Link to original article

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Argument Against Impact: EU Is Not an AI Superpower, published by EU AI Governance on January 31, 2022 on The Effective Altruism Forum. One objection to working on AI governance in the European Union (EU) is that the EU is not a superpower in AI. The EU is not currently leading AI development and will probably not lead AI development in the foreseeable future either. According to the objection, if AGI gets developed elsewhere, then EU AI governance will have little impact on ensuring that it gets developed safely and is aligned with human values. As a caveat to this post, it is not entirely clear that the EU is not a superpower in AI even when it is not leading the rankings or that it is not relevant to AGI development even when it is not the leader in AGI investments. Finally, there could be other pathways to impact besides direct AGI development. Top AI companies in the world are US and Chinese-based. Top companies explicitly trying to develop AGI are all US-owned. According to Disfold, among 30 of the best funded AI startups in the world, 16 are in the US, 9 in China, 3 in the UK, 1 in Canada, and 1 in Singapore. “Of the 2,451 AI new companies that Statista reports as of 2018, 675 have a place with European nations (the UK had 245).” According to McKinsey, “Europe attracted only 11 percent of global venture capital and corporate funding in 2016, with 50 percent of total funds devoted to US companies and the balance going to Asia (mostly China). That share was about the same in 2018. Only four European companies are in the top 100 global AI startups.” While funding for startups in Europe seems to be rapidly increasing, with total investments surpassing €120bn in 2021, triple the value of 2020, the regional gap in top AI-companies is unlikely to be significantly reduced in the near future. According to the Center for Data Innovation’s 2021 report, “China’s AI capabilities relative to the EU and the United States have improved in several ways. First, China has surpassed the EU as the world leader in AI publications. Second, the quality of its AI research has generally trended upward year to year. Third, its software and computer services firms have increased their R&D spending. Fourth, China now has nearly twice as many supercomputers ranked in the top 500 for performance as the United States—the United States led in this indicator as recently as 2017. Finally, China likely continues to lead in the amount of data generated. Overall, however, China has not significantly reduced the gap in AI between itself and the United States, but its trend of consistent progress could eventually evaporate the U.S. lead.” According to the Global AI Index, the United States of America is leading AI development in the world followed by China on the benchmarks of talent, infrastructure, operation environment, research, development, government strategy, and commercial. European countries such as the UK, Germany, the Netherlands and France are in the top 10. This ranking doesn’t consider the EU on the whole, however. According to the AI Index 2019, Europe has consistently been the largest publisher of AI papers, but recently China has surpassed it and has been growing the most from 2000 to 2018. The US is third in terms of number of AI papers. China has 28% of AI publications, Europe has 27% of AI publications, and the US has about 19%. The citation impact of US publications is about 50% higher than China’s. North America accounts for over 60% of global AI patent citation activity between 2014-2018. These data do not distinguish between the EU and the rest of Europe. According to the Global AI Talent Tracker, “the United States has a large lead over all other countries in top-tier AI research, with nearly 60% of top-tier researchers working for American universities and companies.” And “...