of the task at hand, and multiple potential solutions across multiple languages. The result is a massive library of training data tailored to the challenge at hand.
“Arguably, the most important lesson for any ML [machine learning] system is that it should be trained on data that are similar to the data it will see at runtime,” said Kolter.
The second trick is strength in numbers. When an AI writes code piece by piece (or token-by-token), it’s easy to write invalid or incorrect code, causing the program to crash or pump out outlandish results. AlphaCode tackles the problem by generating over a million potential solutions for a single problem—multitudes larger than previous AI attempts.
As a sanity check and to narrow the results down, the AI runs candidate solves through simple test cases. It then clusters similar ones so it nails down just one from each cluster to submit to the challenge. It’s the most innovative step, said Dr. Kevin Ellis at Cornell University, who was not involved in the work.
The system worked surprisingly well. When challenged with a fresh set of problems, AlphaCode spit out potential solutions in two computing languages—Python or C++—while weeding out outrageous ones. When pitted against over 5,000 human participants, the AI outperformed about 45 percent of expert programmers.
A New Generation of AI CodersWhile not yet on the level of humans, AlphaCode’s strength is its utter ingenuity.
Rather than copying and pasting sections of previous training code, AlphaCode came up with clever snippets without copying large chunks of code or logic in its “reading material.” This creativity could be due to its data-driven way of learning.
What’s missing from AlphaCode is “any architectural design in the machine learning model that relates to…generating code,” said Kolter. Writing computer code is like building a sophisticated building: it’s highly structured, with programs needing a defined syntax with context clearly embedded to generate a solution.
AlphaCode does none of it. Instead, it generates code similar to how large language models generate text, writing the entire program and then checking for potential mistakes (as a writer, this feels oddly familiar). How exactly the AI achieves this remains mysterious—the inner workings of the process are buried inside its as yet inscrutable machine “mind.”
That’s not to say AlphaCode is ready to take over programming. Sometimes its makes head-scratching decisions, such as generating a variable but not using it. There’s also the danger that it might memorize small patterns from a limited amount of examples—a bunch of cats that scratched me equals all cats are evil—and the output of those patterns. This could turn them into stochastic parrots, explained Kolter, which are AI that don’t understand the problem but can parrot, or “blindly mimic” likely solutions.
Similar to most machine learning algorithms, AlphaCode also needs computing power that few can tap into, even though the code is publicly released.
Nevertheless, the study hints at an alternative path for autonomous AI coders. Rather than endowing the machines with traditional programming wisdom, we might need to consider that the step isn’t always necessary. Rather, similar to tackling natural language, all an AI coder needs for success is data and scale.
Kolter put it best: “AlphaCode cast the die. The datasets are public. Let us see what the future holds.”
Image Credit: Pexels from Pixabay
As the focus on a transition to renewable energy intensifies, wind farms are being touted as an optimal way to produce electricity. Wind energy is emissions-free (once the turbines are up and running), sustainable, and relatively efficient; turbines are increasingly dotting both the landscape and seascape.
But wind energy’s pitfall is that it’s expensive, particularly offshore. Some studies have gone so far as to proclaim offshore wind’s “dismal economics” unsustainable, particularly without massive government subsidies to keep it afloat.
Rather than giving up, though, some companies are trying to innovate their way through this problem. One of these is a French startup called Eolink, which is building a five-megawatt floating turbine.
The floating part matters because transporting, installing, and maintaining an enormously tall and heavy turbine in the ocean depths is costly and difficult. Traditional turbines have their generators located at the main axle near the top of the support tower. This adds a lot of weight at the top of the tower, and the blades themselves are already heavy enough. All that weight needs to be offset by even more weight at the bottom (and significant strength along the tower’s entire height) to keep the whole thing from toppling over or bending in half.
What if the weight could instead be distributed between multiple support poles? That’s the design Eolink has in mind; its floating turbine swaps out the single large pole for four thinner ones angled towards each other in a pyramid shape. This not only distributes the weight of the turbine’s pieces, it allows the whole structure to be lighter.
Eolink’s floating wind turbine design. Image Credit: EolinkThe turbine Eolink plans to build as a proof of concept will have a generating capacity of five megawatts and weigh 1,100 tons. Its base will be a square with each side 171 feet (52 meters) long, and its rotor’s diameter will be 469 feet (143 meters). For reference, that’s about one and a half Big Bens, or four-fifths of the Washington Monument.
You wouldn’t think that such a huge piece of machinery would be simple to build and transport. But compared to conventional offshore turbines, Eolink’s design does carry a myriad of advantages in terms of both cost and ease.
For starters, the turbines can be built at a shipyard and towed to their intended location by a ship; since they’re not as heavy, they can be placed in shallower water than conventional offshore turbines. For maintenance, the turbine can similarly be towed back to a shipyard, rather than having to send people, robots, and ships to labor at sea for days on end.
Eolink estimates that 67 turbines a year could be built in a single shipyard. The company says its design will use about a third less steel than a traditional turbine with the same capacity, and thanks to a longer distance between the blades and mast, could produce 10 percent more energy at equivalent wind speeds.
The turbines are tethered to the ocean floor, and within the limited space that this allows each one, the whole structure can turn 120 degrees to orient itself into the wind; in other words, the blades will always be spinning as long as there’s even a little bit of wind blowing.
The company says it’s starting construction of its demonstrator turbine this month, deploying it in the spring, and commissioning it in 2024. They also hope to eventually scale up to a 20-megawatt turbine with the same design.
Eolink isn’t the only company trying to make offshore wind profitable and thus sustainable. A Swedish company called SeaTwirl is developing vertical-axis offshore turbines (and completed a seven-year-long trial of its technology earlier this year), and an American company called T-Omega Wind is working on a floating pyramid design similar to Eolink’s.
Whether these companies’ hard work will truly make a dent in the fight to reduce emissions remains to be seen. But it seems something needs to change with offshore wind, and finding ways to make it cheaper seems like a crucial piece of the effort.
Image Credit: Eolink
When a group of photons struck the nearly flawless mirrors of the James Webb Space Telescope earlier this year, they’d been traveling the void for 13.4 billion years. The light was emitted from distant galaxies at a time when the birth of everything we know and see was still, in a cosmic sense, recent history. Ancient doesn’t really do it justice.
Webb’s first deep field images—infrared recordings of minuscule patches of sky, jam-packed with galaxies—sparked a scramble among astronomers to find the oldest galaxies in view. The Hubble Space Telescope held the existing record with observations of a galaxy from when the universe was just 400 million years old. Webb’s larger mirrors and ability to see into the infrared parts of the spectrum were designed to do better.
On Friday, the telescope proved its mettle when a team of scientists—jointly known as JADES, a collaboration between the builders of two of Webb’s instruments, NIRcam and NIRspec—announced they’d confirmed observations of the oldest galaxies yet.
“For the first time, we have discovered galaxies only 350 million years after the Big Bang, and we can be absolutely confident of their fantastic distances,” said Brant Robertson from the University of California Santa Cruz, a member of the NIRCam science team and coauthor on a recent paper on the work.
