These are the days of miracle and wonder.
Or at least, this is what Ideogram.ai made from the lyrics of Paul Simon’s “The Boy in the Bubble”.

Great news!

We are all going to live for many more long and healthy years, basking in the wealth and leisure of a society powered by abundant cheap energy and an explosion of medical and scientific miracles.

That is seriously how I felt by the end of my week at the TED conference, and I have to tell you, it made for a nice break from feeling like we all about to die in a ball of fire.

Apparently we are on the precipice of a new world where AIs figure out nuclear fusion and robots carry all our heavy packages…but where we somehow still have enough work and money to pay for pill-sized robots that give us colonoscopies from the inside out, and enough leisure time to create AI-enabled artworks in mediums we haven’t invented yet.

I am a bit hazy on the details of how we get from here to there, however. It definitely has something to do with AI (which is going to become more powerful than we can imagine, sooner than we expect) and also with quantum computing (which a helpful TED attendee explained much more clearly than the Google quantum computing expert who actually delivered a TED talk).

Perhaps it would all have been clearer to me if I hadn’t dropped science after grade 9…but a week of science talks also made it clear that science alone can’t save us.

The missing link: Not just for scientistsBecause the missing link in getting from here to there isn’t just science: It’s social science. No matter how many inventions and cures get unlocked in this era of imminent scientific revolution, they will only be transformative if they can be delivered at scale. That’s a matter of politics, economics and social organization, rather than scientific innovation.

Perhaps AI will help us untangle these challenges, too, but social solutions are unlikely to emerge fully formed from some lab or machine. Social solutions require consideration, negotiation, and usually some difficult trade-offs. Social solutions require people, and people are slow, messy and inconsistent.

People are also brilliant, creative and compassionate—which is why we need to be at the table while the AIs and the scientists and the quantum computers hatch our new era of miracles. Unfortunately, the big-picture promises of technology are often so big-picture that it’s hard to see where we fit in, where to take our seat at the table, or even how to get started.

That’s why the enormous scale and speed of transformation promised at TED left me hungry to get small and specific: To focus on the immediate, comprehensible steps that can close a little bit of the gap between the imperfect now and the soon-to-be awesome future.

The cure for AI utopianism: More AI!For me, that means trying a new AI tool, tactic or approach every week—or better yet, every day.

It’s tempting to just keep using AI in the ways that are now routine for me:

• Review this text and return a bulleted list of essential fixes (typos, errors) and improvements to consider
• Create a “Happy Birthday” image for my friend, with an image based on her interests. ( It’s easy to do in Ideogram.ai.)
• Write a Python script that can calculate the percentile values for each column in this CSV (Thanks to AI, I now use Python scripting to clean up my messy Excel files. Github’s Copilot makes writing Python code even easier.)

But each of these was new for me at one point, not very long ago! So the only way to keep reaching towards the future is to find new things to try every week.

How do I do that? With AI, of course!

Ask AI to bake learning into your task listI created a simple Coda doc to bake AI into my task list—and I’ve published it online for free, so you can use it to find AI learning opportunities related to your own tasks.

Here’s how I got Coda and ChatGPT to help me find the learning opportunities in my current tasks.

1. I added context to my task list.

In addition the column listing each task I have to do this week, I created columns that challenge me to specify my goal, obstacle(s) and ideal coach on each task; I used a formula to add that context to each task. The result is a task list that looks like…

• My task is to send new story pitches and my goal is to get story assignments but my obstacle is need focused time to think through best angle. I wish I had a top-notch business news editor who could help me.
• My task is to create outline for my upcoming presentation and my goal is to deliver a great and compelling AI talk but my obstacle is I have too many examples and it overwhelms the audience. I wish I had a top-notch speaking coach who could help me.

You don’t have to use Coda to do this. You just need to annotate your task list so you can tell an AI what you want to achieve, where you’re stuck, and who could help.

2. I created a ChatGPT prompt.

Once I had my annotated task list, I pasted it into ChatGPT with this instruction.

You are an AI productivity coach and trainer who gradually expands the AI savvy of the clients you work with. You are working with Alexandra Samuel, who already uses Perplexity, tome, chatgpt, Coda.io, scite.ai, Ideogram, Claude and Github Copilot (to write basic python scripts). Review her task list for the week and suggest 5 possible AI use cases that would extend her skill set by using AI in an innovative way—while helping her get through her task efficiently. In each case, provide SURPRISING uses she wouldn’t think of, along with a very specific first step (like a prompt) that is a little beyond her current knowledge or skill set.
The results were only OK, so I pushed the AI a little:

OK you don’t have to suggest a different tool for each task….Please try again with 5 more suggestions based on the same task list, but choosing the best tool for each task even if that means repeating tools. Also provide a little more creative thinking and detail.

3. I got suggested AI experiments from ChatGPT.

For example…

• Task: Create outline for my upcoming presentation.
• AI Tool: ChatGPT.
• Creative Use: Develop an interactive presentation outline in a dynamic format that includes simulated audience feedback to refine content. Use ChatGPT to generate questions and reactions from a virtual audience based on your presentation’s main points.
• First Step: Outline your presentation’s main points using ChatGPT. Then, simulate a Q&A session where ChatGPT generates typical audience questions and critiques. Use this feedback to adjust and prioritize your content effectively.

I can’t wait to give that a try! You can see more of ChatGPT’s suggestions in this transcript of my session.

Using AI as your AI coach—and tying that coach to your task list—is the best way to continually expand your AI skills and use cases.

And that’s how we’ll get enough AI-savvy humans to ensure this forthcoming age of miracles and wonders really is shared by all.

This post was originally featured in the Thrive at Work newsletter. Subscribe here to be the first to receive updates and insights on the new workplace.