[人人能懂] 不止会算，更要会想：AI正在经历一场认知革命

00:00:29 AI 进阶之路：不造轮子，而是给高手装上预知未来的眼睛

00:04:38 AI进化新思路：不说人话，怎么教得会？ 

00:09:24 你的 App 突然崩了？别急，AI 程序员正在赶来修复的路上 

00:15:21 AI 也会看走眼？我们如何教它练就一双“火眼金睛” 

00:19:21 如何教AI学会“举一反三”？ 

本期介绍的五篇论文：

[CV] Back to the Features: DINO as a Foundation for Video World Models  

[Meta FAIR]  

https://arxiv.org/abs/2507.19468  

---

[CL] GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning  

[UC Berkeley & Stanford University]  

https://arxiv.org/abs/2507.19457  

---

[LG] Agentic Program Repair from Test Failures at Scale: A Neuro-symbolic approach with static analysis and test execution feedback  

[Meta]  

https://arxiv.org/abs/2507.18755  

---

[CL] PrismRAG: Boosting RAG Factuality with Distractor Resilience and Strategized Reasoning  

[Meta Reality Labs]  

https://arxiv.org/abs/2507.18857  

---

[LG] Scale-Consistent Learning for Partial Differential Equations  

[Caltech & Nvidia]  

https://arxiv.org/abs/2507.18813