EPISODE · Dec 12, 2025 · 33 MIN
[人人能懂] 从系统蓝图到思维手术刀
from AI可可AI生活
你有没有想过,我们不仅能使用AI,还能看透它的“内心世界”吗?本期,我们将跟随几篇最新论文,扮演一次AI的建筑师、心理医生和侦探,看看如何重塑一个更省钱的推荐系统,如何看穿AI从无害信息中“悟”出的危险想法,以及如何用一把“手术刀”解剖并教会它举一反三。让我们一起出发!00:00:28 推荐系统的大难题:如何让它更聪明,还更省钱?00:05:59 你教AI画苹果,它却学会了当海盗00:14:26 我们终于有了一把解剖AI的“手术刀”00:20:44 AI侦探:你的模型到底跟谁学的?00:27:06 教AI举一反三:为什么聪明的模型,一到考场就蒙圈?本期介绍的几篇论文:[IR] Meta Lattice: Model Space Redesign for Cost-Effective Industry-Scale Ads Recommendations [Meta AI] https://arxiv.org/abs/2512.09200 ---[CL] Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMs [Truthful AI] https://arxiv.org/abs/2512.09742 ---[LG] Provably Learning from Modern Language Models via Low Logit Rank [Microsoft Research & UC Berkeley & MIT] https://arxiv.org/abs/2512.09892 ---[LG] Natural Geometry of Robust Data Attribution: From Convex Models to Deep Networks [The University of Texas at Austin] https://arxiv.org/abs/2512.09103 ---[LG] Closing the Train-Test Gap in World Models for Gradient-Based Planning [Columbia University] https://arxiv.org/abs/2512.09929 在小宇宙查看该单集文稿
Embed this episode
NOW PLAYING
[人人能懂] 从系统蓝图到思维手术刀
No transcript for this episode yet
Similar Episodes
No similar episodes found.