今天我们要探讨一个很有意思的问题:聪明的AI和“真正理解”的AI,中间到底隔着什么?本期节目,我们将通过几篇最新论文,一探究竟。我们会看到,AI如何从只会“猜答案”的学生,蜕变为懂得自我批判的“思考者”;也会发现,我们大脑理解语言的秘密,可能在于一个高效的“跨部门协作网络”。接着,我们会揭示AI如何通过巧妙的“团队分工”,同时实现速度与性能的飞跃;最后,我们将探讨一门被忽视的“老手艺”和一种让AI“一句话一句话思考”的新模式,它们或许是让AI真正懂你的关键。准备好了吗?让我们一起开启这场关于AI思考方式的深度探索之旅!00:00:47 从“猜对答案”到“讲清道理”:AI的一次认知飞跃00:07:09 语言的“深加工”:你的大脑里藏着一个协作网络00:12:14 既要马儿跑,又要马儿不吃草?AI世界里有个新答案00:17:27 为什么你的AI总是不懂你?秘密藏在一门20年的老手艺里00:23:32 大模型“一句话一句话”地思考,可能吗?本期介绍的几篇论文:[LG] DeepSeekMath-V2: Towards Self-Verifiable Mathematical Reasoning[DeepSeek-AI]https://github.com/deepseek-ai/DeepSeek-Math-V2/blob/main/DeepSeekMath_V2.pdf---[CL] What does it mean to understand language?[Harvard University & Georgia Institute of Technology & MIT]https://arxiv.org/abs/2511.19757---[CL] Kimi Linear: An Expressive, Efficient Attention Architecture[Kimi Team]https://arxiv.org/abs/2510.26692---[CL] Context Engineering 2.0: The Context of Context Engineering[SJTU & SII]https://arxiv.org/abs/2510.26493---[CL] Continuous Autoregressive Language Models[WeChat AI, Tencent Inc]https://arxiv.org/abs/2510.27688
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