A group of mathematicians, including 25 Fields Medalists, has issued an open letter expressing concern that the movement by AI companies to promote mathematical problem-solving as a benchmark is detrimental to mathematics as a discipline and its community.
Mathematicians Protest AI Use in Mathematics, Issuing Open Letter Following OpenAI Announcement
This article is a translation. Read the Japanese original
This movement follows OpenAI's announcement on September 8, 2026, that it had solved the Navier-Stokes equations—one of the unsolved problems in mathematics—using an undisclosed model. Mathematicians have raised concerns regarding the 166 page paper published by OpenAI, noting that it contains little of the detailed explanation typically used by mathematicians when sharing new discoveries.
In the open letter, the mathematicians pointed out a "profound misalignment" between the goals of AI companies and the mathematical community. They warned that the AI industry's focus solely on problem-solving and the high-speed execution of massive amounts of truth-testing may neglect "conceptual understanding and insight," which is the core of mathematics, and could destroy the fertile ground for generating new ideas.
Furthermore, concerns were raised that the premature announcement of AI-generated achievements could make processes such as citing previous research and writing papers difficult, potentially leading to issues regarding intellectual attribution and plagiarism. While acknowledging the need to adapt to the evolution of AI, the mathematical community emphasized that whether this development brings benefits or destructive impacts depends heavily on human decision-making, calling for a collective response from society.
In response to these concerns, it has been reported that OpenAI has decided to withdraw its sponsorship of the Caltech Mathathon. Additionally, the Mathathon organizers have acknowledged the concerns and updated their official explanations, such as making the submission of papers to arXiv a requirement for the 2 round of applications.
Sources
- AIベンチマークに数学を使うのは有害だとして数学者たちが反発 （GIGAZINE、2026-09-15）
- Declaration — Math and AI