Insider claims OpenAI's math breakthrough relied on NYU and Anthropic researchers using Claude
eyishazyer · x · 2026-09-08
A viral thread claims OpenAI may have solved one of math's hardest problems, but the story behind it is messy. NYU mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge spent a year on fluid dynamics equations: Euler, Boussinesq, and a Navier-Stokes variant.
They used AI heavily — Claude and Codex — to generate candidate proofs, then verified and rewrote them by hand. The post questions who really deserves credit: the two outside researchers or OpenAI. Details are unverified, but it showcases a new AI-assisted mathematical research workflow.
More from Fun
- When AI paper acknowledgments ran for pages — a researcher's nostalgia — DynamicWebPaige · 2026-09-08
- Google's Astra spontaneously picks Ruby to edit a file in live demo — xeophon · 2026-09-08
- We solved Navier-Stokes, but nobody can read the 60 pages of Lean — tarantulae · 2026-09-08
- Dev builds a bookmarklet to tag "AI slop" on any website, not just LinkedIn — philnash · 2026-09-08
- AI-generated puppy fixates on your coffee in a funny demo clip — beholdersai · 2026-09-08
- AI community meme: renaming 'paradigma' to 'moggingface' — tensorqt · 2026-09-08