Axiom Math wants to keep AI slop out of mathematics
techcrunch ·
AxiomMath, a $1.6 billion startup founded by Stanford dropout Carina Hong, has successfully formally verified the Stacks Project, one of the most extensive bodies of human-written mathematics. By translating approximately 7,654 pages into 16 million lines of Lean code, the company aims to create a reliable foundation for AI agents to use in software development and scientific research. This effort addresses growing concerns about low-quality 'AI slop' flooding academic journals with unverified claims. The verification process ensures logical soundness through proof assistants, preserving the interconnected structure of complex theorems in algebraic geometry. Axiom trained its own models to develop an intuitive sense for mathematical correctness, allowing parallel AI agents to work efficiently on the project. The resulting formalized library serves as a reusable substrate, enabling future proofs to build upon checked results rather than starting from scratch. CEO Hong challenges other coding companies to adopt similar large-scale formalization efforts to ensure the integrity of AI-generated infrastructure.
techcrunch