Glossary · Term

Lean

← all terms

Definition

Plain language

A programming language for writing math proofs where the computer refuses to accept any step that doesn't logically follow.

As stated in the literature

A dependently-typed proof assistant whose compiler accepts only programs that constitute valid formal proofs; the verification substrate used by AlphaProof, the DeepMind Erdős work, and most current AI-driven formal mathematics.

Also called: Lean 4

Why it matters: It gives AI math systems an external, unforgeable judge of correctness, so claimed proofs can be checked rather than just trusted.

For example, a proof of the irrationality of sqrt(2) in Lean is only accepted if every algebraic step is justified by a previously proved lemma or axiom.

Heard on the show

“There's a language called Lean 4 where you write math as code, and a small, paranoid core program — the kernel — checks every logical step against a fixed set of rules.”
Episode 188 — A Coding Agent Found a Hole in a Peer-Reviewed STOC Proof for Five Dollars

Mentioned in 9 episodes

  1. 188
    A Coding Agent Found a Hole in a Peer-Reviewed STOC Proof for Five Dollars
  2. 176
    An AI Designed Its Own Psychology Studies, Then Confirmed What It Found
  3. 122
    When Your Coding Agent Lies About the Fix: Verifying the Plan Before the Model Runs
  4. 120
    How an AI Agent Rewrites Its Own Tools, Without an Answer Key
  5. 117
    How an Open AI System Verified 672 Hard Math Proofs for Under $300
  6. 101
    Treating Math Formalization Like a Codebase, and Where the Agents Cheat
  7. 096
    How Treating an AI Agent's Execution Like Git Recovers a Coordination Penalty
  8. 075
    Growing Code and Proof Together: Verified Systems in Ten Hours Instead of a Year
  9. 067
    An AI Just Solved a 1996 Erdős Problem—and the Simplest Agent Won

Related concepts

Related terms