Axiom Math, a Palo Alto startup, has released Axplorer, a free AI tool designed to help mathematicians discover patterns that could solve long-standing problems. This tool evolved from PatternBoost, which required supercomputing power and successfully solved the Turán four-cycles problem. Axplorer runs on a standard Mac Pro, making advanced pattern recognition accessible to individual researchers.

The tool represents part of DARPA's expMath initiative encouraging AI adoption in mathematics. François Charton, now at Axiom, co-developed the original PatternBoost at Meta. AxiomProver, another of the company's tools, solved four math problems last year. Mathematicians remain skeptical about whether these tools will truly accelerate research beyond traditional methods.

While open-sourced and available on GitHub, Axplorer faces competition from other AI tools like Google DeepMind's AlphaEvolve. The code allows users to generate sample solutions and counterexamples, potentially speeding up mathematical discovery. Despite enthusiasm, experts caution that AI tools should complement rather than replace traditional mathematical approaches.