Mathematicians push back on OpenAI's model-generated proofs
ZeeshanZiaML · x · 2026-10-12
A debate over OpenAI's model-generated mathematical proofs: m4h007 argues the community should refuse to treat them as finished work — whoever claims a proof must present it so independent verification is feasible, otherwise it's just a claim. Boosting, @onionesque says he consulted 20 specialists who found most such proofs unreadable gibberish, comparable to Mochizuki's abc situation. The core tension: mathematical truth doesn't depend on consensus, but acceptance as correct does.
More from Models
- Gadsby-4B Hits Hugging Face: Constrained Decoding Turns Gemma 3 4B Into a Lipogram Model — cephaloform · 2026-10-12
- 50 AI Models Have Played Cowcraft, an Online Limit-Testing Arena for LLMs — djcows · 2026-10-12
- After days of testing, dev ditches sol 6.1 and astra 6 for agentic work, returns to Opus — NERDDISCO · 2026-10-12
- $/M tokens is broken: Opus 5.5 costs ~6x Sonnet per task at similar scores — lordmairtis · 2026-10-12
- Anthropic's Claude can't even do substring search in chat history, user shows — AaronBergman18 · 2026-10-12
- Every Recent Claude Loves to Say Things Are 'Carried' or 'Held' — repligate · 2026-10-12