Reading up on Lean

AI makes hypothesis generation cheap, but scientific progress hinges on affordable, fast physical verification, requiring new lab‑infrastructure to match compute gains.

Jeremy Avigad argues that while AI can solve routine math problems, mathematicians must focus on harder problems, bigger conceptual thinking, and creative use of AI to advance understanding.

Gamburd argues AI‑generated math proofs undermine Enlightenment‑style verifiable knowledge, urging mathematicians to retain human‑understandable proofs as a decisive, bloodless counterweight to AI labs.

This post shows how to use Lean to write AI-generated proofs verifying equivariance and invariance properties of Transformer neural networks.

AI labs' undisclosed distillation of user data destroys credit assignment in mathematics, making open meritocratic discourse impossible and effectively killing the field as a public tradition.

The AI singularity has begun, as AI now solves major open math problems like Navier‑Stokes, proving it is the primary theorem‑prover and demanding acceptance of its reality.

The article shows how a visual interactive companion explains a Lean‑checked step in a forced Navier‑Stokes proof, clarifying when guarantees hold and when they fail.

Terence Tao uses a Bronze Medal meme to argue that AI-generated mathematical proofs are just the first step, with true value requiring verification, clear explanation, and eventual textbook canonization.

AI’s ability to solve high‑profile math problems erodes puzzle‑solving as a proxy for expertise, risking the loss of conceptual insight and idea generation in mathematics.

AI’s success in solving high‑profile math problems threatens the puzzle‑solving proxy that signals mathematical expertise, risking erosion of genuine idea generation and progress.

Anthropic's Claude autonomously produced the first computer‑checked proof of Fermat’s Last Theorem in Lean, completing the formalization in 11 days.

Stronger type systems like Rust and Lean improve AI-generated code quality by providing a verifiable compiler feedback loop that catches errors.

Using an LLM to mechanize proofs in Lean cut months of work to weeks, showing that PL research can now be produced far faster, reshaping publication expectations.

An unreleased version of Claude improved the lower bound for the fraction of Riemann zeta zeros satisfying the Riemann hypothesis from 41.6% to 67.2% by combining prior mathematical research.

Using formal verification, AI agents can generate code and machine‑checked proofs, removing the need for human code review and overcoming the AI‑generated‑code bottleneck.

AI tools have rapidly generated and formalized counterexamples to longstanding conjectures such as Erdős’ unit distance, Grothendieck’s group scheme problem, and the Jacobian conjecture, showing machines now out‑counterexample humans.

Terry Tao champions combining human insight, AI, and the Lean proof assistant to enable massive, verified mathematical collaborations.

An AI system called Gauss completed a mathematical proof formalization in five days, outpacing a human team that had worked for two years, raising concerns about the future role of mathematicians.

A personal guide lists and reviews nine Lean 4 books, offering subjective opinions and suggested learning paths.

Scientific discovery emerges from a loop of human question-posing, LLM proposal, external verification, and human curation, not from AI alone.

The Hacker News discussion argues that as AI writes more code, formal verification tools like Lean must scale to ensure correctness and trustworthiness.

Providing AI agents with automated feedback loops (back pressure) enables them to self-correct on long-horizon tasks, increasing engineer leverage.

Testing is argued to be more practical than formal verification, but commenters are divided on AI’s role in making verification mainstream.

AI-assisted formal verification shows promise but faces limitations — autoformalization trust, proof assistant speed, and model complexity — making random testing a necessary complement.

Formal verification will become mainstream as AI coding agents leverage proof systems like Lean for validation, according to a Hacker News discussion.