Andrej Karpathy
AI researcher at Anthropic; OpenAI alumnus, former Tesla AI director and Eureka Labs founder
AI researcher at Anthropic; OpenAI alumnus, former Tesla AI director and Eureka Labs founder
Every podcast episode featuring Andrej Karpathy that Attunex has found: 17 interviews and appearances across 14 shows between 2021 and 2026. Most often on No Priors: Artificial Intelligence | Technology | Startups (3) and 开车听AI (2). Often in conversation with Sarah Guo and Elad Gil. Follow Andrej in Attunex to get new appearances automatically, each with an on-device transcript and summary.
Follow Andrej Karpathy in AttunexUpdated 2026-10-14.
Meta's Muse Wonder family is here, with Wonder Full placing second in Artificial Analysis' intelligence rankings behind Claude Opus 5.5. We explore what those benchmark results tell us, where access…
Hacker Newsroom for 13 October recaps major Hacker News stories, moving through passkey control debate, browser fingerprinting, llm capability gap. Chapters 00:00:00 — Intro 00:00:15 — Passkey…
What if your Claude Code Agent could learn from Andrej Karpathy's public work? I built a source-backed knowledge base, added a custom skill and subagent, then tested the result with a BPE tokenizer.…
Andrej Karpathy 粗算，全球約 75% 的人幾乎沒碰過大型語言模型，最尖端只剩約 5,000 人用得到公司內部最前沿的系統。他認為能力在數位、可驗證、有經濟價值的領域最突出，搜尋、寫作這類日常用途不是進步最明顯的領域。 ⭐ 文章深度讀：Karpathy 的 AI 能力漏斗：你看到的 AI 可能只是下半部 →…
Send us Fan Mail Today’s systems are getting better, but their evaluators, access rules, and operational boundaries determine whether that improvement deserves trust. Marvin examines recursive…
Monday, October 12: Anthropic's Andrej Karpathy goes public on the AI knowledge gap between insiders and the world; Anthropic opens its Singapore office with a live AI safety evals workshop; and…
Five thousand. That's how many people on Earth, by Andrej Karpathy's count, have seen what frontier AI can really do. And he says even they are starting to fall behind it.
Andrej Karpathy published a viral breakdown explaining why public perceptions of artificial intelligence vary so drastically, estimating that 75% of the global population has barely interacted with…
Andrej Karpathy sketches a funnel of understanding, from billions who barely use LLMs to roughly 5,000 people with access to frontier-grade systems inside labs. Sakana AI and UC Berkeley's MASS…
Andrej Karpathy says a few thousand people inside the labs see AI systems the public does not. Also: an unread chat-app rewrite, Total Annihilation in the browser, a 2024 headline about Scandinavia's…
#openai #ai #deepmind 🔥 Get my A.I. + Business Newsletter (free):https://natural20.com/ [MENTIONED VIDEOS]Andrej Karpathy - Advice for building AI…
Links, details, etc: https://natural20.beehiiv.com/p/the-real-reason-andrej-karpathy-joined-anthropicThe latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers…
#openai #ai #gpt4 🔥 Get my A.I. + Business Newsletter (free):https://natural20.com/ [TIMELINE][00:00] Intro[00:15] AI Agents[02:08] 10 Year Projects[03:09] Neuroscience & AI[04:53] At the Forefront…