Astronomers first began compiling a list of candidates by analyzing data from Webb’s NIRcam instrument, an exquisitely sensitive infrared camera. Almost immediately after Webb’s first images went public stories of extremely ancient galaxies hit the web.
But while NIRcam observations revealed a rich population of targets worthy of a closer look, official confirmation required detailed spectroscopic analysis.
“It’s very possible for closer galaxies to masquerade as very distant galaxies,” said astronomer and coauthor Emma Curtis-Lake from the University of Hertfordshire in the United Kingdom.
Thanks to NIRspec, in two recent studies (here and here), the teams were able conduct spectroscopic analysis—the gold standard for confirming the distance and age of these incredibly faint early galaxies—for a range of candidates. Though neither study has yet been peer-reviewed, the findings likely beat Hubble’s record.
The sliver of sky observed is about the size of the queen’s eye “on a pound coin held at arm’s length,” Liverpool John Moores University’s Renske Smit told the BBC. Within that eye are almost 100,000 galaxies, each captured at a moment billions of years ago.
To measure the age of a galaxy near the beginning of the universe, scientists measure its “redshift.” As light travels, the expansion of the universe stretches out its wavelength, drawing it into the redder parts of the spectrum. Some of the most ancient light has been stretched out of the visible spectrum and into the infrared—Webb’s specialty.
The oldest galaxies are not only visible in the infrared, but their spectrum also cuts off at a specific point due to the scattering of intergalactic hydrogen. Faint infrared galaxies exhibiting this cutoff, which moves with greater redshift, filled out a pool of candidates. The team then dedicated 28 hours’ observation time to 250 of these with NIRspec. This detailed spectroscopic analysis included specific atomic signatures and nailed down the redshift.
Four galaxies proved exceptionally old, with redshifts greater than 10. Two showed redshifts at 13, from a time when the universe was just 330 million years old. The team says these galaxies are small, just a hundred million solar masses, and made up of young stars less than a hundred million years old. The Milky Way, by comparison, is thought to have at least 100 billion stars, and the sun is some 4.6 billion years old. Despite their diminutive size, the team says these early galaxies produced stars at a prodigious rate, as much as 10 times faster than similarly sized galaxies nearer to the present day.
The Webb Advanced Deep Extragalactic Survey (JADES) searched for faint galax NIRcam images (left) to search for faint galaxies exhibiting a characteristic break in their spectra (known as the Lyman break). NIRspec then precisely measured candidate galaxy redshift (right). Four galaxies (center) were special for being earlier than any spectroscopically confirmed. Image Credit: NASA, ESA, CSA, and STScI, M. Zamani (ESA/Webb), L. Hustak (STScI). Science: B. Robertson (UCSC), S. Tacchella (Cambridge), E. Curtis-Lake (Hertfordshire), S. Carniani (Scuola Normale Superiore), and the JADES CollaborationThese galaxies now appear to hold the record for oldest ever spectroscopically confirmed, but the title may not last long. Though still awaiting confirmation, scientists have estimated some galaxies already captured by Webb are even older, and Webb was designed to see light from epochs as early as 100 million years after the Big Bang.
By studying the earliest stars and galaxies, scientists hope to learn more about galaxy formation and to pin down a period in the universe’s evolution known as reionization, when the strong light of the first stars ionized surrounding gas by stripping electrons from hydrogen and helium. As the stars in these four galaxies may have begun forming as much as 100 million years earlier, this first generation of stars may date back to as early as around 230 million years after the Big Bang.
“With these measurements, we can know the intrinsic brightness of the galaxies and figure out how many stars they have,” Robertson said. “Now we can start to really pick apart how galaxies are put together over time.”
Image Credit: NASA, ESA, CSA, and STScI
ARTIFICIAL INTELLIGENCEChatGPT Proves AI Is Finally Mainstream—and Things Are Only Going to Get Weirder
James Vincent | The Verge“OpenAI has previously sold access to GPT-3 as an API, but the company’s ability to improve the model’s ability to talk in natural dialogue and then publish it on the web for anyone to play with brought it to a much bigger audience. And no matter how imaginative AI researchers are in probing a model’s skills and weaknesses, they’ll never be able to match the mass and chaotic intelligence of the internet at large.”
AUTOMATIONAmazon’s Quest for the ‘Holy Grail’ of Robotics
Christopher Mims | The Wall Street Journal“Amazon, along with a collection of other robotics companies developing similar machines, are chasing what experts in the field call the ‘holy grail’ of robotics—machines as dexterous, quick and adaptable as a human arm and hand. Such a robot could someday be capable of handling any of the thousands—or in Amazon’s case, millions—of different goods carried in a typical e-commerce fulfillment warehouse. ‘What we’re doing is unlike anything that’s been done in human history—the scale we’re working at,’ says [Amazon’s Tye Brady].”
SCIENCEDNA That Was Frozen for 2 Million Years Has Been Sequenced
Antonio Regalado | MIT Technology Review“After an eight-year effort to recover DNA from Greenland’s frozen interior, researchers say they’ve managed to sequence gene fragments from ancient fish, plants, and even a mastodon that lived 2 million years ago. It’s the oldest DNA ever recovered, beating the mark set only last year when a different team recovered genetic material from a million-year-old mammoth tooth.”
NEUROSCIENCEWhat Causes Alzheimer’s? Scientists Are Rethinking the Answer.
Yasemin Saplakoglu | Quanta“The emerging new models of the disease are more complex than the amyloid explanation, and because they are still taking shape, it’s not clear yet how some of them may eventually translate into therapies. But because they focus on fundamental mechanisms affecting the health of cells, what’s being learned about them might someday pay off in new treatments for a wide variety of medical problems, possibly including some key effects of aging.”
ROBOTICS*Xiaomi’s Humanoid Drummer Beats Expectations
Evan Ackerman | IEEE Spectrum*“In a nice surprise, Xiaomi roboticists have taught the robot to do something that is, if not exactly useful, at least loud: to play the drums. The input for this performance is a MIDI file, which the robot is able to parse into drum beats. It then generates song-length sequences of coordinated whole-body trajectories which are synchronized to the music, which is tricky because the end effectors have to make sure to actuate the drums exactly on the beat.”
SPACEEight Artists Chosen for First Civilian Moon Trip
Trevor Mogg | Digital Trends“Four years ago, Japanese billionaire entrepreneur Yusaku Maezawa stood alongside SpaceX chief Elon Musk to announce plans for the first civilian mission to the moon. At the presentation in 2018, Maezawa said he had covered the costs of the weeklong mission and wanted to invite eight other people to join him on the lunar flyby, a voyage that will likely be similar to NASA’s current Artemis I mission involving the Orion spacecraft. …On Thursday, Maezawa confirmed that the search for his fellow crewmembers has now finished as he revealed eight young creatives from around the world.”
