Lean-based neural theorem prover solves AMC12/AIME/IMO olympiad problems

AI Impact Summary

A neural theorem prover for Lean has been built to tackle a subset of high-school olympiad problems, including AMC12, AIME, and adapted IMO questions. This pushes automated reasoning within Lean into problem domains traditionally solved by heuristics, enabling automated proof search and potential educational tooling. The change creates opportunities for automated tutoring, problem-solving demos, and benchmarking, but it also demands rigorous correctness assurance and traceable proofs to prevent misleading results in production.