OpenAI releases nearly 400 AI-generated math results across 700+ manuscripts
OpenAI says 300 of 719 manuscripts have been formalized, about 42 percent, and that it will keep updating the repository.
OpenAI has released nearly 400 AI-generated mathematical results spread across more than 700 manuscripts, publishing the material in a GitHub repository along with guidance on how to navigate it.
The results span combinatorics, several branches of geometry, number theory, theoretical computer science, algebra, topology, probability and statistical mechanics, and mathematical physics, according to OpenAI.
OpenAI acknowledged that the results are at different stages of verification and that many, but not all, of the manuscripts have been formalized. By its own count, 300 top-line results out of 719 manuscripts have been formalized, around 42 percent. The company said it will update the repository with more formalizations as it obtains them.
Mathematicians say the material will take time to absorb
Mathematicians who spoke with The Verge after the release said simply understanding what OpenAI put out could take years. Several said working through the roughly 40-page table of contents and abstracts took them the better part of an hour.
Kevin Buzzard, a mathematics professor at Imperial College London, said he identified numerous theorems in algebraic number theory of which only around six immediately stood out.
Quick answers
Where did OpenAI publish the mathematical results?
OpenAI published them on GitHub, along with guidance on how to navigate the repository.
How many of the manuscripts have been formalized?
OpenAI says 300 top-line results out of 719 manuscripts have been formalized, around 42 percent, and that the repository will be updated with more formalizations.
What areas of mathematics do the results cover?
They cover combinatorics, several branches of geometry, number theory, theoretical computer science, algebra, topology, probability and statistical mechanics, and mathematical physics.