00:01:15 AI变“聪明”的代价：那根看不见的绳子

00:05:47 给AI“长考”时间，它怎么反而变笨了？

00:11:35 AI的“潜移默化”：看不见的信号，看得见的风险

00:15:45 AI教育的秘诀：一句“再试试”的力量

00:19:28 高手记忆的秘密：不是记得多，而是会“扔”

本期介绍的几篇论文：

[LG] The Invisible Leash: Why RLVR May Not Escape Its Origin  

[Stanford University & University of Tokyo]  

https://arxiv.org/abs/2507.14843 

---

[LG] Inverse Scaling in Test-Time Compute  

[University of Edinburgh & EPFL & University of Texas at Austin]  

https://arxiv.org/abs/2507.14417 

---

[LG] Subliminal Learning: Language models transmit behavioral traits via hidden signals in data  

[Anthropic Fellows Program & Truthful AI]  

https://arxiv.org/abs/2507.14805 

---

[LG] A Simple "Try Again" Can Elicit Multi-Turn LLM Reasoning  

[Imperial College London & Northwestern University & University of Washington]  

https://arxiv.org/abs/2507.14295 

---

[LG] LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models  

[Georgia Tech]  

https://arxiv.org/abs/2507.14204