如果AI团队开会只用“眼神”交流，会发生什么？本期节目，我们就来探索AI世界的奇妙新思路。我们将看到，驯服AI的秘诀，可能只是把一个“开关”换成“旋钮”，而让模型更聪明，也许只需给它一条“虚拟”的超宽车道。我们还会发现，机器人正从“打工仔”变身“修炼者”，而最顶尖的AI，甚至正在为它的同伴搭建一个“元宇宙”操场来练兵。让我们一起深入这些最新论文，看看智能的未来是如何被巧妙构想的。00:00:37 AI开会，从此只用“眼神”交流00:04:46 驯服AI：开关与旋钮的差别00:09:21 AI大模型的新玩法：不花大钱，也能办大事00:13:48 机器人不是在打工，是在“修炼”00:18:41 AI练兵新思路：没有操场，咱就造个“元宇宙”操场本期介绍的几篇论文：[CL] Latent Collaboration in Multi-Agent Systems[Princeton University & University of Illinois Urbana-Champaign]https://arxiv.org/abs/2511.20639---[LG] Soft Adaptive Policy Optimization[Qwen Team, Alibaba Inc.]https://arxiv.org/abs/2511.20347---[LG] Virtual Width Networks[ByteDance Seed]https://arxiv.org/abs/2511.11238---[RO] Robot-Powered Data Flywheels: Deploying Robots in the Wild for Continual Data Collection and Foundation Model Adaptation[Stanford University]https://arxiv.org/abs/2511.19647---[LG] Simulating Environments with Reasoning Models for Agent Training[University of Washington & Microsoft & CMU]https://arxiv.org/abs/2511.01824

Ask this episode

“What did the council decide on item 42?” — answered with the speaker and the moment.

Transcript

Related