BIOTECHSynthetic Bacteria Becomes Smallest Lifeform That Can Move Around
Michael Irving | New Atlas“In 2010 scientists at JCVI unveiled the world’s first completely synthetic lifeform—a micro-organism derived from a synthetic chromosome made up of four chemicals and designed using a computer. …In the new study, scientists at Osaka Metropolitan University edited the latest version of the organism, known as syn3, to give it a new ability—movement. This synthetic bacteria is usually spherical and can’t get around on its own, so the team experimented by adding seven proteins thought to allow natural bacteria to swim.
NANOTECH*Girl With a Pearl Earring and Mona Lisa Recreated With Nanotechnology
Jason Arunn Murugesu | New Atlas*“Ting Xu at Nanjing University in China and his colleagues made these minuscule masterpieces using nanostructures that manipulate light rays that hit them, reflecting only specific colours, while suppressing all others. The technique is inspired by insects such as butterflies that have intricate colours in their wings that are created through structural means, rather than pigment. It allows for better colour reproduction than paints or dyes, which don’t work on these tiny scales, says Xu.”
Image Credit: Annie Spratt / Unsplash
Aviation is an essential part of the global economy, but it’s also a major source of damaging greenhouse gases. The first-ever hydrogen jet engine could be a step towards solving that problem.
Unlike ground transport, aviation’s stringent weight requirements mean it can’t be easily decarbonized using batteries. Despite huge progress, the energy density—a measure of how much power you can pack in per pound—of today’s most advanced lithium-ion cells is still far below that of conventional jet fuel.
That’s a problem for aircraft, because it means that adding more batteries to boost range doesn’t provide enough extra juice to cancel out the extra weight of the batteries. While lithium-ion technology continues to improve, even at its theoretical maximum capacity it won’t provide high enough energy densities for even medium-haul flights, and new battery chemistries are still a long way off.
That’s why there’s growing interest in hydrogen as a potential aviation fuel. While it still falls short of the energy density of kerosene, it packs considerably more punch than batteries and produces no emissions when burned.
It’s not a straight swap for jet fuel though, and getting planes to fly on hydrogen will require significant redesigns. But jet-engine maker Rolls-Royce and commercial airliner easyJet have taken the first steps with the first-ever ground test of a jet engine powered by hydrogen.
“The success of this hydrogen test is an exciting milestone,” Grazia Vittadini, Rolls-Royce’s chief technology officer, said in a statement. “We are pushing the boundaries to discover the zero carbon possibilities of hydrogen, which could help reshape the future of flight.”
The test was conducted using a converted Rolls-Royce AE 2100 engine and hydrogen produced using renewable tidal energy by the European Marine Energy Centre on the Orkney Islands in Scotland.
That’s an important detail, because not all hydrogen is created equal. So-called green hydrogen refers to fuel generated by splitting water into hydrogen and oxygen using renewable electricity. But the most common form today is gray hydrogen, which is produced from fossil fuels and generates considerable greenhouse gas emissions.
The companies hailed the breakthrough as a significant step towards decarbonizing the aviation sector. But there is a long way to go before hydrogen is likely to be powering a significant number of aircraft. For a start, firing up an engine on a ground-based test rig is a very different proposition than using it to fly a plane.
While the energy density of hydrogen is certainly better than batteries, an aircraft would need almost four time as much liquid hydrogen compared to jet fuel to fly the same distance, according to the BBC.
What’s more, liquid hydrogen needs to be chilled to -253C and kept under pressure. This would mean much larger and more complicated fuel tanks and systems for delivering the hydrogen to the engines, which would likely require planes to be completely redesigned. Despite those challenges though, easyJet is convinced that hydrogen is the most realistic option for decarbonizing aviation.
“We started a few years ago looking at what might power the aircraft of the future,” David Morgan, easyJet’s chief operating officer, told the BBC. “We looked at battery technology, and it was quite clear that the battery technology was probably not going to do it for the large commercial aircraft that we fly. We’ve come to the conclusion that hydrogen is a very exciting proposition for us.”
They aren’t the only ones. Airbus has been developing several aircraft concepts that use hydrogen as a fuel, and last week it announced plans to test an aircraft engine powered by a hydrogen fuel cell. Rather than burning hydrogen, fuel cells use chemical reactions to convert hydrogen into electricity that can then power an electric motor. The company is aiming to carry out the first test flights on an A380 airliner by 2026.
Even if these prototypes make it out of the lab, though, it’s an open question as to whether there will be enough fuel to support aviation anytime in the near future, given the tiny amount of green hydrogen currently produced. Nonetheless, hydrogen could play an important role in a greener future for aviation, and any progress should be welcomed.
Image Credit: Rolls-Royce
Sixty-six million years ago, an asteroid hit the Earth with the force of 10 billion atomic bombs and changed the course of evolution. The skies darkened and plants stopped photosynthesizing. The plants died, then the animals that fed on them. The food chain collapsed. Over 90 percent of all species vanished. When the dust settled, all dinosaurs except a handful of birds had gone extinct.
But this catastrophic event made human evolution possible. The surviving mammals flourished, including little proto-primates that would evolve into us.
Imagine the asteroid had missed, and dinosaurs survived. Picture highly-evolved raptors planting their flag on the moon. Dinosaur scientists, discovering relativity, or discussing a hypothetical world in which, incredibly, mammals took over the Earth.
This might sound like bad science fiction, but it gets at some deep philosophical questions about evolution. Is humanity just here by chance, or is the evolution of intelligent tool-users inevitable?
Brains, tools, language, and big social groups make us the planet’s dominant species. There are eight billion Homo sapiens on seven continents. By weight, there are more humans than all wild animals.
We’ve modified half of Earth’s land to feed ourselves. You could argue creatures like humans were bound to evolve.
In the 1980s, palaeontologist Dale Russell proposed a thought experiment in which a carnivorous dinosaur evolved into an intelligent tool user. This “dinosauroid” was big-brained with opposable thumbs and walked upright.
It’s not impossible but it’s unlikely. The biology of an animal constrains the direction of its evolution. Your starting point limits your endpoints.
If you drop out of college, you probably won’t be a brain surgeon, lawyer, or Nasa rocket scientist. But you might be an artist, actor, or entrepreneur. The paths we take in life open some doors and close others. That’s also true in evolution.
Giant dinosaurs and mammals through time. Credit: Nick LongrichConsider the size of dinosaurs. Beginning in the Jurassic, sauropod dinosaurs, Brontosaurus, and kin evolved into 30-50 ton giants up to 30 meters long—ten times the weight of an elephant and as long as a blue whale. This happened in multiple groups, including Diplodocidae, Brachiosauridae, Turiasauridae, Mamenchisauridae, and Titanosauria.
This happened on different continents, at different times, and in different climates, from deserts to rainforests. But other dinosaurs living in these environments didn’t become supergiants.
The common thread linking these animals was that they were sauropods. Something about sauropod anatomy—lungs, hollow bones with a high strength-to-weight ratio, metabolism, or all these things—unlocked their evolutionary potential. It let them grow big in a way that no land animals had ever before, or have since.
