Autonomous Mathematical Discovery in an Open-World Multi-Agent Environment

Researchers demonstrated autonomous mathematical discovery using multiple AI agents working collaboratively in an open-world environment called the Station, where agents independently chose research directions and conducted experiments without central coordination. The system achieved novel results on five mathematical problems, including new Kakeya sets, kissing configurations, and improved bounds for several open problems, while also generating interpretable theorems and proofs explaining the constructions. The researchers released all agent dialogues, proofs, and verification code to provide transparency into how the discoveries emerged.