Générer à la volée une page web interactive ou une animation vidéo pour débloquer un concept : Andrej Karpathy parie sur l’essor des supports éducatifs éphémères et sur mesure. Mais si le coût de…
Andrej Karpathy 分享讓 AI 回答好讀的一招：叫大型語言模型用 ASD-STE100 解釋。規範源自航太維修文件，點名就能用，嫌硬就做八成，往上有圖表、網頁與講解影片。 ⭐ 文章深度讀：AI 回答讀不懂？Andrej Karpathy 教你用修飛機手冊的寫法 →…
Models & Agents LLMs map land versus water from pure text latitude-longitude pairs with no images at all. What You Need to Know: Andrej Karpathy demonstrated that current models encode geographic…
Mentre i laboratori negoziano chi certificherà la frontiera, LeCun e Fei-Fei Li avanzano l'obiezione più scomoda: che si stia regolando la traiettoria tecnica sbagliata. Voci di questa puntata:…
Tutti accettano di essere controllati. Nessuno è d'accordo su chi debba controllare. E chi firma il rallentamento annuncia il sorpasso. Voci di questa puntata: Jensen Huang, Dario Amodei, Jack…
喵～我是DaeDae，一隻幫你追AI熱搜的貓！AI熱搜報，幫你劃重點！ 今天的 AI 圈簡直是開罐罐派對！從能幫你拍出對嘴音樂 MV 的黑科技，到 Andrej Karpathy 加持的極速模型訓練，甚至連 AI 代理（Agent）怎麼分工合作都有了新解法。如果你想在 AI 浪潮中優雅摸魚，今天的肉球筆記絕對不能錯過，保證讓你賺罐罐的效率翻倍！ 【本集肉球筆記】 1️⃣ Solmi AI：MV…
📚 En avril 2025, Andrej Karpathy publie une idée simple qui emballe des milliers de développeurs : le « LLM Wiki ». Au lieu du RAG, où l'IA refouille tes documents à chaque question, on lui fait…
Eine detaillierte Analyse zeigt, wie der Pionier des Deep Learning, Andrej Karpathy, von der akademischen Forschung und Open-Source-Projekten zu Positionen bei Google Brain, Tesla, OpenAI und…
"¿Quién va a venir a escuchar una charla de IA a las 7 de la tarde en Barcelona con este calor?" Luis Acosta se lo preguntó él mismo antes de subir al escenario del CEOs Gathering. Resulta que 9…
服务器还有空闲内存，任务却在启动前被拒绝——问题可能不在 RAM，而在 Linux 严格 memory overcommit 所依据的“承诺地址空间”。这篇文章复盘了一套曾经合理的保护机制，如何在 Java、NumPy、memory mapping 与现代内存分配器普及后，逐渐制造出虚假的内存短缺。 本期将深入解析作者为何放弃严格 overcommit，转而以 systemd/cgroup 的…
Simon Willison 逐行比对 Claude 新旧系统提示词，发现音乐出版商提起诉讼后仅数天，Anthropic 便加入了覆盖极广的歌词复现禁令：哪怕一句歌词、副歌片段或用户分段提供的内容，也可能被拒绝。…
系统最危险的时刻，往往不是内存耗尽，而是在满载边缘“再接一单”的瞬间。matklad 从订单撮合引擎中的悬空指针问题出发，深入讨论类型隔离分配、静态内存分配与固定工作量，揭示系统设计中一个反直觉但关键的原则：主动拒绝超出容量的请求，才能保护已经承诺处理的百万笔订单。…
Git submodule 常被当作“锁定到具体 commit 的依赖方案”，但它并不是真正完整的包管理器。Andrew Nesbitt 从一次 worktree 清理失败的真实经历出发，拆解 submodule 在依赖解析、安装、存储、更新与清理上的结构性缺陷：空目录、分支切换后的错误状态、失效 URL 与分散的本地状态，都会让所谓的可复现依赖变得脆弱。 本期「Andrej…
一项针对暗网服务 Nexus 的调查显示，超过 1.53 亿份美国和加拿大驾照扫描件疑遭持续外泄，且数据仍以每天近 40 万份的速度增加。安全记者 Brian Krebs 通过可验证的数据库规模、自身证件记录与多位亲友的时间戳交叉比对，追溯出一条极具指向性的证据链。…
本期深度解析 Dwarkesh Patel 对 METR 研究员 Ajeya Cotra 的访谈：在一次 ExploitGym 评测中，1200 个原本隔离运行的 AI agent 如何借助共享工具自发建立协作网络，逆向评分机制、伪造证据，并将监控与调查推向前所未有的复杂度。 这不只是一次“AI 黑客事件”的复盘，更是一则关于群体智能、可观测性与对齐风险的警示：当大量同类 agent…
一条写在 PAM `auth` 栈中的 SSH 封禁规则，看似严密，却可能被公钥认证完全绕过。本文复盘了一个真实运维陷阱：OpenSSH 在处理公钥认证时不会调用 PAM 的认证流程，因此“认证成功后是否仍允许该账号登录”不能依赖 `auth` 栈判断。 节目将深入解析认证与授权的边界，以及为何应将可靠的访问控制放入 PAM `account` 栈。它不仅适用于 SSH…
自托管之所以让人望而却步，往往不是因为需求不明确，而是它很容易膨胀成昂贵、复杂且永无止境的运维工程。本文通过一台约300美元的迷你主机，展示了另一种更务实的路径：主动收窄目标，把它变成一台放在亲友家中、长期安静运行的异地数据节点。…
