As of publication, twenty-five mathematicians awarded the Fields Medal — including Terence Tao, Peter Scholze, Maryna Viazovska and James Maynard — have signed an open letter warning that artificial intelligence (AI) could erode a core aim of mathematics: human understanding.
What triggered the protest?
Signatories were particularly alarmed after OpenAI announced, a few days earlier, that it had solved the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems designated by the Clay Mathematics Institute in 2000. That problem carries a $1,000,000 prize and lacked a generally accepted proof for decades.
According to OpenAI, it deployed roughly ten thousand AI agents to attack the problem, completed the work in 88 hours, and published a 166-page proof. OpenAI also said it would not claim the Millennium Prize. The company estimates the baseline cost of the effort at about $10 million; the report also refers to OpenAI as “nearly a one-trillion-dollar” company.
Disputes and counterclaims
The announcement prompted further controversy. Two well-known mathematicians claimed they had been working on the same problem and that information about their progress reached OpenAI; OpenAI denies accessing any user data in its work on Navier–Stokes and says its proofs differ substantially from the other teams’ approaches and even from their precise results. The Millennium Prize authority has not yet verified OpenAI’s result.
However, the letter’s authors say their primary concern is not this specific dispute but a broader conceptual risk: AI may generate mathematical results faster than the human community can assimilate and understand them.
The role of understanding in mathematics
The letter argues that mathematics values not only results but the path to those results. That path yields conceptual understanding, raises new questions, and often creates new research directions. If AI merely produces proofs, the underlying ideas and intuitions that fuel subsequent human-driven inquiry may remain obscure, potentially stunting long-term progress.
Terence Tao, one of the signatories, summarized the issue as: “We have become too good at optimization.” Hugo Duminil-Copin used an analogy: dropping someone from the air onto Mount Everest’s summit is very different from climbing the mountain — reaching the summit does not substitute for the knowledge gained in the climb.
Signatories also note that OpenAI’s 166-page proof contains relatively few of the explanatory, pedagogical accounts mathematicians typically provide when introducing a new discovery. Thus it is unclear whether the core human-understandable idea behind the proof will become visible.
Counterarguments: enduring appeal and cumulative work
Not everyone agrees with these concerns. Alexander Paseau, a mathematics professor at the University of Oxford, told The Guardian that mathematics will not lose much of its appeal even if AI outperforms humans in research. Paseau argued that people will still want to understand mathematics and appreciate the beauty of proofs. He also emphasized that many AI-derived results build on human work: for example, OpenAI’s breakthrough allegedly relied heavily on the work of Madrid-based mathematicians Diego Córdoba and Luis Martinez-Zoroa.
Consequences and a call to action
The letter’s authors stress that the ultimate impact of these technologies will largely depend on decisions made by the people and institutions that govern them. They therefore call for urgent dialogue among mathematical communities, companies developing AI technologies, and society at large so that AI’s integration into mathematics supports human understanding rather than replacing it.
The debate continues: some mathematicians fear AI will disconnect mathematical discovery from human insight, while others believe that the desire to understand and the appreciation of mathematical beauty will persist even as machines outpace humans in certain tasks.
Key figures and facts
- Signatories: 25 Fields Medalists, including Terence Tao, Peter Scholze, Maryna Viazovska and James Maynard.
- Problem: Navier–Stokes existence and smoothness, one of the Clay Mathematics Institute’s seven Millennium Prize Problems (designated in 2000).
- Prize: $1,000,000 (Millennium Prize).
- OpenAI’s claims: used about 10,000 AI agents; completed the task in 88 hours; published a 166-page proof; baseline cost estimated at $10 million.
- Status: The Millennium Prize authority has not yet verified the result; disputes exist about whether OpenAI accessed external researchers’ data and about the relationship between the AI proof and prior human work.
