Axiom Math wants to keep AI slop out of mathematics

You can now access a foundational library of mathematics that is computer-verified, reducing the risk of errors in AI-generated code and scientific research.

Axiom Math, a startup valued at $1.6 billion, has announced the formal verification of the Stacks Project, a massive and foundational collection of mathematical work. The company's AI agents translated approximately 7,654 pages of the Stacks Project into 16 million lines of computer-checked code using the Lean programming language and proof assistants. This process ensures the logical soundness and consistency of the mathematical theorems and definitions, akin to formal verification used in chip design. The Stacks Project, initiated in 2005, covers advanced topics in algebra, geometry, and number theory. Axiom Math aims to build a reliable, verified foundation that future AI agents can use for writing software and conducting scientific research, preventing the spread of "AI slop" in these fields. The company raised $200 million in a Series A funding round led by Menlo Ventures.

Axiom Math's successful formal verification of the Stacks Project signifies a potential paradigm shift in how AI agents will interact with and generate complex knowledge. By establishing a robust, computer-checked foundation, the company is creating a reliable substrate for AI development, aiming to mitigate the risks of "AI slop"—inaccurate or unsound outputs—that plague emerging AI applications. This effort not only addresses the immediate need for verified mathematical proofs but also sets a precedent for formalizing other complex domains, potentially impacting software engineering and scientific discovery.

The company's approach, utilizing AI agents that operate with a loose hierarchy and self-directed task identification, reflects a scalable method for tackling vast and intricate bodies of knowledge. This contrasts with traditional, more rigid verification methods and suggests a future where AI collaboratively builds upon verified knowledge bases. Axiom's focus on developing specialized AI models for mathematical reasoning further distinguishes it, positioning the startup as a key player in ensuring the integrity and reliability of AI-driven innovation across various sectors.