Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Pricing
Podcast Image

AI可可AI生活

[人人能懂] 从灵感溯源、速读秘诀到诚实AI

18 Nov 2025

Description

你有没有想过,最顶尖的AI,它的智慧可能不是体现在无所不知,而是敢于坦诚地说出“我不知道”?本期节目,我们将一起探索AI如何学会这项宝贵的品质。我们还会揭秘,如何给AI装上一双“眼睛”让它在嘈杂派对里也能跟你轻松对话,如何用一个优美的公式教会它“速读”长篇报告,甚至让一份200页的PDF自己开口说话,并在一秒内找到AI画作的灵感“祖先”。准备好了吗?让我们一起进入AI更深邃、更智慧的内心世界。00:00:39 AI画画的灵感,能秒速溯源吗?00:06:29 大模型读书慢?给它一副聪明的“速读眼镜”00:12:13 给AI一双眼睛,让它学会“察言观色”00:16:37 AI的最高智慧,是承认自己不知道00:22:56 如何让一份200页的PDF,自己开口说话?本期介绍的几篇论文:[CV] Fast Data Attribution for Text-to-Image Models[CMU & Adobe Research & UC Berkeley]https://arxiv.org/abs/2511.10721---[LG] Optimizing Mixture of Block Attention[MIT]https://arxiv.org/abs/2511.11571---[CL] AV-Dialog: Spoken Dialogue Models with Audio-Visual Input[University of Washington & Meta AI Research]https://arxiv.org/abs/2511.11124---[LG] Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation[Toyota Technological Institute at Chicago & University of California, San Diego]https://arxiv.org/abs/2511.11500---[CL] Information Extraction From Fiscal Documents Using LLMs[Google Inc & XKDR Forum]https://arxiv.org/abs/2511.10659

Audio
Featured in this Episode

No persons identified in this episode.

Transcription

This episode hasn't been transcribed yet

Help us prioritize this episode for transcription by upvoting it.

0 upvotes
🗳️ Sign in to Upvote

Popular episodes get transcribed faster

Comments

There are no comments yet.

Please log in to write the first comment.