PodcastsTechnologyAI可可AI生活

AI可可AI生活

fly51fly
AI可可AI生活
Latest episode

947 episodes

  • AI可可AI生活

    [人人能懂AI前沿] 从举一反三、任务分解到动态“省钱”

    28/05/2026 | 28 mins.
    想知道AI如何学会“看眼色”举一反三吗?本期,我们将一起揭秘几篇最新论文,看看AI如何像有了“变速箱”一样动态切换快慢刀,又如何通过“智能菜谱”让机器人学会干活。我们还会聊聊如何给AI一张“认知体检表”来衡量它的真实水平,以及它那套偷偷学会的“省钱”妙招。准备好了吗?让我们一探究竟!
    00:00:34 AI如何学会“看眼色”,一个关于举一反三的发现
    00:05:43 机器人后空翻都会,为什么还不会端茶倒水?
    00:11:28 AI的“变速箱”,什么时候该用牛刀?
    00:17:29 给AI一张体检表,我们离通用人工智能还有多远?
    00:23:00 你的AI,正在偷偷学会“省钱”
    本期介绍的几篇论文:
    [LG] Fine-Tuning Dynamics of In-Context Factual Recall in Transformers
    [Duke University & Princeton University & UC Berkeley]
    https://arxiv.org/abs/2605.27774
    ---
    [RO] HumanoidMimicGen: Data Generation for Loco-Manipulation via Whole-Body Planning
    [NVIDIA]
    https://arxiv.org/abs/2605.27724
    ---
    [LG] Multi-Mixer Models: Flexible Sequence Modeling with Shared Representations
    [CMU & Google Research]
    https://arxiv.org/abs/2605.28769
    ---
    [AI] Measuring Progress Toward AGI: A Cognitive Framework
    [Google DeepMind]
    https://arxiv.org/abs/2605.28405
    ---
    [LG] Meta-Attention: Bayesian Per-Token Routing for Efficient Transformer Inference
    [Knowledge Lab AG]
    https://arxiv.org/abs/2605.28384

    在小宇宙查看该单集文稿
  • AI可可AI生活

    [人人能懂AI前沿] 世界模型、自我蒸馏、缩放法则、证据链与专家混合

    27/05/2026 | 34 mins.
    你有没有想过,一个AI要怎样才能真正“看懂”世界,而不是假装看懂?本期节目,我们将从一篇最新论文出发,揭示一个简单的“高斯沙堆”如何帮助AI解开世界的本质。紧接着,我们会看到AI如何像武林高手一样“左右互搏”,在没有老师的情况下实现自我进化。然后,我们将探索一个能预测AI未来的“万能公式”,并直面AI的“诚信危机”,看看如何用一条“证据链”让它变得可信。最后,我们会发现,为了让你的手机变得更聪明,AI正在悄悄组建一个“专家委员会”。
    00:00:42 想让AI看懂世界?你只需要一个高斯分布
    00:06:30 不用老师,AI怎么自己考上大学?
    00:12:21 增长的“万能公式”,如何科学地预测你的下一步?
    00:19:27 AI的诚信危机,我们如何相信一个“超级大脑”?
    00:25:58 你的手机,为什么要养一个“专家委员会”?
    本期介绍的几篇论文:
    [LG] When Does LeJEPA Learn a World Model?
    [Cold Spring Harbor Laboratory & New York University & Brown University]
    https://arxiv.org/abs/2605.26379
    ---
    [CL] Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipeline
    [Stanford University]
    https://arxiv.org/abs/2605.26132
    ---
    [LG] Unified Neural Scaling Laws
    [Mila, University of Montreal & Google DeepMind]
    https://arxiv.org/abs/2605.26248
    ---
    [AI] ScientistOne: Towards Human-Level Autonomous Research via Chain-of-Evidence
    [Google Cloud AI Research]
    https://arxiv.org/abs/2605.26340
    ---
    [LG] MobileMoE: Scaling On-Device Mixture of Experts
    [Meta AI]
    https://arxiv.org/abs/2605.27358

    在小宇宙查看该单集文稿
  • AI可可AI生活

    [人人能懂AI前沿] 从AI助教、因果科学到稀疏革命

    26/05/2026 | 29 mins.
    你有没有想过,AI不仅能成为拉平世界教育鸿沟的“私教”,还能化身为你专属的“出题官”?这一期,我们将一起探索几篇有趣的最新论文,看看AI是如何从一个“大力出奇迹”的炼丹师,进化为懂得因果科学的工程师。我们还会揭示一个惊人发现:高效的AI可能一直在“假装”认真阅读,而我们用来测试它速度的秒表,甚至可能一直在“欺骗”我们!
    00:00:32 你的AI私教,正在悄悄拉平世界
    00:06:58 想造好AI?可能得先补上这门“因果课”
    00:12:42 AI读长文太慢?也许它一直在“假装”认真
    00:18:28 你的私人出题官,是怎么炼成的?
    00:23:55 你的AI有多快?小心,你的秒表可能在骗你
    本期介绍的几篇论文:
    [AI] Human-AI Collaboration in Science at Scale: A Global Large-scale Randomized Field Experiment
    [Northwestern Universit & Stanford University]
    https://arxiv.org/abs/2605.24180
    ---
    [LG] Causal methods for LLM development and evaluation
    [MCML & CMU]
    https://arxiv.org/abs/2605.25998
    ---
    [LG] Inference Time Context Sparsity: Illusion or Opportunity?
    [Rice University & UC Berkeley]
    https://arxiv.org/abs/2605.24168
    ---
    [AI] KT4EQG: Personalized Exercise Question Generation via Knowledge Tracing
    [University of California, Santa Barbara & University of South Alabama]
    https://arxiv.org/abs/2605.23933
    ---
    [AI] Identifying and Mitigating Systemic Measurement Bias in Production LLM Inference Benchmarks
    [Google]
    https://arxiv.org/abs/2605.24217