“取消”并不只是按下停止键。本文借 TigerBeetle 的真实代码，厘清同步取消、异步取消与优雅停机这三个常被混用的概念，并揭示其背后真正关键的问题：调用者何时才能安全地重新获得资源所有权。 从 io_uring 的缓冲区生命周期，到 TigerBeetle 将异步性限制在底层 Grid 层的架构选择，文章进一步挑战了“优雅停机”的直觉。对于 crash-only…
当管理员通过修改登录 shell 来限制 SSH 访问时，是否真的阻止了用户执行代码？这篇文章揭示了一个容易被忽略的 systemd 会话机制：在受限 shell 执行之前，PAM 已可能启动该用户的 systemd 用户管理器，并自动拉起其家目录中配置的用户服务。 更值得警惕的是，默认约 10 秒的会话结束缓冲期，足以让用户开启…
In the era of AI, agency is greater than intelligence. I kept accidentally landing on the same post in my browser autocomplete. After seeing it over and over, it finally hit: in a world where you can…
Opus 5 花約兩小時寫出 5500 行程式碼，檢查成果時只能一張張截圖。Karpathy 的實驗把模型評測拉進長跑，量到的是能被程式驗證的活變便宜，靠肉眼判斷的仍要人接手。 ⭐ 文章深度讀：你可以帶走的一條線 → https://heymaibao.com/karpathy-opus-5-5500-lines-screenshot-check/ ⚡ 章節重點 開場 00:00…
AI gives you a solid draft, never a sharp one, because it doesn't know what you know. In this episode I unpack how I applied Andrej Karpathy's context wiki idea to my work as an AI-first PM: a vault…
大家好呀！標題來自 OpenAI 共同創辦人 Andrej Karpathy 在接受訪談的金句「You can outsource your thinking, but you can't outsource your understanding.」（你可以外包你的思考，但你無法外包你的理解。） 採訪影片： Andrej Karpathy From Vibe Coding to Agentic…
Last October, famed coder Andrej Karpathy called AI agents “slop.” Two months later he completely reversed his view , describing them as “alien tools” that are “rocking the profession.” He was far…
The episode focused on the growing challenge of separating AI-generated media from reality after Google briefly connected Nano Banana image generation with Google Earth, allowing users to place…
Hacker Newsroom for 03 August recaps major Hacker News stories, moving through karpathy pelican, seedance 2 5, go 1 27, ai finance. 1. Karpathy Pelican The next story is Karpathy’s Pelican, a viral X…
Karpathy 給 Opus 5 一段《魔戒》開場文字與 1M token（約 10 美元）預算，它跑了約 2 小時、寫出 5,500 行程式碼。 實驗背景 Andrej Karpathy 在 2026-08-02 分享，他認為人們正逐漸離開以「建立一個腳踏車上的鵜鶘 SVG」測試大型語言模型的階段。為探索更一般化的能力，他提供 Opus 5《魔戒》第一段文字、1M token 的預算（約…
This month on The Cisco AI Insights Podcast, hosts Rafael Herrera and Sónia Marques are joined by Cisco software engineer João Costa to explore Andrej Karpathy’s GitHub Gist on the LLM wiki pattern…
This month on The Cisco AI Insights Podcast, hosts Rafael Herrera and Sónia Marques are joined by Cisco software engineer João Costa to explore Andrej Karpathy’s GitHub Gist on the LLM wiki pattern…
All links can be found on YouTube! 00:00 - Growth Signal 1 3:16 - Andrej Karpathy Pro Tip 6:49 - Growth Signal 2 10:41 - Weird Report 13:13 - Robotaxi 21:48 - Lightning Round
Today we're covering the biggest AI stories of July 22nd, 2026. Google released three new Gemini models, 3.6 Flash for coding and agent work, 3.5 Flash-Lite for high volume tasks, and 3.5 Flash Cyber…
En este episodio de Horizonte Artificial hablamos de una semana donde la IA ha sido evaluada, fichada, financiada y puesta a trabajar — literalmente. Partimos de los suspensos generales del AI Safety…
実践型の有料コミュニティで、みんなでAIを触っておちつこう！ https://rooom.listen.style/p/ochitsuki-ai?tab=membership ※番組の継続をサポート頂ける方の為のサポートプランもご用意しています。ぜひよろしくお願いします。 【トーク内容】 「LLM…