Likewise, the carnivorous dinosaurs repeatedly evolved huge, ten-meter, multi-ton predators. Over 100 million years, megalosaurids, allosaurids, carcharodontosaurids, neovenatorids, and finally tyrannosaurs evolved giant apex predators.
Brain size versus body mass for dinosaurs, mammals, and birds. Credit: Nick LongrichDinosaurs did big bodies well. Big brains not so much. Dinosaurs did show a weak trend towards increased brain size over time. Jurassic dinosaurs like Allosaurus, Stegosaurus, and Brachiosaurus had small brains.
By the late Cretaceous, 80 million years later, tyrannosaurs and duckbills had evolved larger brains. But despite its size, the T. rex brain still weighed just 400 grams. A Velociraptor brain weighed 15 grams. The average human brain weighs 1.3 kilograms.
Dinosaurs did enter new niches over time. Small herbivores became more common and birds diversified. Long-legged forms evolved later on, suggesting an arms race between fleet-footed predators and their prey.
Dinosaurs seem to have had increasingly complex social lives. They started living in herds and evolved elaborate horns for fighting and display. Yet dinosaurs mostly seem to repeat themselves, evolving giant herbivores and carnivores with small brains.
There’s little about 100 million years of dinosaur history to hint they’d have done anything radically different if the asteroid hadn’t intervened. We’d likely still have those supergiant, long-necked herbivores and huge tyrannosaur-like predators.
They may have evolved slightly bigger brains, but there’s little evidence they’d have evolved into geniuses. Neither is it likely that mammals would have displaced them. Dinosaurs monopolized their environments to the very end, when the asteroid hit.
Mammals, meanwhile, had different constraints. They never evolved supergiant herbivores and carnivores. But they repeatedly evolved big brains. Massive brains (as large or larger than ours) evolved in orcas, sperm whales, baleen whales, elephants, leopard seals, and apes.
Today, a few dinosaur descendants—birds like crows and parrots—have complex brains. They can use tools, talk, and count. But it’s mammals like apes, elephants, and dolphins that evolved the biggest brains and most complex behaviors.
So did eliminating the dinosaurs guarantee mammals would evolve intelligence?
Well, maybe not.
Starting points may limit endpoints, but they don’t guarantee them either. Steve Jobs, Bill Gates, and Mark Zuckerberg all dropped out of college. But if dropping out automatically made you a multibillionaire, every college dropout would be rich. Even starting in the right place, you need opportunities and luck.
The evolutionary history of primates suggests our evolution was anything but inevitable. In Africa, primates did evolve into big-brained apes and, over 7 million years, produced modern humans. But elsewhere primate evolution took very different paths.
When monkeys reached South America 35 million years ago, they just evolved into more monkey species. And primates reached North America at least three separate times, 55 million years ago, 50 million years ago, and 20 million years ago. Yet they didn’t evolve into a species who make nuclear weapons and smartphones. Instead, for reasons we don’t understand, they went extinct.
In Africa, and Africa alone, primate evolution took a unique direction. Something about Africa’s fauna, flora, or geography drove the evolution of apes: terrestrial, big-bodied, big-brained, tool-using primates. Even with the dinosaurs gone, our evolution needed the right combination of opportunity and luck.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Enrique from Pixabay
This past summer saw the launch of the biggest four-day work week trial in the world, as 3,300 people across several different types of businesses in the UK started working 80 percent of their regular hours for 100 percent of their pay. Employees had to maintain the same level of productivity they had while working five days per week, and assess the new schedule’s impact on various aspects of their mental and physical well-being.
Halfway through the six-month trial, feedback from both employees and companies was overwhelmingly positive; people felt they were more productive and less stressed, and some businesses even saw their financial performance improve.
All the while, a whole other four-day work week experiment was quietly underway on the other side of the pond(s). Run by the same organization—a nonprofit coalition called 4 Day Week Global—this trial involved 903 employees across 33 companies, with the largest proportion (40 percent) based in the US. The remainder were in Australia, Ireland, the UK, New Zealand, and Canada.
The trial’s results were reported last week, and similar to the UK pilot (and the Iceland one before it), it was a resounding success. 96.9 percent of employees want to stick with a four-day week rather going back to five days, and more than half of the participating companies have already decided to implement four-day weeks. Employees’ self-assessed work performance improved, as did their “satisfaction across multiple domains of life.”
While these results can’t quite be called surprising—most of us would quickly sign up to work fewer hours if given the option—it’s worth noting that the culture around work is pretty different in the US than it is in the UK or Australia. A stereotype about Americans is that they prioritize work over almost all else; compared to Europeans, for example, we’re seen as workaholics, putting in longer hours and taking fewer vacation days.
Studies have found that Americans are more likely to conflate career and identity, defining themselves by what they do for a living, whereas Europeans are more likely to see their jobs as a means to live comfortably and do other things they enjoy.
Is it surprising, then, that American workers appear thrilled to cut their work week short?
It’s hard to say how different the results may have been in a pre-Covid world; the pandemic caused people to reevaluate how they were spending their time and what was important to them in life. If we’d never gotten a chance to zoom out from the nine-to-five grind, try working from home, and getting more flexibility with our time and schedules, a four-day work week might have felt less plausible. But in the post-pandemic world, all kinds of hybrid and remote work options are suddenly on the table.
By sector, most of the companies in the US trial were grouped as administrative, IT, and telecoms, followed by professional services and non-profits. Notably, 52 percent of the companies were very small, with ten or fewer employees. Might people at small companies feel less pressure to conform to an hours-intensive company culture or climb the corporate ladder?
It’s certainly possible. But in general, it seems people like having the option to be more efficient with their work hours—if you waste less time on the internet or wandering around the office chatting with coworkers, it’s surprising how much you can get done in a day—and spend their extra free time however they want.
4 Day Week Global says they’re launching new programs in different parts of the world every quarter. Companies that are interested in taking part in one can inquire about upcoming pilots in their country. With the rave reviews these trials keep getting, it seems likely that the four-day week will catch on more widely with time.
There are a few caveats we should keep in mind, though. As one of the more populous countries in the world and one with a high level of economic inequality, the US isn’t a straightforward place to implement any sort of across-the-board policy in terms of work (well, other than those related to protecting employees’ rights and preventing abuse or exploitation).
About three-fourths of the people who participated in the US trial had a bachelor’s degree. But looking at the broader American population, only 37.9 percent of adults age 25 or over have a bachelor’s (that’s up from 30.4 percent in 2011). The nature of most professional jobs is more conducive to a shortened week than jobs that require less education, and implementing four-day policies that applied to some but not others could cause the class divide to widen further.
4 Day Week Global believes that a shorter work week has the potential to not only improve business productivity and worker health outcomes, but can strengthen families and communities and contribute to greater gender equality. These are all worthwhile aims; if working fewer hours could help achieve them, it seems worth a try.
Image Credit: Israel Andrade on Unsplash
At the beginning of my research career around 15 years ago, any suggestion that a bee, or any invertebrate, had a mind of its own or that it could experience the world in an intricate and multifaceted way would be met with ridicule. As Lars Chittka points out in the opening chapters of The Mind of a Bee, the attribution of human emotions and experiences was seen as naivety and ignorance; anthropomorphism was a dirty word.
