--- title: Richard Sutton leaves Carmack to start his own AI lab description: Richard Sutton, co-founder of modern reinforcement learning and 2024 Turing Award winner, has left John Carmack's Keen Technologies to start Oak Lab with... canonical: https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Richard Sutton leaves Carmack to start his own AI lab | daily.dev og:description: Richard Sutton, co-founder of modern reinforcement learning and 2024 Turing Award winner, has left John Carmack's Keen Technologies to start Oak Lab with... og:url: https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy og:image: https://api.daily.dev/og/posts/l7jfhlgjy.png og:image:alt: Richard Sutton leaves Carmack to start his own AI lab og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Richard Sutton leaves Carmack to start his own AI lab **[The Next Web](https://daily.dev/sources/tnw)** · 4 min read · 1 upvotes · 0 comments ## Summary Richard Sutton, co-founder of modern reinforcement learning and 2024 Turing Award winner, has left John Carmack's Keen Technologies to start Oak Lab with collaborator Khurram Javed. The new lab argues that current deep learning methods are fundamentally flawed because they learn from curated datasets rather than real-time experience. Oak Lab is developing algorithms like NetworkIDBD that selectively assign credit to meaningful signals rather than absorbing noise like SGD does. Their stated goal is a trillion-parameter agent that learns continuously in real time on just 20 watts — comparable to the human brain — rather than training once in megawatt data centers. Sutton joins Yann LeCun and David Silver in betting that experience-based learning, not scaling LLMs, is the path to real intelligence. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why did Richard Sutton leave John Carmack's Keen Technologies? Richard Sutton left Keen Technologies to start his own AI lab, Oak Lab, with collaborator Khurram Javed, pursuing what he called a slightly different path toward understanding intelligence. He believes current deep-learning methods are weak and inefficient and need fundamentally new ideas rather than more tweaks, a view he shared publicly on X. _Track how leading researchers are rethinking AI architecture beyond LLM scaling on daily.dev._ ### What is NetworkIDBD and how does it differ from SGD or Adam optimizers? NetworkIDBD is a neural version of Sutton's older IDBD algorithm that assigns credit selectively during learning, rewarding only signals that actually predict something. Standard optimizers like SGD and Adam spread the blame for every error across all parameters, so they absorb noise instead of filtering it out, while NetworkIDBD learns the real pattern where SGD gets drowned by junk data. _Developers evaluating new training methods can follow optimizer research like this on daily.dev._ ### What is Oak Lab's goal for energy-efficient AI compared to today's frontier models? Oak Lab aims to build a trillion-parameter agent that learns and plans in real time on roughly 20 watts, about what the human brain consumes. This contrasts with today's frontier models, which train once in data centers drawing megawatts of power and then remain frozen, rather than continuously learning from experience. _Anyone weighing compute costs against model capability can follow efficiency-focused AI research on daily.dev._ ## Similar posts on daily.dev - [A Turing Award winner says the industry’s fix for running out of data is ‘a big mistake’](https://daily.dev/posts/a-turing-award-winner-says-the-industry-s-fix-for-running-out-of-data-is-a-big-mistake--a5ss4ktbt) · The Next Web · 0 upvotes · 0 comments - [AI research lab NeoCognition lands $40M seed to build agents that learn like humans](https://daily.dev/posts/ai-research-lab-neocognition-lands-40m-seed-to-build-agents-that-learn-like-humans-abxahezpu) · TechCrunch · 0 upvotes · 0 comments - [NVIDIA, Ineffable Intelligence Team Up to Build the Future of Reinforcement Learning Infrastructure](https://daily.dev/posts/nvidia-ineffable-intelligence-team-up-to-build-the-future-of-reinforcement-learning-infrastructure-sv52l7a22) · NVIDIA · 0 upvotes · 0 comments - [From MIT to IBM, expediting AI and quantum deployment](https://daily.dev/posts/from-mit-to-ibm-expediting-ai-and-quantum-deployment-zojfff78d) · MIT News · 0 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#deep-learning](https://daily.dev/tags/deep-learning), [#reinforcement-learning](https://daily.dev/tags/reinforcement-learning) [View this post on daily.dev](https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Richard Sutton leaves Carmack to start his own AI lab","url":"https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy"},"datePublished":"2026-07-14T12:15:37.863Z","dateModified":"2026-09-14T09:08:56.660Z","description":"Richard Sutton, co-founder of modern reinforcement learning and 2024 Turing Award winner, has left John Carmack's Keen Technologies to start Oak Lab with...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/8cc51f5ab3398d2eb02eaa0af819c899?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/8cc51f5ab3398d2eb02eaa0af819c899?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,deep-learning,reinforcement-learning","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Richard Sutton leaves Carmack to start his own AI lab"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/richard-sutton-leaves-carmack-to-start-his-own-ai-lab-l7jfhlgjy#faq","mainEntity":[{"@type":"Question","name":"Why did Richard Sutton leave John Carmack's Keen Technologies?","acceptedAnswer":{"@type":"Answer","text":"Richard Sutton left Keen Technologies to start his own AI lab, Oak Lab, with collaborator Khurram Javed, pursuing what he called a slightly different path toward understanding intelligence. He believes current deep-learning methods are weak and inefficient and need fundamentally new ideas rather than more tweaks, a view he shared publicly on X. Track how leading researchers are rethinking AI architecture beyond LLM scaling on daily.dev."}},{"@type":"Question","name":"What is NetworkIDBD and how does it differ from SGD or Adam optimizers?","acceptedAnswer":{"@type":"Answer","text":"NetworkIDBD is a neural version of Sutton's older IDBD algorithm that assigns credit selectively during learning, rewarding only signals that actually predict something. Standard optimizers like SGD and Adam spread the blame for every error across all parameters, so they absorb noise instead of filtering it out, while NetworkIDBD learns the real pattern where SGD gets drowned by junk data. Developers evaluating new training methods can follow optimizer research like this on daily.dev."}},{"@type":"Question","name":"What is Oak Lab's goal for energy-efficient AI compared to today's frontier models?","acceptedAnswer":{"@type":"Answer","text":"Oak Lab aims to build a trillion-parameter agent that learns and plans in real time on roughly 20 watts, about what the human brain consumes. This contrasts with today's frontier models, which train once in data centers drawing megawatts of power and then remain frozen, rather than continuously learning from experience. Anyone weighing compute costs against model capability can follow efficiency-focused AI research on daily.dev."}}]} ```