Twenty-Five Winners of Mathematics' Highest Prize Warn That AI Companies Are Treating Their Field as a Scoreboard

In a joint declaration published today, 25 winners of the Fields Medal, the top honor in mathematics, say the race by AI companies to solve famous math problems as proof of their programs' power is "detrimental to the science of mathematics." They are not asking for the tools to stop. Their argument is that solving problems was only ever a way to reach understanding, and that hurried AI answers, announced without proper write-ups or credit, risk breaking the chain by which ideas pass from one mathematician to the next.

Twenty-five mathematicians who have won the Fields Medal, the prize often described as mathematics' Nobel, published a short declaration today titled "A Severe Misalignment of AI in Mathematics." The signers include Terence Tao, Peter Scholze, Maryna Viazovska, Pierre Deligne and Yu Deng, who received the medal this year. It appears on a new website, mathandai.org, with a French version in the newspaper Le Monde, and it invites other mathematicians to add their names. Tao wrote on his blog that the statement "grew out of discussions between ourselves over the last week" and was released quickly because of "the urgency of the situation."

The declaration starts by conceding the thing the AI companies say: "Over the last few months, the mathematical capabilities of LLMs have improved dramatically, to the point that they can solve major outstanding problems in many fields of mathematics." Its complaint is about what the companies do with that ability. "The push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics, and to the mathematical community," it says. A benchmark is a test used to score and compare AI programs; the signers' point is that famous unsolved problems have become that kind of test. The declaration names no company, but it arrives three days after OpenAI announced that a system of its AI agents had produced a proof of one of the seven famous Millennium Prize Problems, a claim mathematicians are still checking.

The heart of the argument is about what a solved problem is for. Famous problems, the declaration says, "have often served as landmarks and lighthouses" for measuring understanding. When a person solves one, the proof is talked through, simplified and argued over for years, until it becomes something a student can learn from a textbook. "But solving problems is only a tool and proxy for achieving the primary goal of conceptual understanding and insight," the signers write. "Forgetting this in the world of AI may turn the tool against the primary goal." They worry that "the mass production at faster and faster pace of 'true/false' statements could destroy fertile ground instead of breathing life into new ideas," and that solutions "announced in a rush, leaving no time for a proper writeup" or for "citing relevant previous work of others" raise "severe attribution and plagiarism questions." Without people willing to do the slow work of absorbing an AI's ideas, they say, "the crucial human transmission chain between mathematicians would be lost."

The signers think their field is a preview. "The mathematical community functions, in many ways, as a miniature version of humanity," they write. In many kinds of work, years of training were never only about producing the final answer; they were how a person learned to understand and to ask the next question. When an AI produces the answer directly, those two purposes come apart. The question they pose is one for everyone: "how to make sure that, as AI changes the way work is done, we do not lose sight of what that work was meant to achieve in the first place."

This is not a call to stop. The declaration says AI "offers the potential of enhancing and accelerating genuine mathematical study and understanding," and Tao himself has been one of the field's most visible experimenters with these tools. The companies' view, stated with each new result, is that solving a problem no human could is the clearest possible evidence that their programs can do real work, and there is no dispute in the declaration that the problems are being solved. What the signers ask is that the companies, the mathematicians and the wider public treat the answer as the beginning of the work rather than the end of it. The credit dispute that already surrounds this week's claimed proof is the very thing they are describing.