Math models were already good at brute-force counterexample hunting, says AI researcher

dyamins · x · 2026-07-21

The post argues that the surprising part of the Jacobian conjecture counterexample is less that a math model could find it, and more that people had not previously applied enough automated search to look for this kind of counterexample.

The author says math models are already good at exhaustively checking conditions, handling combinatorial edge cases, and avoiding infinite search spaces through constraint reasoning.
They point to their own experience using models to surface annoying multivariable calculus and algebraic geometry counterexamples.
The broader claim is that automation often does the boring but crucial search work humans dislike, while human effort historically went into elaborate theory-building and proof attempts that still missed the counterexample.

Original post →

Math models were already good at brute-force counterexample hunting, says AI researcher

More from AGI Musings