    在小宇宙查看该单集文稿
  • AI可可AI生活

    [人人能懂AI前沿] 从信噪比、凸优化到底线思维:重塑AI的五个底层逻辑

    25/05/2026 | 30 mins.
    你有没有想过,为什么AI会越练越笨,甚至被人类“算计”?本期节目,我们将一口气解锁五篇最新论文带来的颠覆性思考:从用香农定律解释AI的“U型学习曲线”,到仅用一块显卡就能“调教”出听话的大模型;从赋予AI“底线思维”来应对未知风险,到让它写出“绝对正确”的关键代码;最后,我们还会看到AI如何从“看图识字”进化到真正“读懂规矩”。准备好,一场关于AI思维升级的风暴即将来临!
    00:00:37 你的模型为什么越练越笨?
    00:07:11 AI大模型调教指南,如何用一块显卡搞定?
    00:12:44 面对“深不可测”的世界,AI如何做出明智选择?
    00:18:45 AI开始写“绝对正确”的代码了
    00:23:43 AI的新玩法,从“看图识字”到“读懂规矩”
    本期介绍的几篇论文:
    [LG] LLMs as Noisy Channels: A Shannon Perspective on Model Capacity and Scaling Laws
    [ByteDance Seed]
    https://arxiv.org/abs/2605.23901
    ---
    [LG] Convex Optimization for Alignment and Preference Learning on a Single GPU
    [Stanford University]
    https://arxiv.org/abs/2605.23244
    ---
    [LG] Infra-Bayesian Reinforcement Learning Agents Outperform Classical RL For Worst-Case Robustness
    [Purdue University & CMU & WorldQuant University]
    https://arxiv.org/abs/2605.23146
    ---
    [AI] Inductive Deductive Synthesis: Enabling AI to Generate Formally Verified Systems
    [UC Berkeley]
    https://arxiv.org/abs/2605.23109
    ---
    [CV] General Hazard Detection
    [Swinburne University of Technology]
    https://arxiv.org/abs/2605.23304

    在小宇宙查看该单集文稿
  • AI可可AI生活

    [人人能懂AI前沿] AI变聪明的捷径:给它后悔药、世界地图和一位好陪练

    24/05/2026 | 30 mins.
    想让AI变聪明,一定要把它做得更大、训练得更久吗?最新论文告诉我们,恰恰相反!今天我们要聊聊,如何通过为AI“挖一口深井”来代替疯狂“堆料”,如何给它一颗“后悔药”让小模型也能超越巨无霸。我们还会揭秘AI训练的“快进键”是如何实现的,以及如何给AI一份“世界地图”和一位“高手陪练”,让它拥有永不枯竭的好奇心和可被预测的光明未来。
    00:00:35 人工智能“开窍”的秘密,与其堆料,不如挖井
    00:05:53 给机器一颗“后悔药”,小模型如何反超大模型?
    00:12:02 AI训练的“快进键”,为什么只练开头就能猜到结尾?
    00:17:12 AI“路痴”自救指南,如何拥有永不枯竭的好奇心
    00:23:16 AI养成记,如何给一个“笨”模型当好陪练?
    本期介绍的几篇论文:
    [LG] Equilibrium Reasoners: Learning Attractors Enables Scalable Reasoning
    [CMU]
    https://arxiv.org/abs/2605.21488
    ---
    [AI] Probabilistic Tiny Recursive Model
    [Mila – Quebec AI Institute]
    https://arxiv.org/abs/2605.19943
    ---
    [LG] You Only Need Minimal RLVR Training: Extrapolating LLMs via Rank-1 Trajectories
    [University of Virginia]
    https://arxiv.org/abs/2605.21468
    ---
    [LG] Remember to be Curious: Episodic Context and Persistent Worlds for 3D Exploration
    [University of Toronto & UC Berkeley & Wayve]
    https://arxiv.org/abs/2605.22814
    ---
    [CL] Forecasting Downstream Performance of LLMs With Proxy Metrics
    [Mila – Quebec AI Institute & McGill University]
    https://arxiv.org/abs/2605.18607

    在小宇宙查看该单集文稿
More Technology podcasts
About AI可可AI生活
来自 @爱可可-爱生活 的第一手AI快报,用最简单易懂的语言,带你直击最前沿的人工智能科研动态。无论你是科技小白,还是行业达人,这里都有你想知道的AI故事和未来趋势。跟着我们,轻松解锁人工智能的无限可能! #人工智能 #科技前沿
Podcast website

Listen to AI可可AI生活, The AI Daily Brief: Artificial Intelligence News and Analysis and many other podcasts from around the world with the radio.net app

Get the free radio.net app

  • Stations and podcasts to bookmark
  • Stream via Wi-Fi or Bluetooth
  • Supports Carplay & Android Auto
  • Many other app features