Terence Tao Clarifies Navier-Stokes Rumor Was Misread Hypothetical; Clay Still Lists Problem Unsolved
机器之心 · wechat · 2026-09-06
Machine Heart fact-checks the 'Claude solved Navier-Stokes' rumor: it originated from Terence Tao's explicitly hypothetical Mastodon posts on Sept 3 imagining an autonomous AI producing a massive Lean-verified proof, and discussing how to preserve intermediate mathematical insight. Speculation that Tao had insider knowledge led to Curran's 'prediction,' which drew 2.5M+ views. Tao clarified he knows of no major new progress on N-S; the scenario was hypothetical, though increasingly realistic given AI's pace.\n\nThe piece reviews real recent AI-math milestones: Claude's 13M-line Lean formalization of Fermat's Last Theorem in 11 days (30k machine-verified theorems); an unreleased Anthropic model raising the Riemann zeta zero bound from 41.6% to 67.2%; OpenAI disproving an Erdős unit-distance conjecture plus ten other math/TCS results. AI's impact is shifting from 'how many problems it can solve' to rising importance of formal verification (Lean) and a new division of labor — humans moving toward problem selection, conjectures, and theory distillation, what Tao calls 'big math.' The rumor itself remains unverified social-media speculation.
Related event: Claude Did Not Solve Navier-Stokes, Clarifies Terence Tao(2 posts)→
More from Models
- Insider teases that next week's demos will far outshine OpenAI's official blog and trailer — ChrisGPT · 2026-09-06
- GPT-6 Astra system card: first OpenAI model to hit Critical cybersecurity threshold — RyanGreenblatt · 2026-09-06
- ChatGPT usage-reset cards only extend the date, users find late use a bad deal — dotey · 2026-09-06
- GPT-6 Astra builds browser games with three.js, even modeling cars in Blender — gaganghotra_ · 2026-09-06
- Dev feed consensus: GPT Astra clearly outperforming Fable 5.1 — gaganghotra_ · 2026-09-06
- OpenAI insider: 'Astra genuinely shocked me' and 'ARR doesn't matter anymore' — Yuchenj_UW · 2026-09-06