Why won't AI simplify old proofs? Researchers spar over elegance in AI for math
thegautamkamath · x · 2026-10-06
- Mathematician Gautam Kamath flags an overlooked gap: AI-for-math efforts focus almost entirely on producing new results, while almost nobody uses AI to find simpler, more elegant proofs of old theorems or clean up the literature.
- Su Chen Zang reposts with a pessimistic take: getting AI to produce "elegant maths" is as hard and futile as cleaning up code slop to be less verbose, less redundant, more human-readable.
- Her core argument: models won't pursue elegance unless it improves end-to-end performance — "bitter lessons have bitter aftertaste."
- The exchange highlights a mismatch between current AI optimization goals and mathematical aesthetics.
Related event: Professor Calls for Using AI to Find Simpler Proofs of Old Theorems(8 posts)→
More from AGI Musings
- Lenny interviews OpenAI Codex lead Thibault Sottiaux: why PMs will thrive in the AI era — lennysan · 2026-10-06
- Half of social science job postings now want AI researchers — RexDouglass · 2026-10-06
- NYT Opinion: Ada Lovelace Already Answered the Big Questions About AI — nytopinion · 2026-10-06
- As AI does the thinking, will 'brain gyms' become the new gyms? — birchlse · 2026-10-06
- threepointone: If intelligence gets too cheap to meter, repos will replace docs and specs — threepointone · 2026-10-06
- Researchers puzzled by absence of an org doing open alignment science — sebkrier · 2026-10-06