OpenAI Reportedly Sitting on Hundreds of Unreleased Math Proofs; Navier-Stokes Proof Cost ~$15M in Compute
新智元 · wechat · 2026-09-17
UT Austin professor and ex-OpenAI alignment researcher Scott Aaronson reports AI labs are withholding solutions to 'very significant problems' after hostile reactions to the Navier-Stokes proof. NYU's Scott Armstrong says OpenAI has sat on 'hundreds' of proofs since July's ICM.
Key details: after hearing two Millennium Problems were solved, OpenAI fed all unsolved ones to an internal model stronger than GPT-6 Astra. 100 agents solved the Euler equations variant in 50 hours; 10,000 concurrent agents cracked Navier-Stokes in 88 hours, with 17 more hours for Lean verification — 130B output tokens, $15M in compute for a 166-page proof almost no human has read.
Terry Tao warns AI compute can flatten human projects before they mature, chilling open science; 25 Fields medalists signed a letter criticizing labs for treating math as benchmarks. Priority, credit, and peer review — centuries-old norms — are all breaking down.
More from AGI Musings
- Master foundational math to stay irreplaceable, argues mathematician — TivadarDanka · 2026-11-02
- Narayanan: labs racing toward Recursive Self-Improvement may hit Recursive Self-Extinction of their business model — random_walker · 2026-11-02
- Editor fully resigns from JMLR, arguing AI reviews beat rushed humans and journals need a rethink — roydanroy · 2026-11-02
- Alpha School investigation: 17 campuses on probation, Bill Ackman as shadow lead, Holmes involved — benjaminjriley · 2026-11-02
- Google's chief economist: AI will transform more jobs than it kills — gains come from redesigning workflows — danielrock · 2026-11-02
- Daniel Lemire: AI is exploding demand for boring compute, server CPUs to double — lemire · 2026-11-02