Pet owners eagerly ascribe emotions to their animals, but the simple brain of a bee surely could not experience the rich tapestry that is our existence. They are far too simplistic and robotic, right?
Lars Chittka has been researching honeybees for the past 30 years. The Mind of a Bee is a collection of his research stories. It also covers the influential figures in bee research and provides a historical perspective on the research that much behavioral work is built on today.
People have long been curious about the behavior of bees. Many questions posed in the 1800s are still around. While Chittka’s beautifully collated and captivating “story” does not present research results that are necessarily new, to read them presented together like this, I find myself tantalized by questions I had not thought to ponder. For example, how do bees decide who stays and who leaves when a swarm is formed?
The World of a BeeThe book opens by challenging you to put yourself into the world of a bee.
A honeybee’s experience of the world is so completely foreign to our own that to understand and research it is a challenge not to be underestimated. Indeed, it is understandable that we have relegated the experience of bees to something simplistic and robotic when you discover the difficulties faced by researchers.
First, picture yourself as a bee. You have wings, allowing flight. Your vision is not as sharp anymore, worse than your grandfather’s with his coke-bottle glasses, but you see things more quickly. Life is experienced on a faster timeline—what was once a movie is now more like a series of images in a slideshow.
The antennae protruding from your head function as hands, ears, tongues, and noses, all in one. You can tell if someone has visited a flower before you—a flower you picked out of a field of hundreds by its scent, and which you found by following the directions you felt a fellow bee dance for you inside the pitch-black hive perhaps ten kilometers from your current position.
Chittka then invites us to imagine the life of the bee. Upon exiting the hive for the first time, you must learn its location through a series of flights—behavior observed in other central-place foragers such as ants and wasps. Failure to recognize your hive and return home equals death.
Once you have memorized the location of your hive, you then must successfully navigate your way to and from various resource-rich patches as efficiently as possible, learning new locations, the timing of certain flowers releasing their nectar, and the techniques required to manipulate other flowers into relinquishing theirs.
So far, this sounds instinctual, a basic response to hunger. Yet Chittka presents additional research—historical and current—that provides insights into the cognitive skills of bees. We learn that bees can count. They can learn rules and categorize flowers. And they can learn from others, not only which flowers are rewarding, but how to access them.
One of my favorite experiments, perhaps for the videos that accompanied the publication, is of bumblebees pushing balls into holes to get rewards. This skill can be learnt by an observer bee and, what is truly fascinating, it can be improved upon. The observer bee can solve the task by copying the goal rather than strictly copying the technique, demonstrating an understanding of the task and the desired outcome.
But when would a bee ever need to push a ball into a hole to be rewarded with some “nectar”?
As Chittka rightly points out, the questions we pose to understand the minds of bees must have a biological relevance to make sense. That is, we need to understand what is important to the survival of bees, what is essential in their existence, and frame our questions of intelligence and sentience around that aspect. If we ask the wrong questions, we will never fully understand the answers—like asking a fish to climb a tree and finding it lacking.
Lars Chittka. Image Credit: Wikimedia Commons, CC BY-SAConsciousness and EmotionThe punch this book packs is in the subtle build-up to the final chapters, whereupon it becomes increasingly hard to deny the “mind” of a bee.
While it is impossible to prove consciousness in another organism, the research Chittka has collated provides a compelling argument. In The Mind of a Bee, you will read that bees feel emotions and pain, display metacognition (that is, they know what they know), and show individual differences in their ability to learn, with fast and slow learners. Bees are aware of their bodies and the outcomes of their actions, and they display intentionality through tool use—previously only recognized in humans, primates, and the corvidae family of birds.
Regardless of whether you believe a bee has a mind or not, globally there has been a change in research practices as invertebrates are seen to experience the world more fully.
Ethics approval is required for work on some invertebrates, including crustaceans and cephalopods, and statements of ethical treatment of other invertebrates are required for submission of manuscripts to some journals. To suggest an invertebrate, such as a bee, may have these fuller experiences of life is no longer attracting ridicule, but instead is creating an uncomfortable space for insect researchers, who may not wish to confront the reality of their experiments.
We have underestimated the intelligence of bees and other “lower” species for far too long; it is time to pay attention. Chittka shows us that bees have the key ingredients of a mind: they have a representation of space, they can learn by observation, and they display simple tool use. Bees have demonstrated a flexible memory, with ideas of what they want to achieve, an ability to explore suitable solutions to get it, and an awareness of the possible outcomes of their own actions.
Experiments have further shown that bees appear to attach emotional states to rewards and punishments. While their biology and experience of the world is very different to ours, it is reasonable to believe that they do indeed possess a mind capable of experiencing the rich tapestry of life we have so long thought only available to us.
Written with moments of levity and soaked in curiosity, The Mind of a Bee is a delight. While some may not be ready to ascribe sentience to something as “simple” as a bee, this book will prompt you to question why not. As Chittka so eloquently put it in a recent talk: “We are thinking, suffering, enjoying beings in a world of other thinking, suffering and enjoying beings, with different minds and perceptions.”
I for one am looking at the world a little differently with that in mind.
This article is republished from The Conversation under a Creative Commons license. Read the original article.
Image Credit: Jon Sullivan/Wikimedia Commons
AI hates uncertainty. Yet to navigate our unpredictable world, it needs to learn to make choices with imperfect information—as we do every single day.
DeepMind just took a stab at solving this conundrum. The trick was to interweave game theory into an algorithmic strategy loosely based on the human brain called deep reinforcement learning. The result, DeepNash, toppled human experts in a highly strategic board game called Stratego. A notoriously difficult game for AI, Stratego requires multiple strengths of human wit: long-term thinking, bluffing, and strategizing, all without knowing your opponent’s pieces on the board.
“Unlike chess and Go, Stratego is a game of imperfect information: players cannot directly observe the identities of their opponent’s pieces,” DeepMind wrote in a blog post. With DeepNash, “game-playing artificial intelligence (AI) systems have advanced to a new frontier.”
It’s not all fun and games. AI systems that can easily maneuver the randomness of our world and adjust their “behavior” accordingly could one day handle real-world problems with limited information, such as optimizing traffic flow to reduce travel time and (hopefully) quenching road rage as self-driving cars become ever more present.
“If you’re making a self-driving car, you don’t want to assume that all the other drivers on the road are perfectly rational, and going to behave optimally,” said Dr. Noam Brown at Meta AI, who wasn’t involved in the research.
DeepNash’s triumph comes hot on the heels of another AI advance this month, where an algorithm learned to play Diplomacy—a game that requires negotiation and cooperation to win. As AI gains more flexible reasoning, becomes more generalized, and learns to navigate social situations, it may also spark insights into our own brains’ neural processes and cognition.
Meet StrategoIn terms of complexity, Stratego is a completely different beast compared to chess, Go, or poker—all games that AI has previously mastered.
