GoKawiilOpenAI claims three math/CS proofs, including disproof of ETH
OpenAI posted results to GitHub including a proof extending Sogge's local smoothing for waves to all dimensions above two, which also yields related Kakeya and Stein restriction estimates. It also released a paper claiming a deterministic subexponential-time, polynomial-space algorithm for counting satisfying assignments of fixed-width CNF formulas, which for 3-SAT would refute the Exponential Time Hypothesis. A group led by Russell Impagliazzo, Ramamohan Paturi, and Ryan Williams will review the ETH claim, and OpenAI released materials immediately at that group's request.
GoKawiil's interpretation of the reporting above, not reported fact.
If verified, a disproof of the Exponential Time Hypothesis would be a landmark result in theoretical computer science, reshaping assumptions underlying complexity theory and cryptography. The decision to invite independent scrutiny before peer review suggests OpenAI is trying to build credibility for AI-generated mathematics while acknowledging such claims require external validation. The wave-equation result, if correct, would close a long-standing open problem in harmonic analysis spanning multiple conjectures.
- OpenAI published two major theoretical results directly to GitHub rather than through traditional peer-reviewed channels.
- A claimed refutation of the Exponential Time Hypothesis is being independently reviewed by a panel of prominent computer scientists.
- The releases reflect a new model for engaging academics to vet AI-generated scientific claims before formal publication.
Source: theverge.com — Richard Lawler, 2026-10-26
Published there as: “OpenAI announces three big science discoveries and a new way to work with academics”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.