「技术不无聊」是投资人孟醒和朋友们共同发起的一档技术播客，在这里我们带着第一视角、好奇心和“这可能是个坑”的心态，把大时代中的小技术展开讲讲。
这些分享也许不够成熟、不够客观甚至有些粗糙，但一定足够鲜活且充满创造力，毕竟，冒险和创新仍是这个世界上最性感的命题。
技术不无聊，创业有得聊。
Podcast Overview
「技术不无聊」是投资人孟醒和朋友们共同发起的一档技术播客，在这里我们带着第一视角、好奇心和“这可能是个坑”的心态，把大时代中的小技术展开讲讲。 这些分享也许不够成熟、不够客观甚至有些粗糙，但一定足够鲜活且充满创造力，毕竟，冒险和创新仍是这个世界上最性感的命题。 技术不无聊，创业有得聊。
Language
🇨🇳
Publishing Since
8/26/2024
1 verified contact email on file for 技术不无聊
Pitch yourself as a guest, propose sponsorships, or reach out directly to the host.
Recent Episodes
March 23, 2025
09. 解读Anthropic报告：MCP爆火背后的Agent野心
<p>近期，Anthropic发布了一篇名为《Which economic tasks are performed with AI? Evidence from millions Claude conversation》——从数百万条Claude对话中探索AI能更为擅长的工作任务。这份报告给我们带来了一些启发，尤其是关于如何量化模型带来的经济价值，以及大家最为关心的，它们能取代哪些人类工作任务？更准确地说，应该是人们如何跟AI协作的问题，当然这期内容不止报告本身，我们也聊聊这份报告制作方Anthropic这家公司本身。</p><ul> <li><strong>主播</strong></li></ul><p>孟醒：五源资本合伙人，滴滴前自动驾驶COO</p><p>大奎：Looki.AI品牌负责人</p><ul> <li><strong>时间轴&关键概念</strong></li></ul><p>1:28 Anthropic是一家什么样的公司？</p><p>Star Gate</p><p>ADEPT</p><p>3:23 Sonnet 3.7的发布备受关注</p><p>5:07 Sonnet 3.7高达70%的准确率意味着什么？</p><p>imageNET</p><p>AlexNet</p><p>7:30 什么是SWE Bench？</p><p>8:16 报告如何展开研究？</p><p>https://assets.anthropic.com/m/2e23255f1e84ca97/original/Economic_Tasks_AI_Paper.pdf</p><p>Clio：<a href="https://www.anthropic.com/research/clio">www.anthropic.com</a></p><p>9:16 报告第一个结论：AI的使用广度大于深度</p><p>10:53 为什么大部分人使用AI浅尝辄止？</p><p>12:42 AI的垂直化是应该基于工种而不是行业</p><p>MCP</p><p>Cursor</p><p>Augment Code</p><p>Context window</p><p>RAG</p><p>Retrival</p><p>17:10 toB和toC的coding产品，哪一个想象空间更大？</p><p>MarsCode</p><p>19:35 toB行业负反馈大于正反馈</p><p>20:40 Anthropic是在一开始就考虑好做toB生意吗？</p><p>21:30 什么是federation learning联邦学习？</p><p>25:13 Sonnet3.7 为什么不是一个推理模型？</p><p>27:05 Sonnet从3.5升级到3.7对应用公司意味着什么？</p><p>deep research</p><p>30:07 第二个核心结论：极高端职业使用AI频率极低</p><p>Parallel legal</p><p>32:23 对精准度要求极高的行业如何克服模型幻觉问题？</p><p>Harvey AI</p><p>34:19 toB模型服务公司对RAG有极高要求</p><p>35:39 医疗行业和大模型的结合点</p><p>Co-pilot</p><p>Auto-pilot</p><p>41:12 报告第三个核心结论：增强型任务使用率57%高于自动化任务43%</p><p>43:40 重新认知人的需求变化</p><p>46:29 用“杨贵妃吃荔枝”类比人的需求变化</p><p>49:52 co-pilot和auto-pilot产品市场的发展趋势</p><p>55:26 当AI过剩，生产力会如何重新分配?</p><p>57:32 报告背景的补充</p>
February 13, 2025
2025年AI硬件的变量：Agent和供应链创新 | 对话本末科技创始人张笛
<p>本期节目是孟醒受邀参加《五源小酒馆》播客录制，与五源资本投资人贺开颜、本末科技创始人张笛聊聊未来一年间，技术进步会撬动哪些硬件变革以及未来的机会会在哪里，以下是本次对谈实录。</p><ul> <li><strong>主播</strong></li></ul><p>贺开颜 五源资本投资人</p><ul> <li><strong>嘉宾</strong></li></ul><p>张笛 本末科技创始人</p><p>孟醒 五源资本合伙人、滴滴前自动驾驶COO</p><ul> <li><strong>时间轴</strong></li></ul><p>3:35 眼镜、指环、代步都是新兴的硬件领域，针对“指环”展开讲讲</p><p>8:05 2025CES中最让人眼前一亮的产品形态是什么？</p><p>11:25 2025年的大变量：Agent和供应链创新。</p><p>15:16 机器人泛化能力上装vs.通用类操作设备</p><p>18:57 2025年三个关于机器人的判断</p><p>21:09 针对AR眼镜的的解读</p><p>26:06 眼镜长期合适创业公司吗？</p><p>34:47 能高效调用大模型能力的硬件究竟是什么？</p><p>38:35 我们在陪伴机器人上把“情感”和“功能”区分太开</p><p>41:40 情感陪伴需要有足够的留白</p><p>44:27 Cat person vs. Dog person?</p><p>50:39 AI的情感陪伴是副产品</p><p>56:28下一代的超级硬件会出现在哪里？</p>
November 8, 2024
07.对scaling law不乐观，o1很难创造新概念，我们距离大模型共识有多远？
<p>本期节目是孟醒受邀参加TGO鲲鹏会与张俊伟博士和田渊栋博士的对谈，他们围绕大模型领域的scaling Law、涌现能力、大模型的双系统应用以及AI Agent话题并结合当下多个研究展开讨论。</p><ul> <li><strong>主播</strong></li></ul><p>张俊伟 IEEE T-CSVT Associate Editior、TGO 鲲鹏会 硅谷董事</p><ul> <li><strong>嘉宾</strong></li></ul><p>田渊栋 MetaAI研究团队（FAIR）研究科学家总监、卡耐基梅隆大学机器人系博士</p><p>孟醒 五源资本合伙人、滴滴前自动驾驶COO</p><ul> <li><strong>时间轴</strong></li></ul><p>3:58 无人驾驶中是否存在scaling law效应？</p><p>9:12 对scaling law并不乐观至少在目前看来大模型的能力并不会断崖式地变好</p><p>12:14 田渊栋博士的新工作：神经网络的解可以通过某种代数的方式构造出来</p><p>18:30 Dualformer模型中的双系统system1和system2的分工</p><p>21:36 Apple GSM和Danny Zhou对于COT的分歧和争议</p><p>24:24 大模型无法做到过滤无关信息进行推理</p><p>29:04 大语言模型的物理学：一个小变量的改变能影响大语言模型的逻辑</p><p>31:28 大模型能抽象出数学中的定义吗？