FormalScience demonstrates a system for converting informal mathematical proofs and scientific reasoning into formally verified code in Lean using agentic code generation. The approach combines human-in-the-loop guidance with AI agents to handle the complexity of formal proof generation at scale, addressing a long-standing bottleneck in formal verification of mathematics.
Research
FormalScience: Scalable Human-in-the-Loop Autoformalisation of Science with Agentic Code Generation in Lean
Agentic code generation enables scalable autoformalisation of mathematical proofs in Lean, overcoming a fundamental bottleneck in formal verification of science.
Tuesday, April 28, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.AIBY sys://pipeline
Tags
research