Mathematics has changed. Let's act like it

340 points · 428 comments on HN · read original →

Points and comments are a snapshot, not live.

AI has changed mathematics, requiring new evaluation and hiring practices.

OpenAI released over 700 research papers on October 6, 2026, proving AI can make groundbreaking mathematical contributions, though many papers are poorly written and some wrong. Frank Calegari argues mathematicians should not mourn but adapt: evaluate all work as if AI-assisted, emphasize interviews in hiring, demand more from journals, and reward exploratory work over mere proof production. He draws parallels to the classification of finite simple groups, which cooled interest in group theory. Calegari views AI as a tool to pursue bolder, broader questions, not merely solve existing conjectures.

What commenters are saying

Dominant sentiment is skeptical acceptance: commenters debate whether AI will replace human mathematicians or simply become a powerful tool. A key split emerges between those who argue AI can advance any problem better than humans (citing trillions in investment and low-hanging fruit) and those who emphasize human choices-which problems to value, paradigms shifts, and social construction of mathematics. Several commenters note that AI proofs in Lean may push corner cases not well tested, and that the real value of mathematics often lies in machinery and understanding, not just results.