今天我们要聊一个根本问题:聪明的AI到底是什么样?是像一位能极速压缩思考时间的解题大师,还是一个能在想象的沙盒里自我进化的机器人?我们还会一起探究,为什么AI时而是无所不知的学霸,时而又是死记硬背的书呆子,甚至还会一本正经地胡说八道。最后,你会发现,让AI变得更懂你的终极秘诀,可能恰恰是先教会它如何“装傻”提问。让我们跟随几篇最新论文,一起解剖AI的思考内核。00:00:36 AI变聪明的秘密:不是知道更多,而是想得更快00:06:29 给机器人一个“沙盒”,让它在想象中进化00:12:37 为什么AI既是学霸,又是书呆子?00:17:19 AI的“发挥失常”:一个问题,两种症状00:23:35 如何让机器更懂你?答案是:先让它学会“装傻”本期介绍的几篇论文:[LG] AI Agents as Universal Task Solvers: It’s All About Time[AWS Agentic AI]https://arxiv.org/abs/2510.12066---[RO] Ctrl-World: A Controllable Generative World Model for Robot Manipulation[Stanford University & Tsinghua University]https://arxiv.org/abs/2510.10125---[LG] LLM Knowledge is Brittle: Truthfulness Representations Rely on Superficial Resemblance[FAIR at Meta & University of Zurich]https://arxiv.org/abs/2510.11905---[CL] Generation Space Size: Understanding and Calibrating Open-Endedness of LLM Generations[Stanford University]https://arxiv.org/abs/2510.12699---[LG] Asking Clarifying Questions for Preference Elicitation With Large Language Models[Google]https://arxiv.org/abs/2510.12015
No persons identified in this episode.
This episode hasn't been transcribed yet
Help us prioritize this episode for transcription by upvoting it.
Popular episodes get transcribed faster
Other recent transcribed episodes
Transcribed and ready to explore now
SpaceX Said to Pursue 2026 IPO
10 Dec 2025
Bloomberg Tech
Don’t Call It a Comeback
10 Dec 2025
Motley Fool Money
Japan Claims AGI, Pentagon Adopts Gemini, and MIT Designs New Medicines
10 Dec 2025
The Daily AI Show
Eric Larsen on the emergence and potential of AI in healthcare
10 Dec 2025
McKinsey on Healthcare
What it will take for AI to scale (energy, compute, talent)
10 Dec 2025
Azeem Azhar's Exponential View
Reducing Burnout and Boosting Revenue in ASCs
10 Dec 2025
Becker’s Healthcare -- Spine and Orthopedic Podcast