AI News · Models ·
OpenAI releases AI-generated results in math, computer science and physics
OpenAI announced new results produced by its models in mathematics, computer science and physics, weeks after releasing more than 700 pages of manuscripts on 372 major math problems. One new manuscript presents a proof that the Exponential Time Hypothesis is false, a claim that outside experts are now examining.
Key points
- OpenAI says its models produced new results in mathematics, computer science and physics.
- One manuscript claims a proof that the widely accepted Exponential Time Hypothesis is false.
- OpenAI will not seek academic credit; outside expert groups are reviewing the results.
What happened: OpenAI announced a new set of discoveries in mathematics, computer science and physics that it said were generated by its models, Engadget reported. The release follows manuscripts published weeks earlier that ran to more than 700 pages of solutions and progress on 372 major math problems. The company has not disclosed which models produced the new results.
The details: New manuscripts posted on GitHub include one establishing Sogge's local smoothing for waves in every spatial dimension, a conjecture about solutions to the wave equation. Another presents a proof that the Exponential Time Hypothesis is false. The hypothesis is widely believed to be true and underpins hundreds of papers in areas such as databases, computational biology and geometry. If it is false, better algorithms for problems in those fields could in theory exist. The scientists who proposed it are part of the group examining the result. In physics, the report describes a proof of interacting quantum field theories with gauge fields and fermions in four dimensions, but the group handling that work has not yet published anything.
Background: OpenAI used a different release approach for each field. Math results were published for researchers to examine independently, computer science results were explained by outside experts in their own writeups, and physics materials were left to the examining group. OpenAI said it will not seek academic credit for its models' work and is asking for feedback on how to broaden participation.
What to watch: Independent confirmation of the results was not reported, and the physics findings remain unpublished.
Our take
Claims of model-driven research are moving from benchmarks to expert scrutiny, so teams should weigh vendor claims against independent confirmation.