![[人人能懂AI前沿] 从反刍记忆、辩证学习到分治与预见](https://image.xyzcdn.net/FuDP4HpAp8ezgVZMmEel3mblKCmJ.jpg@middle)
主播
节目简介
来源：小宇宙
今天，我们不聊堆算力的“大力出奇迹”，而是要探索几条让AI变得更智慧、更可靠的巧妙路径。我们会看到，一个简单的“反刍”机制，如何让AI不再健忘；一场内部“辩论赛”，又如何教会它诚实；“专家分工”的智慧，怎样让它在变强的同时还更省钱。最后，我们还会探究AI是如何学会像高手一样“抬头看路”地做决策，甚至在数学领域领悟“功夫在题外”的道理。准备好了吗？让我们一起揭开这些最新论文背后的绝妙构思。
00:00:37 AI的“反刍”，一个让它更聪明的简单魔法
00:05:10 如何让AI变得更聪明，同时还不变坏？
00:09:50 AI 进化新思路，从“大力出奇迹”到“聪明分工”
00:14:55 高手决策的秘密，既要埋头拉车，又要抬头看路
00:20:36 AI做数学，功夫在诗外
本期介绍的几篇论文：
[LG] Recirculation
[Google DeepMind]
https://arxiv.org/abs/2608.17981
---
[LG] Debate Training Reduces Reward Hacking in RLAIF
[Google DeepMind]
https://arxiv.org/abs/2608.17776
---
[CV] MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video Understanding
[Meta]
https://arxiv.org/abs/2608.17402
---
[LG] Q-Learning With World Models
[Stanford University & Peking University]
https://arxiv.org/abs/2608.17163
---
[AI] The Problem Is the Problem: Towards Scalable Mathematical Discovery
[CMU]
https://arxiv.org/abs/2608.16977
00:00:37 AI的“反刍”，一个让它更聪明的简单魔法
00:05:10 如何让AI变得更聪明，同时还不变坏？
00:09:50 AI 进化新思路，从“大力出奇迹”到“聪明分工”
00:14:55 高手决策的秘密，既要埋头拉车，又要抬头看路
00:20:36 AI做数学，功夫在诗外
本期介绍的几篇论文：
[LG] Recirculation
[Google DeepMind]
https://arxiv.org/abs/2608.17981
---
[LG] Debate Training Reduces Reward Hacking in RLAIF
[Google DeepMind]
https://arxiv.org/abs/2608.17776
---
[CV] MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video Understanding
[Meta]
https://arxiv.org/abs/2608.17402
---
[LG] Q-Learning With World Models
[Stanford University & Peking University]
https://arxiv.org/abs/2608.17163
---
[AI] The Problem Is the Problem: Towards Scalable Mathematical Discovery
[CMU]
https://arxiv.org/abs/2608.16977
外观