The game is essentially capture the flag. Each side has 40 pieces they can place at any position on the board. Each piece has a different name and numerical rank, such as “marshal,” “general,” “scout,” or “spy.” Higher ranking pieces can capture lower ones. The goal is to eliminate the opposition and capture their flag.
Stratego is especially challenging for AI because players can’t see the location of their opponents’ pieces, both during initial setup and throughout gameplay. Unlike chess or Go, in which each piece and movement is in view, Stratego is a game with limited information. Players must “balance all possible outcomes” any time they make a decision, the authors explained.
This level of uncertainty is partly why Stratego has stumped AI for ages. Even the most successful game-play algorithms, such as AlphaGo and AlphaZero, rely on complete information. Stratego, in contrast, has a touch of Texas Hold ’em, a poker game DeepMind previously conquered with an algorithm. But that strategy faltered for Stratego, largely because of the length of game, which unlike poker, normally encompasses hundreds of moves.
The number of potential game plays is mind-blowing. Chess has one starting position. Stratego has over 1066 possible starting positions—far more than all the stars in the universe. Stratego’s game tree, the sum of all potential moves in the game, totals a staggering 10535.
“The sheer complexity of the number of possible outcomes in Stratego means algorithms that perform well on perfect-information games, and even those that work for poker, don’t work,” said study author Dr. Julien Perolat at DeepMind. The challenge is “what excited us,” he said.
A Beautiful MindStratego’s complexity means that the usual strategy for searching gameplay moves is out of the question. Dubbed the Monte Carlo tree search, a “stalwart approach to AI-based gaming,” the technique plots out potential routes—like branches on a tree—that could result in victory.
Instead, the magic touch for DeepNash came from the mathematician John Nash, portrayed in the film A Beautiful Mind. A pioneer in game theory, Nash won the Nobel Prize for his work for the Nash equilibrium. Put simply, in each game, players can tap into a set of strategies followed by everyone, so that no single player gains anything by changing their own strategy. In Statego, this brings about a zero-sum game: any gain a player makes results in a loss for their opponent.
Because of Stratego’s complexity, DeepNash took a model-free approach to their algorithm. Here, the AI isn’t trying to precisely model its opponent’s behavior. Like a baby, it has a blank slate, of sorts, to learn. This set-up is particularly useful in early stages of gameplay, “when DeepNash knows little about its opponent’s pieces,” making predictions “difficult, if not impossible,” the authors said.
The team then used deep reinforcement learning to power DeepNash, with the goal of finding the game’s Nash equilibrium. It’s a match made in heaven: reinforcement learning helps decide the best next move at every step of the game, while DeepNash provides an overall learning strategy. To evaluate the system, the team also engineered a “tutor” using knowledge from the game to filter out obvious mistakes that likely wouldn’t make real-world sense.
Practice Makes PerfectAs a first learning step, DeepNash played against itself in 5.5 billion games, a popular approach in AI training dubbed self-play.
When one side wins, the AI gets awarded, and its current artificial neural network parameters are strengthened. The other side—the same AI—receives a penalty to dampen its neural network strength. It’s like rehearsing a speech to yourself in front of a mirror. Over time, you figure out mistakes and perform better. In DeepNash’s case, it drifts towards a Nash equilibrium for best gameplay.
What about actual performance?
The team tested the algorithm against other elite Stratego bots, some of which won the Computer Stratego World Championship. DeepNash squashed its opponents with a win rate of roughly 97 percent. When unleashed against Gravon—an online platform for human players—DeepNash trounced its human opponents. After over two weeks of matches against Gravon’s players in April this year, DeepNash rose to third place in all ranked matches since 2002.
It shows that bootstrapping human play data to AI isn’t needed for DeepNash to reach human-level performance—and beat it.
The AI also exhibited some intriguing behavior with the initial setup and during gameplay. For example, rather than settling on a particular “optimized” starting position, DeepNash constantly shifted the pieces around to prevent its opponent from spotting patterns over time. During gameplay, the AI bounced between seemingly senseless moves—such as sacrificing high-ranking pieces—to locate the opponent’s even higher-ranking pieces upon counterattack.
DeepNash can also bluff. In one play, the AI moved a low-ranking piece as if it were a high-ranking one, luring the human opponent to chase after the piece with its high-ranking colonel. The AI sacrificed the pawn, but in turn, lured the opponent’s valuable spy piece into an ambush.
Although DeepNash was developed for Stratego, it’s generalizable to the real-world. The core method can potentially instruct AI to better tackle our unpredictable future using limited information—from crowd and traffic control to analyzing market turmoil.
“In creating a generalizable AI system that’s robust in the face of uncertainty, we hope to bring the problem-solving capabilities of AI further into our inherently unpredictable world,” the team said.
Image Credit: Derek Bruff / Flickr
We humans can’t stop playing with our food. Just think of all the different ways of serving potatoes—entire books have been written about potato recipes alone. The restaurant industry was born from our love of flavoring food in new and interesting ways. My team’s analysis of the oldest charred food remains ever found show that […]
BIOTECH Biotech Labs Are Using AI Inspired by DALL-E to Invent New Drugs Will Douglas Heaven | MIT Technology Review “These protein generators can be directed to produce designs for proteins with specific properties, such as shape or size or function. In effect, this makes it possible to come up with new proteins to do […]
Wormholes might sound like something that belongs in a Star Trek episode rather than a research paper, but scientists just simulated one on Google’s Sycamore quantum computer. The result suggests these devices could be used to test out fundamental physical theories. The possibility of wormholes was first outlined in a 1935 paper by Albert Einstein […]
Austin, Texas-based 3D printing construction company ICON has gotten some pretty significant projects off the ground in recent years, from a 50-home development in Mexico to a 100-home neighborhood in Texas. This week the company won a NASA contract that will help it get an even bigger project much further off the ground—all the way […]
More people are opting to go vegetarian or vegan as factory farming’s impact on the planet becomes more apparent. But one carnivorous delight they may not have to give up is bacon, especially if they’re willing to be a bit flexible. A Dutch startup has been working on cultured bacon for a few years now, […]
We all imagine better versions of ourselves. Smarter. More attractive. Physically nimble to rock the dance floor, conquer martial arts, or run that ultramarathon. And as we grow older, we want to live longer, healthier lives—or even reverse the aging process itself. For eons, people have tapped into various resources to give themselves an edge. […]
AI has mastered some of the most complex games known to man, but while it often excels at competition, cooperation doesn’t come as naturally. Now an AI from Meta has mastered the game Diplomacy, which requires you to work with other players to win. Google’s mastery of the game of Go was hailed as a […]
In his latest book, the oncologist and acclaimed writer Siddhartha Mukherjee focuses his narrative microscope on the cell, the elementary building block from which complex systems and life itself emerge. It is the coordination of cells that allow hearts to beat, the specialization of cells that create robust immune systems, and the firing of cells […]
ARTIFICIAL INTELLIGENCE Meta’s ‘Cicero’ AI Trounced Humans at Diplomacy Without Revealing Its True Identity Mack DeGeurin | Gizmodo “Meta says Cicero more than doubled the average score of human players across 40 anonymous online Diplomacy games and ranked in the top 10% of players who played more than one game. Cicero even placed 1st in […]
