[人人能懂] 从梦境修炼、真实互动到思维几何

今天，我们将一起潜入AI的“内心世界”，探索几篇极具启发性的最新论文。我们将见证AI如何通过“梦中修炼”掌握复杂技能，又如何走出实验室，向真实的普通用户拜师学艺。我们还会首次窥见AI“大脑”发育的完整三部曲，并学习如何像侦探一样，从AI的“思考过程”中识别出它独特的性格。最后，我们将揭示一个惊人的记忆悖论：为何对AI来说，“记性太好”反而可能是一种诅咒？

00:00:36 AI的“梦中修炼”：如何在虚拟世界里，炼成绝世武功？

00:05:50 AI走出“象牙塔”：真实的用户，才是最好的老师

00:10:55 文如其人，AI如其思：我们能从“思考过程”看出AI的性格吗？

00:16:34 AI“大脑”的发育三部曲：从牙牙学语到融会贯通

00:21:42 AI的记忆悖论：为何“记性太好”，反而让你“忘得更快”？

本期介绍的几篇论文：

[LG] Training Agents Inside of Scalable World Models  

[Google DeepMind]  

https://arxiv.org/abs/2509.24527 

---

[LG] The Era of Real-World Human Interaction: RL from User Conversations  

[FAIR at Meta]  

https://arxiv.org/abs/2509.25137 

---

[CL] Your thoughts tell who you are: Characterize the reasoning patterns of LRMs  

[Meta Superintelligence Labs & Harvard University]  

https://arxiv.org/abs/2509.24147 

---

[LG] Tracing the Representation Geometry of Language Models from Pretraining to Post-training  

[McGill University & UC Berkeley]  

https://arxiv.org/abs/2509.23024 

---

[LG] Short window attention enables long-term memorization  

[Ecole Normale Supérieure Paris Saclay & Johannes Kepler University Linz & Meta FAIR]  

https://arxiv.org/abs/2509.24552