Terence Tao published 'A Severe Misalignment of AI in Mathematics' (Sept 11), presenting a declaration co-signed by 25 Fields Medalists (Deligne, Scholze, Villani, Viazovska and others; open signatures at mathandai.org). Core claims: 'the push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics'; solutions are 'only a tool and proxy' for conceptual understanding; mass-produced true/false statements 'could destroy fertile ground'; rushed announcements raise 'severe attribution and plagiarism questions.' Tao is not anti-AI, but outcomes 'will in large part be determined by the decisions of the humans in control.'
25 Fields Medalists declare the goals of AI companies and mathematics severely misaligned
2026-09-11Frontier watchWATCH
Terence Tao published 'A Severe Misalignment of AI in Mathematics' (Sept 11), presenting a declaration co-signed by 25 Fields Medalists (Deligne, Scholze, Villani, Viazovska and others; open signatures at mathandai.org). Core claims: 'the push by AI companies to solve mathematical problems as a benchmark is detrimental to the science of mathematics'; solutions are 'only a tool and proxy' for conceptual understanding; mass-produced true/false statements 'could destroy fertile ground'; rushed announcements raise 'severe attribution and plagiarism questions.' Tao is not anti-AI, but outcomes 'will in large part be determined by the decisions of the humans in control.'
Referencemathandai.org · OverviewReferenceterrytao.wordpress.com · a severe misalignment of ai in mathematics
Imported from the existing research archive. Migration preserves the recorded judgment and is not a fresh verification.
What this means for you
The research community's first coordinated pushback against benchmark-driven math AI, landing the same week as Anthropic's FLT formalization and coverage of mathematicians' objections to OpenAI's methods. For evaluation teams: math benchmarks are contested territory — expect community norms (attribution, writeup requirements) to start gating how AI math results can be claimed; treat 'solved famous problem' announcements as needing community-verifiable artifacts.
FRONTIER WATCH
Public artifacts and validated results are different.
The recorded fact is the public work and its artifacts. Author-reported outcomes do not establish independent reproduction. Next checks include usable code, task conditions and third-party validation.
Evidence & related material
Technical details & scores
| Date | 2026-09-11 via Tao's blog; open signatures at mathandai.org |
|---|---|
| Signatories | 25 Fields Medalists (1978-2026 cohorts) |
| Claims | benchmark-driven solving detrimental to mathematics; solutions are proxies; attribution/plagiarism risks |
| Stance | not anti-AI; outcomes determined by human decisions |
Technical impact 2
Engineering value 2
Adoption signal 4
Maturity 4
Verification cost 1
Risk 1