So, this happened.
The Times is reporting that 41 mathematicians and Fellows of the Royal Society have sent a letter to its president warning that the pace of AI development is becoming an “emergency.” The names on it include Fields Medal winners Peter Scholze, Wendelin Werner, Timothy Gowers, Martin Hairer, and James Maynard. So this is not exactly a fringe group.
A big part of the concern is math. The letter points to how quickly AI is getting better at serious mathematics, including OpenAI’s recent claimed Navier Stokes breakthrough, and asks the Royal Society to push the issue further into government and public debate.
At almost the same time, Anthropic’s Claude was used to help find elliptic curves of rank at least 30 and 31, beating the previous record of 29. Getting from rank 28 to 29 took mathematicians about 18 years. Then the record moved twice in a matter of days.
It does not mean Claude independently solved everything. Humans were involved. Obviously. But 18 years versus a few days is hard to ignore, and it helps explain why some mathematicians are getting increasingly uneasy about where this is going.
I get the concern. Where I start to lose them a little is the government part.
But once you bring government regulation into it, things can get pretty stupid pretty fast. Slowing down AI sounds simple until somebody has to decide what “too fast” actually means, which systems get restricted, and who gets to make those kinds of decisions. And I am not exactly sure that governments are going to figure this out faster than the technology is moving
Maybe regulation ends up being part of the answer. I have no idea. I am just not convinced that handing the problem to politicians suddenly makes it less of a mess.
Also, how many of these letters are we getting? We had the Mathathon letter, the Fields Medalists letter, and now this one. At this point, the AI warning letters are starting to pile up.
Sources:
The Times: Mathematicians fear AI curbs may come too late to save humanity
International Mathematical Union: Fields Medal
OpenAI: On the Navier-Stokes Millennium Prize Problem
anothernews.io/news/claude-math-record
Elliptic Curve Rank Leaderboard: Rank 31 curve
History of elliptic curve rank records
More from Dogmathic
For more math news, explanations, proofs, examples, and commentary, check out Dogmathic Math Notes and the latest Dogmathic videos. You can also grab free PDFs and LaTeX resources or sign up for occasional email updates when something worth sharing gets published.
Leave a Reply