[人人能懂AI前沿] AI如何深度思考？怎样成为说服大师？能创造游戏吗？

你有没有想过，一个更聪明的AI，是靠更大的模型，还是更深的“思考”？本期节目，我们将看到AI如何通过“反复打磨”超越百倍于自己的对手，如何用惊人的“信息密度”在辩论中战胜人类世界冠军。我们还会一起探索，当AI开始尝试建立自己的“世界观”、从零创造一个完整的游戏、甚至像我们一样行动时，一个怎样的新世界正在向我们走来。

00:00:29 聪明人的“笨功夫”，AI世界的新思考维度

00:06:39 那个最会“说话”的，已经不是人了

00:13:11 让AI拥有“世界观”，而不只是个“美图秀秀”

00:19:56 AI当“码农”可以，当“游戏制作人”呢？

00:25:33 向人学习，机器才能像人一样行动

本期介绍的几篇论文：

[LG] Looped World Models
[FaceMind Research Asia]
arxiv.org

[AI] AI systems out-persuade expert humans
[University of Oxford & UK AI Security Institute & Stanford University]
arxiv.org

[AI] Kairos: A Native World Model Stack for Physical AI
[Kairos Team]
arxiv.org

[CL] GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?
[The Chinese University of Hong Kong & Shenzhen Loop Area Institute]
arxiv.org

[LG] Looped World Models[FaceMind Research Asia]arxiv.org

[AI] AI systems out-persuade expert humans[University of Oxford & UK AI Security Institute & Stanford University]arxiv.org

[AI] Kairos: A Native World Model Stack for Physical AI[Kairos Team]arxiv.org

[CL] GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?[The Chinese University of Hong Kong & Shenzhen Loop Area Institute]arxiv.org

[RO] Human Universal Grasping[New York University]arxiv.org

[LG] Looped World Models
[FaceMind Research Asia]
arxiv.org

[AI] AI systems out-persuade expert humans
[University of Oxford & UK AI Security Institute & Stanford University]
arxiv.org

[AI] Kairos: A Native World Model Stack for Physical AI
[Kairos Team]
arxiv.org

[CL] GameCraft-Bench: Can Agents Build Playable Games End-to-End in a Real Game Engine?
[The Chinese University of Hong Kong & Shenzhen Loop Area Institute]
arxiv.org

[RO] Human Universal Grasping
[New York University]
arxiv.org