RL Pioneer Csaba Szepesvari Grills Anthropic on Claude's Fermat Proof
CsabaSzepesvari · x · 2026-09-05
RL pioneer Csaba Szepesvari publicly pressed Anthropic on how we can know the formalization behind Claude's proof is correct, asking for an honest account of verification efforts and their limits.
Context: Anthropic said Claude last month completed the first formalized proof of Fermat's Last Theorem in Lean — a project experts expected to take years and the largest Lean project to date.
More from AGI Musings
- Aligned vs. Monitorable: An X Debate Over Whether CoT Monitoring Is Necessary — sandersted · 2026-09-05
- China may join US AI safety talks; GPT-6 first model rated Critical cyber risk, newsletter finds — gleech · 2026-09-05
- JeffLadish: even AI insiders haven't internalized that AI swarms will dwarf us — HaydnBelfield · 2026-09-05
- davidad cites pandemic panic suppression as caution against hiding AI truths — davidad · 2026-09-05
- davidad on infohazards: don't suppress discussion of impending AI upheaval — davidad · 2026-09-05
- Even AI insiders haven't grasped that AIs will collectively surpass humans soon — JeffLadish · 2026-09-05