De Branges' failed 1964 proof invoked in OpenAI math manuscripts debate
Michael_Druggan · x · 2026-10-11
Invoking a historical precedent, MichaelDruggan quoted the story of Louis de Branges' 1964 claim to have proved the invariant subspace problem: dozens of mathematicians verified the 70-page manuscript line by line — Paul Halmos worked on it for two months before asking to leave the refereeing committee — and an unresolvable error was eventually found.
Druggan argues the difference is that 'the vast majority of OpenAI's results are correct,' even offering a bet; littmath counters that many are 'not exactly proofs' though most could be converted with effort, and notes de Branges was available to answer questions while the internal model is not. The exchange highlights math community concerns over verifying AI-generated proofs.
Related event: OpenAI's 722-Paper Math Dump Sparks Fierce Debate in Academia(32 posts)→
More from Research
- All of Science embedded and free: semantic search for your agents, 10 req/min — IgorCarron · 2026-10-11
- dioscuri: 2027 will be the year AI does for science what it did for math — dioscuri · 2026-10-11
- Tsinghua's AAArena finds AI agents hit rank 1 with replays but stall on complex games — rohanpaul_ai · 2026-10-11
- 100 Claude agents no better than 10: Ramez Naam shows agent scaling tops out fast — zsakib_ · 2026-10-11
- Turing Post's 2026 guide: which LLM benchmarks to use for reasoning, coding, math and agents — TheTuringPost · 2026-10-11
- OCR benchmark fairness: postprocessing and formatting conventions skew scores — VikParuchuri · 2026-10-11