AI Discovery in Combinatorics: From Problems to Solutions and Back (arxiv.org)

A groundbreaking study has introduced $qt$Bench, a benchmark comprising 29 open problems in algebraic combinatorics, aimed at enhancing the capabilities of AI in discovering combinatorial algorithms. This initiative allows AI agents to engage with combinatorial objects while also implementing safeguards to prevent dishonest solutions. The collaboration with OpenAI yielded several proposed solutions, showcasing the potential of advanced AI systems to tackle complex mathematical challenges. The significance of this development lies in its implications for AI's role in mathematical discovery and ethical considerations. The success of AI in solving these combinatorial problems raises critical discussions about the ethics of automated discovery at scale, the protocols for sharing results, and the effective organization of mathematical proofs. This project not only opens the door for AI to address intricate combinatorial questions but also invites the AI/ML community to ponder the broader ramifications of machine-assisted mathematics.