Did OpenAI solve the wrong Navier-Stokes problem?
Two weeks ago OpenAI announced that an internal large language model had produced a proof that satisfies the Navier‑Stokes problem as defined by the Clay Mathematics Institute, a result that would earn the $1‑million prize. The proof relies on a formulation that includes an optional external force, a choice many mathematicians describe as unnatural and disconnected from the core question about…
Key points
- openai’s LLM‑generated proof earned the $1‑million Clay prize but targets a variant with an external force.
- three mathematicians posted a proof last Thursday showing the method cannot extend to the full problem.
- on september 7 two mathematicians used AI to produce a blowup; openai completed a similar result less than a day later.
Luis Silvestre of the University of Chicago says the original Clay problem remains unsolved, and three mathematicians posted a proof last Thursday showing OpenAI’s approach cannot be extended to the full, force‑free version. Diego Córdoba notes that all known fluids experience some external force, while Luis Martínez‑Zoroa points out that most research focuses on the scenario without one. On September 7 two other mathematicians used AI to generate a blow‑up for a frictionless fluid, and OpenAI produced a similar result less than a day later. Gonzalo Cao‑Labor a says the situation is “really uncertain at this moment” and that the community’s view of OpenAI’s contribution will depend on how the external‑force loophole is judged.
The story so far
2 episodes →- Did OpenAI solve the wrong Navier-Stokes problem?this story
Did OpenAI solve the wrong Navier-Stokes problem?
scientificamerican.com · 22 September 2026
Loading the full article…
This text was published by scientificamerican.com and written by Joseph Howlett. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Hacker News discussion · 51 pointsnews.ycombinator.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- John Platt describes Google’s Empirical Research Assistance tool for scientific scoring · 1 src
- Anthropic reports Claude leads 26% of its AI research and development tasks · 5 src
- Generative AI designs de novo thiolation domains that boost nrps yields up to ~3‑fold · 1 src
- Google DeepMind uses 7‑hour role‑playing game to explore AI’s impact on science · 1 src
- Opinion: big tech’s AI cancer‑cure promises outpace proven progress · 1 src
Comments
via GitHub Discussions