你有没有想过，让AI学会“做白日梦”去预演触感，竟然能让它的动手能力提升90%？我们常说的“深度思考”，在AI那里可能只是一种高效的“循环播放”。本期节目，我们将从几篇最新的AI论文出发，一起探寻AI如何像高手一样进行“跨界”经验调用，看看AI界的“秦始皇”又是如何通过“统一度量衡”，为智能体打造一个强大的行动底座，揭开那常常被我们忽视的、冰山下的98%。

00:00:34 学会“做白日梦”，才能把活儿干好

00:05:23 AI的冰山，我们看不见的那98%

00:11:19 AI的“深度思考”，原来是“循环播放”？

00:16:46 高手，都善于“跨界”调用经验

00:23:38 AI 界的“秦始皇”，如何统一智能体的“度量衡”？

本期介绍的几篇论文：

[RO] Learning Versatile Humanoid Manipulation with Touch Dreaming  

[CMU]  

https://arxiv.org/abs/2604.13015 

---

[AI] Dive into Claude Code: The Design Space of Today's and Future AI Agent Systems  

[Mohamed bin Zayed University of Artificial Intelligence]  

https://arxiv.org/abs/2604.14228 

---

[LG] A Mechanistic Analysis of Looped Reasoning Language Models  

[University of Oxford & Mila]  

https://arxiv.org/abs/2604.11791 

---

[LG] Memory Transfer Learning: How Memories are Transferred Across Domains in Coding Agents  

[KAIST]  

https://arxiv.org/abs/2604.14004 

---

[AI] UniToolCall: Unifying Tool-Use Representation, Data, and Evaluation for LLM Agents  

[University of Science and Technology of China & Eastern Institute of Technology]  

https://arxiv.org/abs/2604.11557