Kevin Buzzard's sphere PR vs an AI-coded Ricci flow formalization sparks a Lean math debate
JasonRute · x · 2026-10-19
Jason Rute highlighted an apparent paradox: Kevin Buzzard filed a mathlib PR on Sept 26 proving the unit sphere is a Riemannian manifold, while a full Ricci flow formalization appeared Sept 28. His quip: one is mathlib, the other is vibe coded — and it could even be the other way around.
The thread grew into a debate on the state of Lean formalization:
- eigenstate asks whether there is actually a usable differential geometry library in Lean, and which workflows are viable — e.g. quickly formalizing research-level differential geometry statements with confidence.
- jacobaustin132 argues Tau Ceti covers gaps mathlib lacks, but AI can vibe up its own foundations faster than either library evolves; he predicts hobbyists burning Claude tokens on Prove2Me/Tau Ceti will formalize most major results within 3 months, after which AI can leverage those finished repos.
- The pair also clashed on whether we need to understand what AI actually proves, versus trusting a vibe sense that the AI is smart enough not to prove something wrong.
More from Research
- Embedding arithmetic demo: cartoon − animation − comic = diagram — mayfer · 2026-10-23
- Domingos Cites New Result Claiming RL Is 'Basically Useless' vs Supervised Learning — pmddomingos · 2026-10-23
- Wolfram Opens Rule 30 Prizes to AIs: $30,000 for Three Open Problems — cephaloform · 2026-10-23
- Stanford and Databricks use AI agents to reproduce medical findings and test analytical robustness — ChengleiSi · 2026-10-23
- EMNLP paper: diffusion LMs internally encode latent denoising timestep representations — gsarti_ · 2026-10-23
- KU Leuven and VIB open two Group Leader professorships in AI for Biology, start 2027 — jpillowtime · 2026-10-23