[人人能懂] AI新思路：偷师自然、过程淘宝、为知识画像

00:00:32 AI的“统一场”：它怎么看懂世界？00:04:10 AI的“半成品”，竟然是个宝库？00:07:43 AI学会“听话”的秘密：回到耳朵本身00:11:39 给数据“画个像”：一种全新的学习“渔”法本期介绍的四篇论文：[CL] Beyond the Rosetta Stone: Unification Forces in Generalization Dynamics  [Google DeepMind]  https://arxiv.org/abs/2508.11017 ---[CL] Diffusion is a code repair operator and generator  [Microsoft]  https://arxiv.org/abs/2508.11110 ---[CL] Representing Speech Through Autoregressive Prediction of Cochlear Tokens  [MIT & Stanford University]  https://arxiv.org/abs/2508.11598 ---[LG] Compressive Meta-Learning  [Stanford University & University of California, Santa Cruz & Polytechnic University of Catalonia]  https://arxiv.org/abs/2508.11090 在小宇宙查看该单集文稿