Leading mathematicians fear AI is making their field dumber, and warn the rest of us is next
decoder ·
Twenty-five Fields Medal winners have signed a joint statement warning that the goals of the AI industry and mathematics are "severely misaligned," arguing that mass-producing AI-generated solutions to problems could undermine the discipline's real purpose of conceptual understanding. They note that large language models have recently become capable of cracking major outstanding problems across many areas of mathematics, and that AI companies treat such problems as benchmarks to conquer. The signatories argue that famous unsolved problems function as landmarks for measuring improved understanding, and that when one is solved, the new thinking it required matters more than the answer itself—thinking that mathematicians normally unpack over years of talks, discussions and simplifications. AI threatens to short-circuit this process, with solutions announced without proper writeups, isolation of new methods, or citation of prior work, raising severe attribution and plagiarism questions. The statement comes amid a controversy involving two mathematicians and OpenAI, which is accused of trying to beat researchers to a partial solution of a Millennium Prize Problem for publicity. The signatories describe a broader threat to intellectual work, since training in many fields develops understanding and the ability to formulate new questions, not just final answers—aims that come apart when AI produces results directly. They do not call for a ban, saying AI could enhance genuine mathematical study, but insist that whether the changes benefit or damage the field depends largely on decisions by the humans controlling the technology, and that these issues must be addressed urgently.
decoder