𧟠Google's AI agent team just cracked a math puzzle that stood open since 2013
Google says its Gemini 4 models, working as teams of AI agents in Antigravity's /teamwork, helped prove information theory's CourtadeâKumar conjecture with researchers from CUHK, USC and CMU. The full proof was checked by computer in 50 million lines of Lean code. Another agent system fully autonomously formalized the Weil conjectures, a landmark result in mathematics, in about 1.8 million lines across 28,800 subagent runs. Google is now giving 10,000 academic researchers six months of free access to its top usage tier.
â AI is moving from answering homework to helping settle real open problems, and scientists can now try it themselves.
#AI
Google says its Gemini 4 models, working as teams of AI agents in Antigravity's /teamwork, helped prove information theory's CourtadeâKumar conjecture with researchers from CUHK, USC and CMU. The full proof was checked by computer in 50 million lines of Lean code. Another agent system fully autonomously formalized the Weil conjectures, a landmark result in mathematics, in about 1.8 million lines across 28,800 subagent runs. Google is now giving 10,000 academic researchers six months of free access to its top usage tier.
â AI is moving from answering homework to helping settle real open problems, and scientists can now try it themselves.
#AI
