rick awsb ($people, $people)(@rickawsb):google的dream rsi是harness的rsi openai的astra开发下一代模型，是模型参数的rsi ASPIRE、meta depth等方法，说的是验证的rsi 此外，还有工具调用的rsi，递归深度的rsi，目标设置的rsi。。。。。。 rsi本身正在多维度扩展，rsi本身正在scaling 这就是为什么，ai能力增长快得令人发指

2026.09.17 15:35

google的dream rsi是harness的rsi openai的astra开发下一代模型，是模型参数的rsi ASPIRE、meta depth等方法，说的是验证的rsi 此外，还有工具调用的rsi，递归深度的rsi，目标设置的rsi。。。。。。 rsi本身正在多维度扩展，rsi本身正在scaling 这就是为什么，ai能力增长快得令人发指

Dr Singularity@Dr_Singularity

2026.09.16 12:46

big AI news Google just demonstrated a recursive self improvement loop for AI discovery Google/DeepMind researchers introduced Dream-RSI, a system where an AI agent improves how it explores problems by replaying its past discovery attempts, testing thousands of alternative strategies cheaply, then deploying the better strategy in the next round. Across algorithm design, mathematical optimization, and GPU kernel engineering, it matched or improved discovery quality while cutting search costs dramatically, in one setting reducing agent calls by up to 162x. 👀 Importantly, it improves the exploration policy, not the underlying model weights.