Decoder with Nilay Patel Decoder with Nilay Patel

Kevin Roose is AGI-pilled. Why aren't you?

Oct 26, 2026 · 1h 8m

Summary

Kevin Roose discusses his new book, The AGI Chronicles, and his departure from the New York Times to launch the video-first podcast Machine Gods. The conversation explores the AI industry’s shift from research to commercial enterprise, highlighting the intense competition between OpenAI, Anthropic, and Google. Roose analyzes the "AGI" belief driving these labs, the risks of existential threats, and the evolving audience for AI journalism in the creator economy.

Topics discussed

Sponsorship and show introduction Welcome and guest introduction Kevin's role as CEO and background Transition to YouTube and Hard Fork Challenges in media and video journalism YouTube's role in high-quality journalism Machine Gods as a video-first podcast The AI story and platform shift Stability of Instagram for new stories Replicating visual content on new platforms Surprises in the creator economy Running a business vs. journalism Evaluating AI products as they evolve Audience demand for AI realism AI realism and understanding the technology The book's meticulous reporting Preserving the history of deep learning Challenges in documenting private AI efforts Kevin's role as a historian Google's shift towards AGI The long history of AGI concepts The transformer architecture and OpenAI The light bulb moment for LLMs Sponsorship from Quo Sponsorship from Framer Framer's enterprise solutions Framer's e-commerce features Next token prediction's success The mystery of next token prediction Cosmic mysteries in AI development Feeling the AGI in AI labs Google's admission of building AGI The singularity and Google I/O AI companies' shared vision Religious fervor and business decisions Belief as a key ingredient in AI success OpenAI's compute constraints The blip and OpenAI's pivot OpenAI's bet on AGI Focusing on coding and agentic workloads B2B SaaS and true passions Anthropic's predictions and reality The cost and time of building AGI Idealists running an enterprise company Zealots and missionaries in AI The teleological end of capitalism Effective altruism and accelerationism Bill Gates and the AI race Inevitabilism in AI development Technology as a force for good The cost of empowering corporations AI backlash and company responses Chatbots before cancer cures Google's task-based approach to AI Google's reaction to OpenAI DeepMind's reinforcement learning Other paths to AGI LLMs as the most promising path Sponsorship from CrowdStrike Sponsorship from OutShift OutShift's agentic AI insights AI systems and the illusion of life Public perception of AI capabilities The subjective step change in AI GPT-4 and the Turing test Talking to computers for a long time The experience of GPT-4 class models Saturated human-like communication The culture of AI parties A small subculture making big decisions The ethos of the AI scene The weight of ambition and consequences Futurism in San Francisco Accepting the weight of ambition The future of AI companies The singleton theory of AGI Inference and open-weight models Distillation and second-tier companies Idealistic fervor in AI development Competitive strategies in AI Being AGI-pilled Brittleness in AI systems Novel problem solving and intelligence Learning from young children AI systems acting like computers Computational functionalism The nature of human cognition Religiosity in Silicon Valley Building gods and seeking God AI psychosis and the plateau The future of UGC ads Show credits and LinkedIn sponsorship Bose sponsorship
Listen ad-free on Castria

Episode summaries and topics are generated from the audio by Castria — a free podcast app that skips ads automatically. IP Geolocation by DB-IP