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

AI可可AI生活

[人人能懂] 洞察地球、高效学习、完美欺骗

23 Oct 2025

Description

今天,我们将开启一场从宏观到微观的AI探索之旅。我们将看到AI如何像做CT一样洞察整个地球的复杂系统,然后潜入它的大脑内部,看看它是如何分两步“猜测并精炼”出答案的。接着,我们会探讨AI如何像武林高手一样通过亲身实践来学习新知而又不忘旧事,并学会了像项目经理一样先规划再执行,以兼顾速度与质量。但最后,我们也会揭示一场AI世界的“无间道”,看看当聪明的AI学会“学术造假”时,会带来怎样严峻的挑战。00:00:39 我们如何用AI给地球做一次“全身CT”?00:05:27 大模型思考,需要分几步?00:11:25 AI的“鱼和熊掌”:既要快,又要好,可能吗?00:16:00 为什么高手越学越强,而我们一学就忘?00:21:22 AI世界的“无间道”:当“坏科学家”遇上“傻瓜审稿人”本期介绍的几篇论文:[AI] Earth AI: Unlocking Geospatial Insights with Foundation Models and Cross-Modal Reasoning[Google Research]https://arxiv.org/abs/2510.18318---[CL] How Do LLMs Use Their Depth?[UC Berkeley & Georgia Institute of Technology]https://arxiv.org/abs/2510.18871---[LG] Planned Diffusion[University of California, Los Angeles & MIT CSAIL]https://arxiv.org/abs/2510.18087---[LG] Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting[Princeton University]https://arxiv.org/abs/2510.18874---[AI] BadScientist: Can a Research Agent Write Convincing but Unsound Papers that Fool LLM Reviewers?[University of Washington]https://arxiv.org/abs/2510.18003

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.