25 Fields Medallists, including Terence Tao, warn that AI benchmark-chasing in mathematics is misaligned with science

A public declaration announced by Terence Tao and signed by 25 Fields Medallists — including Peter Scholze and Manjul Bhargava — argues that AI companies' push to solve mathematical problems as benchmarks is misaligned with conceptual understanding and could…

Twenty-five Fields Medal-winning mathematicians, including Terence Tao, Peter Scholze, and Manjul Bhargava, have published a public statement (described as a 'declaration' in the earlier report and an 'essay' in the later one) arguing that AI companies' push to solve mathematical problems as benchmarks is severely misaligned with the goals of science and the mathematical community. The signatories warn that rushed, headline-driven releases of LLM solutions to major problems raise attribution and plagiarism questions and bypass the human process of integrating results into the mathematical canon, potentially eroding how mathematical ideas are developed, transmitted, and conceptually understood. They frame the issue as a broader misalignment between AI outputs and the purpose of intellectual work, affecting scientific and creative professions and society at large, and call for urgent action by AI developers, the mathematical community, and society. The declaration is posted on a public page and invites further signatures, in the manner of the Leiden declaration, and has been covered by The Economist. The two reports agree on the signatory count (25, all Fields Medallists), the named signatories, and all core claims; the named co-signatories Scholze and Bhargava and the urgent-action call come only from the later report, while the public posting page, Leiden comparison, and Economist coverage come only from the earlier one.

• Declaration announced by Terence Tao and signed by 25 initial signatories, all Fields Medallists; both reports agree on the count of 25.
• Named signatories: Terence Tao (both reports); Peter Scholze and Manjul Bhargava (later report, 17:45:12Z, only).
• Core claim: AI companies' push to solve mathematical problems as benchmarks is misaligned with conceptual understanding and the mathematical community's goals.
• Concerns: rushed, headline-driven releases of LLM solutions to major problems raise attribution and plagiarism questions.
• Signatories say mass-produced AI solutions could bypass integration of results into the mathematical canon and erode the human process that develops and transmits mathematical ideas.
• Framed as a broader misalignment between AI outputs and the purpose of intellectual work, affecting scientific and creative professions and society at large.
• Calls for urgent action by AI developers, the mathematical community, and society (later report only).
• Statement is posted on a public page; further signatures are invited in the manner of the Leiden declaration (earlier report, 17:30:26Z, only).