New arXiv Paper Shows Lean Formalization Agents Deviate from Original Proofs
burny_tech · x · 2026-10-08
A new arXiv paper (2610.08144) reports that Lean formalization agents sometimes follow a different proof route than the informal proof they are formalizing — notably, the Lean formalization of OpenAI's Navier-Stokes paper differs from the original.
Mathematician Scott N. Armstrong boosted the paper, calling the general finding obvious and noting he had explicitly made this point about OpenAI's Navier-Stokes paper in a recent podcast (5:00–7:00).
More from Research
- LIBERO-MAX: mid-task changes cut success rates for all 14 robot policies by 11-25.7 points — DJiafei · 2026-10-08
- SOL: a sample-based distance metric proposed to evaluate diffusion language models — LucaAmb · 2026-10-08
- Top mathematicians debate: is AI exploration a convex hull or a generated subgroup? — gleech · 2026-10-08
- RUC's ME-Decoding uses Mahalanobis distance to preserve semantic diversity in LLM decoding — jiqizhixin · 2026-10-08
- Social contagion and fertility: seeing your boss have kids makes you likelier to follow — EleanorOlcott · 2026-10-08
- KAIST's FastOPD cuts VLA robot inference latency by 78% via on-policy distillation — kaist-ai · 2026-10-08