Researcher: AI and I ran out of ideas on the proof; turning to the Lean-verified one
aran_nayebi · x · 2026-11-21
Researcher Aran Nayebi shares that he tried using AI to fix a mathematical proof, but "at some point me + the AI ran out of ideas." He says it's now more worth trying to understand Levent's proof, since that one has actually been formally verified in Lean.
The post offers a first-hand data point on the current limits of AI in frontier mathematics, amid the ongoing controversy over AI-claimed proofs.
Related event: Riemann proof flaw may be unfixable, even with AI(2 posts)→
More from Research
- Colgate says LLM role-play can generate ~90% accurate survey data, challenging market research — ZeroStateReflex · 2026-11-22
- Continual Causality Bridge returns at AAAI-27, novel-work deadline Dec 1 — mundt_martin · 2026-11-22
- Harvard lab uses Biomni AI agents to decode the hidden "grammar" of DNA — anshulkundaje · 2026-11-22
- PredActor Cuts Diffusion to 5% of Forward Pass for Low-Latency Humanoid Control — chris_j_paxton · 2026-11-22
- Claude's System Card Reveals Protein-Sequence Training, Sparking Debate on Bio Foundation Models — chaitjo · 2026-11-22
- Axiom offers to pay biotechs to run drug programs while they keep 100% of assets — cjmaddison · 2026-11-22