OpenAI releases 372 math results from an unreleased model
Many proofs have been checked in Lean; mathematicians say it will take months to assess their significance.
OpenAI released 372 results from its new, unreleased large language model, each of which the company says resolves or makes substantial progress on a major open question in mathematics or theoretical computer science, Scientific American reported on Tuesday.
The company posted the results in a GitHub repository. Many have already been verified in Lean, a programming language that validates a proof's logic, according to the report. Claimed results include a solution to the four-dimensional Kakeya conjecture and improvements to important computer algorithms.
An OpenAI spokesperson said the model produced almost every result in response to a single prompt handed to a single AI agent, Scientific American reported.
Mathematicians will need months to determine whether the proofs contain novel ideas or are mostly combinations of existing techniques, the publication said.
From ARQS Sponsored
For B.C. teams putting stories like this into practice, ARQS offers Flow automation, SignDesk, Northline course catalogue and ARQS security. In the ARQS Store, ARQS Pipeline Pro Setup, ARQS InvoiceFlow Billing Control Studio, ARQS InvoiceFlow Collections Console are among the newest tools available.