[人人能懂] 从“学傻了”到“我错了” episode artwork

EPISODE · Dec 25, 2025 · 29 MIN

[人人能懂] 从“学傻了”到“我错了”

from AI可可AI生活

你有没有想过,为什么投入巨大的AI模型有时反而会“学傻了”?当AI的“词典”里没有“我错了”这个词时,我们又该如何教会它自我反思?本期节目,我们将一起钻进AI的大脑,从几篇最新的论文出发,看看AI是如何诊断自己内部的“罢工”,如何通过一场“无限游戏”变得更安全,以及它在绘画时,究竟是在搞创作,还是在“背书”。00:00:30 规模的诅咒,AI为何会“学傻”?00:06:29 AI的语言里,没有“我错了”00:11:35 想让AI更安全?答案可能藏在一场“无限游戏”里00:16:13 我们如何看穿世界的规则?AI给了新思路00:23:44 揭秘AI绘画,它“抄袭”的秘密藏在哪?本期介绍的几篇论文:[LG] Understanding Scaling Laws in Deep Neural Networks via Feature Learning Dynamics  [DePaul University & Iowa State University]  https://arxiv.org/abs/2512.21075 ---[CL] Reflection Pretraining Enables Token-Level Self-Correction in Biological Sequence Models  [Fudan University & Shanghai Artificial Intelligence Laboratory]  https://arxiv.org/abs/2512.20954 ---[LG] Safety Alignment of LMs via Non-cooperative Games  [FAIR at Meta & University of Tübingen]  https://arxiv.org/abs/2512.20806 ---[LG] Active inference and artificial reasoning  [University College London & VERSES]  https://arxiv.org/abs/2512.21129 ---[LG] Generalization of Diffusion Models Arises with a Balanced Representation Space  [University of Michigan]  https://arxiv.org/abs/2512.20963 在小宇宙查看该单集文稿

Episode metadata supplied by the publisher feed · Published Dec 25, 2025

Embed this episode

NOW PLAYING

[人人能懂] 从“学傻了”到“我错了”

0:00 29:21

No transcript for this episode yet

We transcribe on demand. Request one and we'll notify you when it's ready — usually under 10 minutes.

No similar episodes found.

Frequently Asked Questions

How long is this episode of AI可可AI生活?

This episode is 29 minutes long.

When was this AI可可AI生活 episode published?

This episode was published on December 25, 2025.

Can I download this AI可可AI生活 episode?

Yes. Use the download control on the episode player to save the publisher-provided media file.
URL copied to clipboard!