In August 2026, OpenAI made headlines with its advanced AI system, Astra, which reportedly solved 13 long-standing open problems in mathematics and theoretical computer science — challenges that had eluded human researchers for decades. The breakthrough included advances in areas such as group theory, coding theory, operator algebras, sphere packing, lattice cryptography, and computational complexity. To achieve these results, the team utilized the Lean theorem prover to formally verify the AI’s mathematical proofs. The announcement sparked significant debate within the scientific community: while some celebrated the milestone as a major step in AI’s reasoning capabilities, others called for further scrutiny to understand the full implications and limitations of the system.
About the Author
