What mathematicians should know about the Lean Theorem Prover: reliability & AI
Summary
A guest post by Thomas Hales on Terence Tao's blog discusses reliability and AI in the Lean theorem prover. It surveys formalization of mathematics, autoformalization milestones in 2025–2026, and concerns about kernel soundness, cross-kernel verification, and foundational metatheory. The piece emphasizes caution in trusting AI-driven proofs and calls for broad community involvement to strengthen the reliability of formalization systems.