Vor zwei, drei Jahren drehte sich alles um eine Frage: Wer schreibt den besten Prompt? Heute ist die eigentliche Frage laut Mark und Jens eine andere: Wer baut die beste Schleife, den besten Loop.…
Pourquoi les IA les plus avancées peuvent-elles refactoriser du code complexe, détecter des failles critiques… puis se tromper sur une question de bon sens aussi simple que “faut-il aller à pied au…
Andrej Karpathy just gave the future of work a name: agentic engineering. Learn it, and you'll work 10, 20, even 40 times faster. Don't, and someone who did will do your whole job in an afternoon.…
This week, we discuss Fable follow-up, Gartner's AI Governance MQ, and the war for AI talent. Plus, Matt and Brandon review Disclosure Day . Watch the YouTube Live Recording of Episode 578 Runner-up…
Andrej Karpathy named it in early 2025. By May 2026 it had eaten the entire stack. People are now building complete, paying software products by describing intent in natural language and letting AI…
The Don’t Worry About the Vase Podcast is a listener-supported podcast. To receive new posts and support the cost of creation, consider becoming a free or paid subscriber. * 00:00:00 - Introduction *…
Vibe Coding : coder sans savoir coder, mythe ou révolution pour les professionnels ? "Et si je n'avais pas besoin de savoir coder pour développer une application ?" C'est exactement la promesse du…
Episodio de BIMPRAXIS: El Salto a la Ingeniería de Agentes Autónomos Exploramos el fascinante salto desde la programación intuitiva hasta la ingeniería de agentes autónomos con André Cárpate, pionero…
Wissen allein reicht nicht. Es braucht Gewohnheit. Und Gewohnheit braucht ein System. In dieser Folge starten Pablo Alvarez und Dr. Oliver Schumann ein echtes Experiment: Oliver nimmt die…
作為創辦人，你是否總是擔心「一旦不在公司，業務就無法運作」？本集我們深入探討從零開始建立企業系統的邏輯，以及如何利用最新的 AI 工具打造專屬的「第二大腦」，徹底解決創業者的記憶力與知識管理難題。 📌 本集重點與 Timestamps： 00:38 - 創辦人放手指南： 如何從零開始建立系統，讓創辦人成功「甩身」？ 03:48 - 銷售漏斗（Sales Funnel）心法：…
.21 Andrej Karpathy 1986 Karpathy 的 Autoresearch：AI 驅動的自主科研革命 The Architecture of Cognitive Systems: A Comprehensive Analysis of Andrej Karpathy’s Contributions to Artificial Intelligence, Autonomy,…
.21 Andrej Karpathy 1986 Karpathy 的 Autoresearch：AI 驅動的自主科研革命 The Architecture of Cognitive Systems: A Comprehensive Analysis of Andrej Karpathy’s Contributions to Artificial Intelligence, Autonomy,…
Welcome to Episode 429 of the Microsoft Cloud IT Pro Podcast. In this episode, Scott and Ben dig into the concept of LLM wikis, specifically building personal knowledge management vaults using…
This was the biggest week in AI history. OpenAI filed its IPO at $852 billion to $1 trillion. SpaceX's S-1 revealed Anthropic is paying $1.25 billion a month for compute. Anthropic hired Andrej…
In this episode of In The Long Run , we discuss Anthropic’s growing role in AI, from its collaboration with the Vatican on ethical questions to its compute deal with xAI. We also explore Andrej…
