Fable’s Jacobian “counterexample” appears to have rediscovered a 1999 paper

dotey · x · 2026-07-21

An X thread says an AI lab mathematician used Fable to find what looked like a counterexample to the famous Jacobian conjecture, but the story later flipped: a very similar counterexample had already been constructed in 1999 by Russian mathematician Vitushkin.

The poster argues Fable likely retrieved the old paper and repackaged it into a different-dimensional claim. They also contrast it with a previous OpenAI episode in which the company confidently claimed an original solution to an Erdős conjecture, only for it to turn out to have been retrieved rather than derived.

Original post →

More from Fun

Fun channel →