OpenAI announces three big science discoveries and a new way to work with academics — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI announced three major scientific discoveries—one in mathematics, one in computer science, and one in physics—produced by its AI models, marking a significant expansion of AI’s role in foundational research.
- OpenAI released model output for the local smoothing and Kakeya conjectures directly on GitHub, following its October 6 precedent, but this time advising researchers to develop their own explanations without designated intermediaries.
- OpenAI says its AI refuted the Exponential Time Hypothesis by producing a deterministic algorithm that counts satisfying assignments for fixed-width CNF formulas in subexponential time, with verification led by original ETH proposers Russell Impagliazzo and Ramamohan Paturi.
- David Gross will lead a group including Michael Douglas, Arthur Jaffe, Igor Klebanov, and Nikita Nekrasov to shepherd OpenAI’s physics result—a claimed rigorous existence proof for interacting quantum field theories in four dimensions, including QCD and a unified Standard Model completion.
- OpenAI adopted new scientific release practices by involving external expert 'shepherds' in computer science and physics, responding to input from researchers on responsible dissemination of AI-generated discoveries.
- AGMAI, OpenAI’s math advisory panel, recommended against exclusive early access to AI outputs, citing unequal access and biased selection, and urged future releases to include Lean-formalized proofs and public funding for interpretation work.
Why it matters: The involvement of original ETH architects in verifying OpenAI’s refutation gives the claim rare credibility, while the shift to open, expert-guided validation changes how AI-driven science may be governed—directly impacting researchers, funding priorities, and the reproducibility standards of AI-aided discovery.
Ask SkimNews
