Mathematicians Criticize OpenAI's Mass Release of AI-Generated Proofs
OpenAI released more than 700 AI-generated mathematical proofs, prompting strong objections from mathematicians and an advisory group that warned against testing advanced problems on proprietary models.
OpenAI announced the publication of over 700 files containing AI-generated proofs for a range of mathematical problems. The release follows earlier controversy surrounding the company's AI agents and a claimed solution to the Navier-Stokes equations.
The Association for Human Mathematics issued a statement rejecting the claim that the release advances the discipline, calling the mass publication "not a demonstration of scholarship, but a demonstration of power." The statement was shared by Fields Medalist Terence Tao, who urged both mathematicians and the public to view the approach with skepticism.
OpenAI said it had consulted an independent Advisory Group on Mathematics and Artificial Intelligence to improve how it shares results with the research community. However, the advisory group later released a statement saying it does not endorse testing advanced mathematical problems on proprietary models and asked OpenAI to cease such testing.
The association criticized OpenAI for ignoring the advisory group's advice, alleging that the company disregarded standard scientific norms that ensure trustworthiness and public interest in mathematical research. While the proofs have been verified using the Lean programming language, experts note that many are difficult for humans to interpret, leaving their practical value uncertain and placing additional verification work on mathematicians.