altryne challenges Gary Marcus: Lean proofs only verify, they don't generate the results
altryne · x · 2026-10-07
altryne pushed back on Gary Marcus, arguing the Lean proofs are a separate process run to VERIFY the results of the non-Lean agentic loop — not part of how the results were produced. It's a sharp technical point in the ongoing debate over AI math results: informal solving and formal verification should be evaluated separately.
Related event: Gary Marcus and altryne Debate Lean Verification in Math AI(2 posts)→
More from Models
- OpenAI's unreleased model reportedly proved the quasi-Riemann hypothesis with a Lean proof — Dr_Singularity · 2026-10-07
- OpenAI reportedly folded open-problem lists into internal math benchmarks, sparking clarification — tszzl · 2026-10-07
- Blogger challenges OpenAI: when will AI invent category-theory-level math? — burny_tech · 2026-10-07
- OpenAI releases new math results from internal frontier model, claims Unique Games proof — thomasahle · 2026-10-07
- HuatuoGPT-3: 27B open medical LLM hits 70.1 on HealthBench, beating GPT-6 Astra — CUHKSZ · 2026-10-07
- Dev Complains Overnight Long-Running Tasks Keep Hitting Usage Limits — willdepue · 2026-10-07