Google and DeepMind Introduce Dream-RSI: A Recursive Self-Improvement Loop for AI Agents

Researchers from Google and DeepMind introduced Dream-RSI, a system where an AI agent improves how it explores tasks by replaying past attempts, cheaply testing thousands of alternative strategies, and selecting better ones for the next round. The approach preserved or improved results in algorithm design, mathematical optimization, and GPU-kernel creation while reducing search cost by up to 162x.