AI just solved a century of math mysteries by translating them into code

Google's new AI system, called LEAP, has achieved a historic leap forward in formal verification—the process of turning mathematical proofs into computer-checkable code. Working with its companion system Argon, LEAP autonomously translated some of mathematics' most profound and long-standing unsolved problems into the Lean 4 programming language, allowing a computer to verify their proofs line-by-line for the first time.

The AI's accomplishments are staggering: it formally verified proofs for the Weil conjectures (a 1974 landmark), Hilbert's tenth problem, the 3D Navier-Stokes equations (a Millennium Prize problem), and the Birkhoff conjecture open since 1927. It even found and fixed a gap in an existing proof. This breakthrough shifts the burden of tedious verification from human reviewers to machines, potentially accelerating scientific discovery by letting researchers focus on creativity while computers handle the meticulous, error-prone checking. The era of AI-augmented mathematics has officially arrived.