你有没有想过，最顶尖的AI是如何思考的？本期节目，我们将从四篇最新论文出发，揭示AI成长的秘密：有时，思路的宽度比深度更重要；有时，机器也需要演化出难以言喻的“品味”；甚至，它还需要学会“左手画方、右手画圆”的协同技巧，并懂得在关键时刻，用恰当的“约束”来避免犯下最聪明的傻错误。准备好了吗？让我们一起潜入AI思考的深水区。

00:00:34 AI搞科研，拼的不是智商，而是“思路宽”

00:05:32 如何让机器拥有“数学品味”？

00:10:30 AI思考，也需要“左手画方，右手画圆”？

00:16:58 为什么最聪明的工具，反而最容易犯傻？

本期介绍的几篇论文：

[AI] What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversity  

[FAIR at Meta]  

https://arxiv.org/abs/2511.15593 

---

[LG] Learning Interestingness in Automated Mathematical Theory Formation  

[UT Austin]  

https://arxiv.org/abs/2511.14778 

---

[CV] Think Visually, Reason Textually: Vision-Language Synergy in ARC  

[The Chinese University of Hong Kong & Shanghai AI Laboratory]  

https://arxiv.org/abs/2511.15703 

---

[LG] CODE: A global approach to ODE dynamics learning  

[University of Stuttgart & Stanford University]  

https://arxiv.org/abs/2511.15619