Google DeepMind's AI agent autonomously solved 9 of 353 attempted open Erdős problems in mathematics, including two problems that had remained unsolved for 56 years, at a cost of only a few hundred dollars per problem solved. The system, built on Gemini models and incorporating formal verification through Lean, represents a breakthrough in AI's mathematical reasoning capabilities beyond pattern-matching to independent mathematical discovery. This signals a potential dramatic reduction in research costs and timeline for solving fundamental mathematical conjectures that have historically required significant human effort and funding.