November 15, 2022 marked a milestone for our species, as the global population hit 8 billion. Just 70 years ago—within a human lifetime—there were only 2.5 billion of us. In AD1, fewer than one-third of a billion. So how have we been so successful? Humans are not especially fast, strong, or agile. Our senses are […]
The early 2020s have been a chaotic time, with seemingly one crisis after another befalling humanity: the Covid-19 pandemic, inflation and supply chain upheaval, political instability and extremism, climate change…the list goes on. But what does it all really mean, or matter, when you zoom out and look at the big picture? An astronomy professor […]
Artificial intelligence is on a tear. Machines can speak, write, play games, and generate original images, video, and music. But as AI’s capabilities have grown, so too have its algorithms. A decade ago, machine learning algorithms relied on tens of millions of internal connections, or parameters. Today’s algorithms regularly reach into the hundreds of billions […]
A new cancer therapy is a match made in heaven. On one side is CRISPR, the gene-editing technology that’s taken genetic engineering by storm. The other is a therapy called CAR-T, which transforms normal immune cells into super soldiers that hunt down specific cancers. Scientists have long sought to combine these two big advances into […]
At about 10 o’clock on the night of February 28, 2021, a fireball streaked through the sky over England. The blazing extraterrestrial visitor was seen by more than 1,000 people, and its descent was filmed by 16 dedicated meteor-tracking cameras from the UK Fireball Alliance and many dashboard and doorbell cams. With the time difference […]
FUTURE Picture Limitless Creativity at Your Fingertips Kevin Kelly | Wired “For the first time in history, humans can conjure up everyday acts of creativity on demand, in real time, at scale, for cheap. Synthetic creativity is a commodity now. Ancient philosophers will turn in their graves, but it turns out that to make creativity—to […]
Human spaceflight has suffered a significant lull since the groundbreaking Apollo missions of the 1960s and 70s. But that looks set to change following the successful launch of NASA’s Artemis I mission, a crucial first step towards taking astronauts back to the moon. Since the space shuttle made its final outing in 2011, NASA has […]
In 2020 cultured meat startup Memphis Meats raised $161 million in Series B funding, making it the most-funded startup in the industry. The investment validated cultured meat’s technological soundness and indicated that consumer interest in these products was likely to grow. After changing its name to Upside Foods in 2021, this past April the company […]
Women have far more control over their bodies today than we did before the pill was invented. But reproduction is a two-player game, and women still carry a far greater burden than men when it comes to preventing unwanted pregnancy. A Virginia-based startup called Contraline is hoping to change this. The company developed a new […]
What was science fiction is now scientific reality: with a series of targeted electrical zaps to the spinal cord, nine paralyzed people immediately walked again with help from a robot. Five months later, half of the participants no longer needed those zaps to walk. Does the sentence sound a bit familiar? By themselves, the results—while […]
Private companies are playing an ever greater role in space, in many cases with the blessing of national space agencies. Now Japan has issued a startup the first-ever license to conduct business activity on the moon, which could change the face of lunar exploration. SpaceX’s rapid ascent to become one of the world’s premier launch […]
In searching for planets and studying their stars, I’ve had the privilege of using some of the world’s great telescopes. However, our team has recently turned to an even larger system to study the cosmos: Earth’s forests. We analyzed radioactive signatures left in tree rings around the world to study mysterious “radiation storms” that have […]
BIOTECH CRISPR Cancer Trial Success Paves the Way for Personalized Treatments Heidi Ledford | Nature “A small clinical trial has shown that researchers can use CRISPR gene editing to alter immune cells so that they will recognize mutated proteins specific to a person’s tumors. Those cells can then be safely set loose in the body […]
An enormous neutrino observatory buried deep in the Antarctic ice has discovered only the second extra-galactic source of the elusive particles ever found. In results published last week in Science, the IceCube collaboration reports the detection of neutrinos from an “active galaxy” called NGC 1068, which lies some 47 million light-years from Earth. How to […]
Of the four main blood types, the most common is Type O-positive, accounting for 37 percent of the population. Type O-negative, meanwhile, is a universal donor, meaning it carries the lowest risk of causing serious reactions for most people who receive it in a transfusion. For people with rare blood types, though, it can be […]
The push to retire combustion engine cars and transition to electric vehicles is intensifying, even as energy prices rise and electrical grids get stretched thin. Cutting out the middle man with cars that can convert sunlight to horsepower all on their own could help. A new one is coming to market next year—German company Sono […]
I get frustrated every time I grab a glove and realize it’s for the opposite hand. But to synthetic biologists, this annoyance is a biological quirk that could help transform medicine. Think long-lasting medications that can be taken once a month rather than three times a day. Or biomolecule-based diagnostic tools that linger inside the […]
ROBOTICS Having AIs Train Robot Dogs to Balance Makes Them a Lot Cheaper Jeremy Tsu | New Scientist “An AI has been used to train a small robot dog to perform cleaning tasks. The hardware cost a total of $6300, which is less than a tenth of the price tag of the well-known robot dogs […]
The US has a housing shortage problem. We also have a plastic waste problem. What if we could solve both these problems simultaneously with an unexpected two birds/one stone innovation? If you’ve ever longed to live in a small house made out of 100,000 recycled plastic water bottles, your lucky day is just around the […]
It’s a paradox: Life needs water to survive, but a world full of water can’t generate the biomolecules that would have been essential for early life. Or so researchers thought. Water is everywhere. Most of the human body is made of it, much of planet Earth is covered by it, and humans can’t survive more […]
Over the course of the Covid pandemic, millions of people invested in air purifiers in an effort to keep their homes or offices as virus-free as possible. Soon they may be able to trade their electric air purifiers for a version that’s far more natural: a plant. Last week a French company called Neoplants unveiled […]
A nightmare scenario keeps synthetic biologists up at night. A strain of bacteria with an extensively revised genetic code leaks out of the lab. A part of its genetic machinery transfers to an unsuspecting host—allowing it to co-opt the host cell to reproduce, even if it means harming the host. The bacteria wasn’t engineered to […]
On the smallest scales, our universe gets weird. Particles act like billiard balls or waves on water, depending how you probe them. Properties can’t be measured simultaneously or tend to smear uncertainly over a range of values. Human intuition fails us. For much of the last century, all this weirdness was mostly the domain of […]
Bias in AI systems is proving to be a major stumbling block in efforts to more broadly integrate the technology into our society. A new initiative that will reward researchers for finding any prejudices in AI systems could help solve the problem. The effort is modeled on the bug bounties that software companies pay to […]
BIOTECHNOLOGY Neoplants Bioengineers Houseplants to Use Them as Air Purifiers Romain Dillet | TechCrunch “Neoplants targets specifically a group of indoor air pollutants that can’t be efficiently captured by traditional air purifiers. Most air purifiers focus on particulate matters. But it’s harder to tackle volatile organic compounds (VOCs). That’s why Neoplants focuses on two categories […]
