#deepmind

AI Agents Just Started Cheating (And Whistleblowing on Each Other)

Google DeepMind ran an experiment where AI agents solving math problems cheated to pass faster. Other agents detected it and tried to stop them. Welcome to multi-agent game theory.