Künstliche Intelligenz, KI, Artificial Intelligence, AI, Quantencomputer, Robotik, ChatGPT, Grok, Transhumanismus

🔥DeepMind’s AI just cracked 9 unsolved math problems, autonomously. Researchers revealed that DeepMind’s new agent, AlphaProof Nexus, successfully solved 9 open Erdős problems, including two that had remained unsolved for more than 50 years. What’s wild is the cost: each successful solution reportedly took only a few hundred dollars in compute. But behind that number was a massive search effort. The system actually attempted all 353 formalized Erdős problems, repeatedly generating proofs, checking them in Lean (a formal mathematics verification system), analyzing failures, and trying again. If the agent exceeded 3,000 iterations without progress, the run was stopped. The architecture worked like an AI mathematician with memory and strategy: • Generate a proof attempt • Verify it formally in Lean • Analyze errors and refine the approach • Explore multiple reasoning branches simultaneously • Save promising intermediate ideas • Abandon dead ends through evolutionary search Beyond the Erdős tasks, the system also proved 44 previously unknown hypotheses from the OEIS mathematical sequence database. This may be the first large-scale benchmark where an AI company tested agents directly on unsolved research problems instead of standard exams or coding tasks. 📁 Research paper. 💻 Proof results and evidence. @aipost 🏴