Meta shares six AI-assisted mathematics papers

12 hours ago 28

Meta shared six mathematics papers developed with Muse Spark on October 2, saying five present answers to previously open research questions.

Mathematicians used Muse Spark 1.1 and 1.2 in Thinking Mode through the standard Meta AI chat interface over several months. Researchers directed the work, and a second group reviewed the arguments. Papers identify passages primarily drafted by humans or AI.

The results span probability, differential equations, group theory, optimization, arithmetic physics and non-associative algebra.

In group theory, Muse Spark generated a GAP search program that found a 384-element counterexample to M. Kida’s 2024 conjecture. Researchers verified it and completed the argument.

Another paper establishes a threshold for fitting random Gaussian points to an ellipsoid, while leaving behavior at the threshold unresolved. A wave-collapse paper addresses a question open since 2015 under specified symmetry and energy conditions.

Meta also reported a three-dimensional counterexample to a conjecture on solvable evolution algebras. It acknowledged independent work on that problem, ellipsoid fitting and the group-theory conjecture.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.

Read Entire Article