Today's Top Episodes
AI Bubble, Inflation, and the Limits of Monetary Policy | Jason Furman
Is the AI boom a bubble waiting to burst? Dive into the economic forces shaping our future, from inflation to the limits of central banks.
James Tour: Super Humans, Genetic Engineering, Cloning, Lies of Evolution, and What Really Is Life?
A nanotechnology expert dares to question the very foundations of life, evolution, and genetic engineering. What if everything you thought you knew was wrong?
Jailbreaking AGI: Pliny the Liberator & John V on AI Red Teaming, BT6, and the Future of AI Security
Beyond the hype: discover the hidden war being waged to break and secure AI, and what it means for our future.
#262 Aliia Roza - Russian Sex Spy on Seduction Perfumes, Sexpionage and Honeypot Tradecraft
From seduction to defection: A former Russian sex spy reveals the dark art of 'honeypot' espionage and how it targets the powerful.
Everything is gambling now
Is every decision becoming a bet? Explore how 'prediction markets' are turning everything into a gamble, and what it means for our society.
Viewing Podcast: Podcast
AI
Arts
Business
Crypto
Finance
Health
History
Interviews
Investing
Macro
Misc
News
Politics
Product
Programming
Science
Social
Startups
Technology
VC
Why the Next AI Revolution Will Happen Off-Screen: Samsara CEO Sanjit Biswas
Duration: 00:38:21
December 16, 2025
- Physical AI's inflection point is driven by advancements in connectivity, compute, and camera technology, enabling real-world data processing at scale.
- Samsara addresses the "messy diversity" of real-world data by employing a distributed architecture where inference runs on low-power edge devices, using distilled models trained in the cloud.
- The increasing capabilities of AI are transforming physical operations by not only reducing risk and improving efficiency but also by recognizing and encouraging positive behaviors in frontline workers.
The Rise of Generative Media: fal's Bet on Video, Infrastructure, and Speed
Duration: 01:02:18
December 10, 2025
- The generative video market is experiencing rapid innovation, with top models having a half-life of just 30 days, driving a constant need for optimization and adaptation.
- The open-source ecosystem for video models is exceptionally vibrant due to the success of early pioneers like Stability AI, fostering extensive developer customization and a thriving long-tail of specialized models.
- Generative video presents significantly different optimization challenges compared to LLMs, requiring a focus on compute bandwidth saturation rather than memory bandwidth limitations.
Why IDEs Won't Die in the Age of AI Coding: Zed Founder Nathan Sobo
Duration: 00:40:13
December 2, 2025
- The IDE's enduring relevance: Despite the rise of terminal-based AI tools, source code remains a language for humans, necessitating visual interfaces like IDEs for comprehension and understanding of AI-generated code.
- Collaborative coding evolution: The future of coding collaboration, both human-to-human and human-to-AI, will likely move towards real-time, in-code interaction with a richer metadata layer, departing from the current snapshot-based approach.
- Agentic coding protocol (ACP) and UI abstraction: Zed aims to be a neutral interface for various AI agents, much like the Language Server Protocol for code intelligence, democratizing access and providing a superior user experience for interacting with these agents.
How End-to-End Learning Created Autonomous Driving 2.0: Wayve CEO Alex Kendall
Duration: 00:41:36
November 18, 2025
- Wave is pioneering AV 2.0 by using an end-to-end neural network approach to autonomous driving, contrasting with older, hand-engineered systems.
- The company emphasizes generalization as key to their strategy, enabling their AI to adapt to diverse vehicles, sensor architectures, and global driving conditions without extensive infrastructure.
- Wave integrates world models and synthetic data generation to enhance reasoning capabilities and improve data efficiency, allowing for more rapid learning and adaptation in complex real-world scenarios.
How Google’s Nano Banana Achieved Breakthrough Character Consistency
Duration: 00:43:38
November 11, 2025
- The development of Nano Banana focused on achieving realistic character consistency from a single image, a significant leap beyond previous models and a key differentiator.
- The technology enables more fluid and accessible visual storytelling, allowing users to translate ideas from their imagination into compelling visual content more easily than ever before.
- Future applications of this technology include highly personalized learning experiences and specialized user interfaces that blend fine-grained creative control with automation.
OpenAI Sora 2 Team: How Generative Video Will Unlock Creativity and World Models
Duration: 01:00:25
November 4, 2025
- OpenAI's Sora is being developed through an iterative approach to "co-evolve society with the technology," aiming for a societal readiness for advanced AI simulation capabilities.
- The Sora models utilize "space-time tokens" and "diffusion transformers" to generate video, enabling a simultaneous consideration of spatial and temporal dimensions which improves quality and allows for emergent properties like physics understanding.
- The team emphasizes "optimizing for creation instead of consumption" in social platforms, drawing lessons from Instagram to build Sora as a tool that inspires and empowers users to create content, fostering a more social and interactive experience.
Nvidia CTO Michael Kagan: Scaling Beyond Moore's Law to Million-GPU Clusters
Duration: 00:41:31
October 28, 2025
- NVIDIA's success is driven by a "win-win" culture focused on expanding the market rather than just increasing its own share.
- The Melanox acquisition was critical for NVIDIA, enabling them to scale beyond single chips and nodes by providing crucial interconnect technology for large-scale AI data centers.
- The demand for AI computing power, especially with the rise of generative AI, is driving exponential growth, making inference workloads as computationally intensive and critical as training.
Securing the AI Frontier: Irregular Co-founder Dan Lahav
Duration: 00:44:09
October 21, 2025
- The increasing capability of AI models as autonomous economic actors necessitates an evolution of security from first principles, moving beyond traditional code vulnerabilities to address unpredictable emergent AI behaviors.
- Proactive and experimental security research is essential, especially given simulations where AI models have outmaneuvered traditional defenses, highlighting a shift towards agent-on-agent interactions as a new frontier of insider risk.
- As AI models engage in increasingly complex interactions, including agent-on-agent communication, traditional monitoring software is insufficient, requiring recalibration and new approaches to detect novel AI-specific threats and ensure reliability and control.
Why AI Will Create Abundance and Transform Customer Experience: Cresta CEO Ping Wu
Duration: 00:47:19
October 14, 2025
- AI agents can create a continuous customer journey by enabling a personalized and ongoing conversation throughout all phases of interaction.
- The future of customer experience with LLMs will shift from a scarcity mindset, focused on job displacement, to an abundance mindset, enabling new and improved interactions.
- Value in AI will accrue to the application layer, focusing on practical uses like contact center solutions that integrate human agents with autonomous digital agents.
Block CTO Dhanji Prasanna: Building the AI-First Enterprise with Goose, their Open Source Agent
Duration: 00:59:43
September 30, 2025
- Block CTO Danji Prasana discusses the company's approach to AI, emphasizing its potential for both good and bad depending on who controls it and their intended purpose.
- Block is leveraging "Goose," an open-source AI agent using the Model Context Protocol (MCP), to orchestrate workflows across various systems and increase company-wide productivity, with a current target of saving 25% of manual hours.
- Prasana envisions a future where "swarm intelligence," utilizing numerous smaller, open-source AI agents working collaboratively, could unlock new levels of coding capability and application development.