</p><p>37:22 学界和产业中目前AI agent是什么样的进展？</p><ul> <li><strong>提及</strong></li></ul><p>Scaling Law｜涌现效应｜Dualformer｜Apple GSM｜Danny Zhou｜Allen Zhu</p><h1>Composing Global Optimizers to Reasoning Tasks via Algebraic Objects in Neural Nets</h1><p><a href="https://arxiv.org/search/cs?searchtype=author&query=Tian,+Y">Yuandong Tian</a></p><p>Dualformer: Controllable Fast and Slow Thinking by Learning with Randomized Reasoning Traces</p><p><a href="https://arxiv.org/search/cs?searchtype=author&query=Su,+D">DiJia Su</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Sukhbaatar,+S">Sainbayar Sukhbaatar</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Rabbat,+M">Michael Rabbat</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Tian,+Y">Yuandong Tian</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Zheng,+Q">Qinqing Zheng</a></p><p>Chain of Thought Empowers Transformers to Solve Inherently Serial Problems</p><p><a href="https://arxiv.org/search/cs?searchtype=author&query=Li,+Z">Zhiyuan Li</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Liu,+H">Hong Liu</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Zhou,+D">Denny Zhou</a>, <a href="https://arxiv.org/search/cs?searchtype=author&query=Ma,+T">Tengyu Ma</a></p><p>Physics of Language Models: <a href="https://physics.allen-zhu.com/">physics.allen-zhu.com</a></p><ul> <li><strong>后期：</strong>大奎</li></ul>
10 total episodes available
Similar Podcasts
Discover related shows you might enjoy
Deep-dive analytics for 技术不无聊
Frequently asked questions
Have a different question and can't find the answer you're looking for? Reach out to our support team by sending us an email and we'll get back to you as soon as we can.
- What is 技术不无聊?
- How often does this podcast release new episodes?
This podcast updates bi-weekly.
- Where can I listen to this podcast?
This podcast is available on 6 platforms including Apple Podcasts, Spotify, and more. You can also use the RSS feed directly.
- Does this podcast accept guests?
Yes, this podcast regularly features guests.
Legal Disclaimer
Pod Engine is not affiliated with, endorsed by, or officially connected with any of the podcasts displayed on this platform. We operate independently as a podcast discovery and analytics service.
All podcast artwork, thumbnails, and content displayed on this page are the property of their respective owners and are protected by applicable copyright laws. This includes, but is not limited to, podcast cover art, episode artwork, show descriptions, episode titles, transcripts, audio snippets, and any other content originating from the podcast creators or their licensors.
We display this content under fair use principles and/or implied license for the purpose of podcast discovery, information, and commentary. We make no claim of ownership over any podcast content, artwork, or related materials shown on this platform. All trademarks, service marks, and trade names are the property of their respective owners.
While we strive to ensure all content usage is properly authorized, if you are a rights holder and believe your content is being used inappropriately or without proper authorization, please contact us immediately at hey@podengine.ai for prompt review and appropriate action, which may include content removal or proper attribution.
By accessing and using this platform, you acknowledge and agree to respect all applicable copyright laws and intellectual property rights of content owners. Any unauthorized reproduction, distribution, or commercial use of the content displayed on this platform is strictly prohibited.