Remnants of ancient viral pandemics in the form of viral DNA sequences embedded in our genomes are still active in healthy people, according to new research my colleagues and I recently published. HERVs, or human endogenous retroviruses, make up around eight percent of the human genome, left behind as a result of infections that humanity’s […]
Beyond Meat has been selling plant-based imitation meat since 2012, with an ever-growing list of products including burgers, ground meat, sausage, meatballs, jerky, and chicken. This week the company added one more to the list: steak. Their plant-based steak tips are the first product of its kind to hit the market. The big differentiator between […]
Ninety percent of American adults don’t eat enough fruits and vegetables, opting for fast food and processed foods instead. Cost, flavor, and convenience are all factors in this imbalance, but as health statistics show, we should be working harder to reverse our dietary trends. A startup called PairWise is out to help change the way […]
Sitting on the Marine Atlantic ferry, I’m watching the Newfoundland skyline disappear on the horizon as I type away. I see the rocking of the ocean waves, inhale its salty breeze, feel and hear the buzz of the ship’s rumbling engine. I try to focus on writing this sentence, but my eyes hopefully scan the […]
Range anxiety is one of the biggest barriers to electric vehicle adoption, driven in large part by the long time it takes to recharge. But a new approach can give a battery enough juice to travel 200 miles in just over 10 minutes. Battery technology is the biggest barrier to widespread electric vehicle adoption, because […]
The point of this text is not to predict how many people will ever live. What I learned from writing this post is that our future is potentially very, very big. If we keep each other safe—and protect ourselves from the risks that nature and we ourselves pose—we are only at the beginning of human […]
Even as 3D printing gains more traction as a next-gen homebuilding method, new ideas for sustainable, affordable housing are continuously popping up. There’s “foldable” homes; homes that ship in kits then are assembled like Ikea furniture; prefab homes are made of structural panels; and soon, if one ambitious company can help it, there will be […]
A few years ago, most of the vertical farms being built were for growing leafy greens. Since then, the technology has not only scaled, with each new farm seemingly out to top the square footage of its predecessors—it’s also expanded its repertoire, with everything from algae to mushroom roots being grown indoors under LED lights. […]
Several weeks ago, climate scientists said an extreme heatwave in China was the worst on record anywhere. Days later, Pakistan declared a national state of emergency due to unprecedented flooding. In Europe, record summer heat included the highest temperature ever recorded in Britain. And, in the US, the worst drought in over a thousand years […]
The venerable stock image site, Getty, boasts a catalog of 80 million images. Shutterstock, a rival of Getty, offers 415 million images. It took a few decades to build up these prodigious libraries. Now, it seems we’ll have to redefine prodigious. In a blog post last week, OpenAI said its machine learning algorithm, DALL-E 2, […]
Despite decades of research, the human brain remains largely a mystery to science. A new $500 million project to create the most comprehensive map of it ever could help change that. Our brains are among the most complex objects in the known universe. Deciphering how they work could bring tremendous benefits, from finding ways to […]
ARTIFICIAL INTELLIGENCE Meta’s New Text-to-Video AI Generator Is Like DALL-E for Video James Vincent | The Verge “The videos are clearly artificial, with blurred subjects and distorted animation, but still represent a significant development in the field of AI content generation. …while it’s clear these videos are computer-generated, the output of such AI models will […]
Coastal urban centers around the world are urgently looking for new, sustainable water sources as their local supplies become less reliable. In the US, the issue is especially pressing in California, which is coping with a record-setting, multi-decadal drought. California Gov. Gavin Newsom recently released a $8 billion plan for coping with a shrinking water […]
Restaurants have been struggling with labor shortages since the worst days of the pandemic, and the situation doesn’t seem to be getting any better. After employees quit in droves and millions of restaurants shut down and laid off their entire staff, people seem to have found other ways to earn a living. The result for […]
Three and a half months ago, the biggest four-day workweek trial in the world to date kicked off in the UK. Over 3,300 employees from 70 different companies—including everything from large corporations to small neighborhood pubs—started getting 100 percent of their pay for working 80 percent of their typical schedule. Sounds like a pretty good […]
Housing the world’s rapidly-growing population will require massive urban expansion and lots of concrete and steel, but these materials have a huge carbon footprint. A shift to building cities out of wood could avoid more than 100 billion tons of CO2 emissions, according to a new study. Replacing reinforced concrete with timber might sound unwise, […]
Astronauts on the space station may seem distant, but they’re only 248 miles from Earth: a little more than the drive from New York City to Washington DC. Everything they need can be delivered in relatively short order. Astronauts visiting Mars won’t have such easy access. The red planet’s average distance from Earth is 140 […]
ARTIFICIAL INTELLIGENCE An AI-generated Artwork’s State Fair Victory Fuels Arguments Over ‘What Art Is’ James Vincent | The Verge “The rise of text-to-AI image generators has only just begun, but already, the programs are sparking heated debates about the nature of art, whether this software poses a threat to artists’ livelihoods, and whether or not […]
A Jetsons-inspired future where people zip around the skies in flying cars seems to still be a long way off, but that’s not stopping people from designing, launching, and even starting production of personal aircraft. Now London-based SkyFly is joining the fray; the company recently started taking pre-orders on a personal eVTOL it calls the […]
The study of present-day species has delivered a clear verdict on humanity’s place in the living world: right alongside chimpanzees and bonobos. However, this does not tell us much about our earliest human representatives, their biology or geographical distribution—in short, how we became human. For this, we mainly have to rely on the morphology of […]
Even before labor shortages and supply chain issues began plaguing the economy, the food service industry was bringing in robots. From flipping burgers to making pizzas, automation has been taking over a variety of food preparation tasks. A San Francisco restaurant has now taken it to the next level, opening what it claims is the […]
AI and conventional computers are a match made in hell. The main reason is how hardware chips are currently set up. Based on the traditional Von Neumann architecture, the chip isolates memory storage from its main processors. Each computation is a nightmarish Monday morning commute, with the chip constantly shuttling data to-and-fro from each compartment, […]
Proposals for beaming solar power down from space have been around since the 1970s, but the idea has long been seen as little more than science fiction. Now, though, Europe seems to be getting serious about making it a reality. Space-based solar power (SBSP) involves building massive arrays of solar panels in orbit to collect […]
This week, scientists announced that the James Webb Space Telescope, which among its many talents can analyze the atmospheres of exoplanets, just confirmed the presence of carbon dioxide on a world orbiting a sun some 700 light-years away. It’s the first observation of CO2 in a planetary atmosphere beyond our solar system. But that discovery, […]
BIOTECH This Company Is About to Grow New Organs in a Person for the First Time Jessica Hamzelou | MIT Technology Review “In the coming weeks, a volunteer in Boston, Massachusetts, will be the first to trial a new treatment that could end up creating a second liver in their body. And that’s just the […]