Axiom Math wants to keep AI slop out of mathematics
As AI floods math with unchecked proofs, Axiom Math is betting on verification. The startup’s AI agents just formally verified the Stacks Project, one of the field’s biggest libraries.
Axiom Math, a startup valued at $1.6 billion, aims to prevent AI from generating inaccurate mathematical results. The company has formally verified the Stacks Project, a vast collection of mathematical knowledge, by translating its 7,654 pages into 16 million lines of computer-checked code. Mathematicians have been working on the Stacks Project since 2005, with Peter Woit, a Columbia mathematician, comparing it to the influential Bourbaki series.
The formal verification process involves rewriting proofs as code in the Lean programming language, which a proof assistant checks line by line. This method, known as formal verification, ensures the soundness of mathematical structures, similar to how engineers use it to validate chip designs. Axiom's CEO, Carina Hong, believes that formalizing math will create a verified foundation for AI agents to rely on, avoiding the "slop" that has plagued the field.
The company has trained its agents to work independently and in parallel, allowing them to identify tasks and collaborate efficiently. By formalizing the Stacks Project, Axiom has addressed a significant gap in existing formalized math, particularly in algebraic geometry. The CEO asserts that the more formalized proofs there are, the more can be built upon them, creating a reusable body of knowledge for future AI developments.
Written by urgent.news from TechCrunch's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
