{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/16f3351475f94dfbafe9ba850c4cf50f\" frameborder=\"0\" width=\"1790\" height=\"1342\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1342,"width":1790,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1342,"thumbnail_width":1790,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/16f3351475f94dfbafe9ba850c4cf50f-7c70446e0dfb4f55.gif","duration":157.276,"title":"Recursive Self Improving AI, Explained in Two Years","description":"This Loom discusses building superintelligence through recursive self-improvement in AI. The speaker notes a recent funding round of $650 million at a $4.65 billion valuation aimed at one goal: creating an AI that improves itself and goes beyond human capability, with a timeline of within about two years. They explain the concept using the scientific method, where an AI identifies its own shortcomings, builds a new version of itself, and loops that process to conduct research on itself. They also clarify how recursive self-improvement works by having the AI generate a different new version in each iteration."}