Mathematicians confront AI’s rapid advance

Cornell mathematician Steven Strogatz says AI is rapidly reshaping research mathematics, turning a field built around human insight into one where corporate labs can chase major breakthroughs with machine-driven systems. OpenAI said it used tens of thousands of agents to solve a 90-year-old math problem tied to a $1 million prize, though the result still needs independent verification.

The announcement has intensified a credit dispute. The solution builds on work by Diego Córdoba and Luis Martínez-Zoroa, while Tristan Buckmaster says OpenAI moved quickly after learning of related work he had done with Anthropic researcher Levent Alpöge and tried to influence who received credit. Strogatz framed the episode as part of a broader race among AI labs seeking attention before large public offerings.

Recent results from Anthropic and OpenAI have deepened the anxiety. Anthropic said Claude proved 29,500 small theorems while formalizing an existing proof of Fermat’s Last Theorem, and OpenAI reported progress in 10 other long-standing problems in August. Strogatz and collaborator Alex Townsend see AI as both democratizing and destabilizing: it can accelerate hard research, but may leave human mathematicians serving mainly as interpreters of machine-generated proofs.