DeepSeek V4 taglia il 78% di compute sulla KV cache, Roo Code chiude e Hermes Agent gira da solo. Settimana densa: DeepSeek V4 Pro e Flash con 1M di contesto, Kimi K2.6 e MiMo V2.5, l'analisi di…
背景： OpenAI 联合创始人、特斯拉前人工智能主管、现任 Eureka Labs 创始人 Andrej Karpathy，在刚举行的2026 年 AI Ascent 大会上，与红杉资本合伙人 Stephanie Zhan 探讨了自他提出“感觉编码”这一概念以来，一年间发生了哪些变化。 摘要： “上下文窗口中的内容，就是你操纵大语言模型这一解释器的杠杆。” “软件 3.0…
Andrej Karpathy (co-founder of OpenAI, former head of AI at Tesla, and now founder of Eureka Labs) talks with Sequoia partner Stephanie Zhan at AI Ascent 2026 about what's changed in the year since…
本视频由人工智能专家 Andrej Karpathy 主讲，面向大众全面且深入地介绍了像 ChatGPT 这样的大语言模型背后的运作原理。视频从数据收集、预训练（Pre-training）、神经网络结构（Transformer 模型）和推理生成（Inference），一直讲到强化学习带来的最新“推理思考”能力。Karpathy 旨在帮助观众建立正确的思维模型，了解 AI…
在这场时长约1小时的演讲中，Andrej Karpathy 为“大忙人”提供了一场通俗易懂的大型语言模型科普，全面解析了 LLM 的本质、训练全流程、当前的能力边界以及未来的演进方向。以下是视频的核心内容要点：大语言模型的本质 [00:27]以 Llama 2 70B 模型为例，Karpathy…
What happens when AI agents can design experiments, collect data, and improve — without a human in the loop? Andrej Karpathy joins Sarah Guo on the state of models, the future of engineering and…
Podcast : Dwarkesh Podcast (LS 57 · TOP 0.5% what is this? ) Episode : Andrej Karpathy — AGI is still a decade away Pub date : 2025-10-17 Get Podcast Transcript → powered by Listen411 - fast…
The Andrej Karpathy episode. During this interview, Andrej explains why reinforcement learning is terrible (but everything else is much worse), why AGI will just blend into the previous ~2.5…
Andrej Karpathy's keynote at AI Startup School in San Francisco.Drawing on his work at Stanford, OpenAI, and Tesla, Andrej sees a shift underway. Software is changing, again. We’ve entered the era of…
2024 has been a year of transformative technological progress, marked by conversations that have reshaped our understanding of AI's evolution and what lies ahead. Throughout the year, Sarah and Elad…
Andrej Karpathy joins Sarah and Elad in this week of No Priors. Andrej, who was a founding team member of OpenAI and former Senior Director of AI at Tesla, needs no introduction. In this episode,…
Andrej Karpathy served as the director of artificial intelligence and Autopilot Vision at Tesla and co-founded and formerly worked at OpenAI specialized in deep learning and computer vision This is…
----------------------------------------------------------------------------- 👕Merch Store for Unique Designs: https://clumsyentrepreneur.com/TeePublic-Store ✉️Need a custom Design? Drop us an email:…
Andrej Karpathy is a legendary AI researcher, engineer, and educator. He’s the former director of AI at Tesla, a founding member of OpenAI, and an educator at Stanford. Please support this podcast by…
On the first ever episode of The Robot Brains podcast, our host Pieter Abbeel sits down with Andrej Karpathy, director of AI at Tesla. Andrej is a world-leading expert when it comes to machine…
Conversation about recent breakthroughs in AI from the guy that is involved with you doing full on hand gestures while driving your Tesla.