{"id":23301,"date":"2025-11-16T12:09:22","date_gmt":"2025-11-16T20:09:22","guid":{"rendered":"http:\/\/jnack.com\/blog\/?p=23301"},"modified":"2025-11-17T10:04:41","modified_gmt":"2025-11-17T18:04:41","slug":"a-brief-history-of-the-world-models","status":"publish","type":"post","link":"http:\/\/jnack.com\/blog\/2025\/11\/16\/a-brief-history-of-the-world-models\/","title":{"rendered":"A Brief History of the World (Models)"},"content":{"rendered":"\n<p>On Friday I got to meet <a href=\"https:\/\/en.wikipedia.org\/wiki\/Fei-Fei_Li\">Dr. Fei-Fei Li,<\/a> &#8220;the godmother of AI,&#8221; at the launch party for her new company, World Labs (see her launch <a href=\"https:\/\/www.worldlabs.ai\/blog\/marble-world-model\">blog post<\/a>). We got to chat a bit about a paradox of complexity: that as computer models for perceiving &amp; representing the world grow massively more <strong>sophisticated<\/strong>, the interfaces for doing common things\u2014e.g. moving a person in a photo\u2014can get <strong>radically simpler<\/strong> &amp; more intentional. I&#8217;ll have more to say about this soon.<\/p>\n\n\n\n<p>Meanwhile, here&#8217;s her fascinating &amp; wide-ranging conversation with Lenny Rachitsky. I&#8217;m always a sucker for a good Platonic allegory-of-the-cave reference. \ud83d\ude42<\/p>\n\n\n\n<figure class=\"wp-block-embed is-type-video is-provider-youtube wp-block-embed-youtube wp-embed-aspect-16-9 wp-has-aspect-ratio\"><div class=\"wp-block-embed__wrapper\">\n<iframe loading=\"lazy\" title=\"The Godmother of AI on jobs, robots &amp; why world models are next | Dr. Fei-Fei Li\" width=\"604\" height=\"340\" src=\"https:\/\/www.youtube.com\/embed\/Ctjiatnd6Xk?feature=oembed\" frameborder=\"0\" allow=\"accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share\" referrerpolicy=\"strict-origin-when-cross-origin\" allowfullscreen><\/iframe>\n<\/div><\/figure>\n\n\n\n<p>From the YouTube summary:<\/p>\n\n\n\n<p>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk\">00:00<\/a>) Introduction to Dr. Fei-Fei Li <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=331s\">05:31<\/a>) The evolution of AI <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=577s\">09:37<\/a>) The birth of ImageNet <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=1045s\">17:25<\/a>) The rise of deep learning <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=1433s\">23:53<\/a>) The future of AI and AGI <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=1791s\">29:51<\/a>) Introduction to world models <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2445s\">40:45<\/a>) The bitter lesson in AI and robotics <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2882s\">48:02<\/a>) Introducing Marble, a revolutionary product <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=3060s\">51:00<\/a>) Applications and use cases of Marble <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=3661s\">01:01:01<\/a>) The founder\u2019s journey and insights <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=4205s\">01:10:05<\/a>) Human-centered AI at Stanford <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=4464s\">01:14:24<\/a>) The role of AI in various professions <br>(<a href=\"https:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=4696s\">01:18:16<\/a>) Conclusion and final thoughts<\/p>\n\n\n\n<p>And here&#8217;s Gemini&#8217;s solid summary of their discussion of world models:<\/p>\n\n\n\n<ul>\n<li><strong>The Motivation:<\/strong> While LLMs are inspiring, they lack the <strong>spatial intelligence<\/strong> and <strong>world understanding<\/strong> that humans use daily. This ability to reason about the physical world\u2014understanding objects, movement, and situational awareness\u2014is essential for tasks like first response or even just tidying a kitchen <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=1943\">32:23<\/a>.<\/li>\n\n\n\n<li><strong>The Concept:<\/strong> A world model is described as the <strong>lynchpin<\/strong> connecting visual intelligence, robotics, and other forms of intelligence beyond language <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2012\">33:32<\/a>. It is a foundational model that allows an agent (human or robot) to:\n<ul>\n<li><strong>Create<\/strong> worlds in their mind&#8217;s eye through prompting <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2101\">35:01<\/a>.<\/li>\n\n\n\n<li><strong>Interact<\/strong> with that world by browsing, walking, picking up objects, or changing things <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2112\">35:12<\/a>.<\/li>\n\n\n\n<li><strong>Reason<\/strong> within the world, such as a robot planning its path <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2131\">35:31<\/a>.<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>The Application:<\/strong> World models are considered the <strong>key missing piece<\/strong> for building effective embodied AI, especially robots <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2168\">36:08<\/a>. Beyond robotics, the technology is expected to unlock major advances in scientific discovery (like deducing 3D structures from 2D data) <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2268\">37:48<\/a>, games, and design <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2251\">37:31<\/a>.<\/li>\n\n\n\n<li><strong>The Product:<\/strong> Dr. Li co-founded <strong>World Labs<\/strong> to pursue this mission <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2065\">34:25<\/a>. Their first product, <strong>Marble<\/strong>, is a generative model that outputs genuinely <strong>3D worlds<\/strong> which users can navigate and explore <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=2951\">49:11<\/a>. Current use cases include virtual production\/VFX, game development, and creating synthetic data for robotic simulation <a target=\"_blank\" rel=\"noreferrer noopener\" href=\"http:\/\/www.youtube.com\/watch?v=Ctjiatnd6Xk&amp;t=3185\">53:05<\/a>.<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>On Friday I got to meet Dr. Fei-Fei Li, &#8220;the godmother of AI,&#8221; at the launch party for her new company, World Labs (see her launch blog post). We got to chat a bit about a paradox of complexity: that as computer models for perceiving &amp; representing the world grow massively more sophisticated, the interfaces [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":[],"categories":[18,66],"tags":[],"_links":{"self":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/23301"}],"collection":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/comments?post=23301"}],"version-history":[{"count":4,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/23301\/revisions"}],"predecessor-version":[{"id":23317,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/posts\/23301\/revisions\/23317"}],"wp:attachment":[{"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/media?parent=23301"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/categories?post=23301"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/jnack.com\/blog\/wp-json\/wp\/v2\/tags?post=23301"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}