[人人能懂] 物理、工具与过程的智慧

00:00:35 AI太耗电？可能我们一开始就想错了00:04:23 AI的“外挂”：我们到底需要什么样的智能？00:09:27 如何让AI“带徒弟”：过程，永远比结果更重要00:13:48 创新的“乐高”：如何像拼积木一样“制造”好点子？00:18:26 AI界的“瑞士军刀”：一个模型搞定所有视频“抠图”本期介绍的几篇论文：[LG] Learning with springs and sticks  [University of Toronto]  https://arxiv.org/abs/2508.19015  ---[LG] Understanding Tool-Integrated Reasoning  [Tencent]  https://arxiv.org/abs/2508.19201  ---[LG] StepWiser: Stepwise Generative Judges for Wiser Reasoning  [FAIR at Meta]  https://arxiv.org/abs/2508.19229  ---[LG] The Ramon Llull's Thinking Machine for Automated Ideation  [CMU & OSU & Stanford University]  https://arxiv.org/abs/2508.19200  ---[CV] Autoregressive Universal Video Segmentation Model  [NVIDIA & CMU]  https://arxiv.org/abs/2508.19242  在小宇宙查看该单集文稿