Post-Cutoff

Gil Kalai: ‘Be a Grothendieck!’, on AI and mathematics

Gil KalaiBlog

Why it matters

A senior combinatorialist argues against slowing AI-assisted mathematics while naming the risks to the community; listed by Tao on Oct 12.

Summary

Posted Sept 22, 2026 on ‘Combinatorics and more’. Summary from an automated reading: Kalai calls AI in mathematics ‘a major event in our lives, perhaps a crisis, perhaps an opportunity, and probably both’ and writes ‘I don’t recommend any attempt to “slow down” or stop progress in AI-assisted mathematics’. He stresses AI’s use for simplifying proofs and generating problems and conjectures, and asks for open debate and tolerance of differing views. He mentions Chen and Klartag’s ‘Digesting the proof of the sharp thin-shell inequality’ (arXiv 2607.23307), Alexandre Eremenko’s MathOverflow report that several of his open problems were solved with AI in August 2026, Alex Kontorovich’s ICM 2026 plenary and the Advisory Group on Mathematics and AI (agmai.org). Quotes were extracted by a summarising tool and should be re-checked against the page.

Archived text

We have not archived the text of this post yet.

Source: gilkalai.wordpress.com/2026/09/22/be-a-grothendieck-on-ai-and-mathematics

Cited in

  1. Science & math 98 days after the cutoff

    OpenAI releases 722 AI-written math manuscripts claiming hundreds of open problems

  2. Science & math 92 days after the cutoff

    Summer 2026 flood: dozens of named conjectures settled on arXiv with disclosed AI help