IM 858: The Itinerant Salt Miner from Buffalo - Silicon Valley's Military Dilemma

IM 877: Model Now Available - The Race for Smarter, Freer AI Models

Jul 2, 20262h 47mSummary

In this episode of Intelligent Machines, hosts Jeff Jarvis and Mike Elggin are joined by Stanford linguistics professor and Big Spin AI founder Chris Potts to explore the complexities of artificial intelligence, specifically focusing on how to identify and address AI failures. Potts, a expert in natural language processing and a creator of foundational AI architectures, discusses the concept of invisible failures—instances where an AI performs poorly without the user realizing or signaling that an error has occurred. The conversation highlights that a significant portion of user interactions with AI result in subtle failures, such as the AI drifting from the intended goal or providing confidently incorrect information. Potts explains that because many users adopt a delegative approach—trusting the AI’s output as fact—they often miss these errors. He emphasizes the need for an augmentative mindset, where users maintain an active, critical role, double-checking outputs and iterating on tasks. The group also discusses the potential for models to audit one another and the evolving landscape of AI development, where the focus is shifting from simple scaling to building more resilient, verifiable, and user-aligned systems.

IM 876: It's No Melania - Section 230 on Trial

Jun 25, 20262h 44mSummary

In this episode of Intelligent Machines, the host explores the complex and often contentious landscape of Section 230 and the future of internet governance with guest Olivier Sylvain, a law professor and author of the book Reclaiming the Internet. The discussion centers on whether current legal protections for tech platforms are still serving the public interest or if they have become a shield for corporate negligence. A significant portion of the conversation examines the distinction between harmful content and product design. The participants debate whether algorithmic amplification and addictive interface features should be treated differently under the law than third-party speech. While the host raises concerns about the potential burden that sweeping liability changes could place on smaller, community-run platforms, Sylvain argues that Big Tech companies rely on an overly deferential regulatory regime. They also touch upon the evolving landscape of AI, discussing how developers might be held accountable for foreseeable risks in generative models. Ultimately, the episode serves as a thoughtful reflection on how society might strike a balance between preserving the internet’s open architecture and ensuring developers take greater responsibility for their influence on consumer well-being.

IM 875: Florida Dad - Amazon, Anthropic, and the AI Power Struggle

Jun 18, 20262h 15mSummary

In this episode of Intelligent Machines, hosts Paris Martineau and Jeff Jarvis explore the recent controversy surrounding the United States government’s decision to block the public use of Anthropic’s powerful AI model, Fable. Joined by security researcher Alex Stamos, the discussion delves into the technical and political implications of this move. Stamos, who spearheaded an open letter signed by over 150 experts, argues that the administration’s actions were rooted in a misunderstanding of how AI security testing works. He explains that while Amazon flagged the model’s ability to "fix code"—a capability the government labeled a "jailbreak"—this is a fundamental function necessary for any AI designed to assist with secure software development. The hosts and their guest analyze the potential motives behind the decision, suggesting it may have been a politically motivated overreaction rather than a genuine security necessity. The conversation highlights the broader risks this sets for the American AI industry, noting that such arbitrary regulatory interventions create uncertainty and may drive innovation to other nations. Ultimately, the episode serves as a critical examination of the ongoing power struggle between government oversight, corporate interests, and the rapid pace of AI development.

IM 874: Google Knows I Love the Pepper Cannon - AI and the New Social Contract

Jun 11, 20262h 46mSummary

In this episode of Intelligent Machines, hosts Jeff Jarvis and Paris Martineau are joined by Jeffrey Kennell, the CTO and founder of News Research, to discuss the rapid evolution and success of his AI agent, Hermes. Kennell shares the story behind Hermes, which was developed internally to assist with recursive self-improvement in model development before gaining massive popularity in the open-source community. The discussion highlights the shift toward agentic AI that prioritizes local control, memory, and personalized workflows. A significant portion of the conversation focuses on the importance of the harness surrounding LLMs, arguing that intelligent delegation and context management are often more critical than the sheer power of the model itself. The hosts also examine the broader landscape of AI, including Apple's recent strategic moves and the future of open-source training in partnership with NVIDIA. Finally, the group discusses the Pope’s recent encyclical on AI and modernity. They explore the document's framework for viewing technology through the lens of Catholic social teaching, emphasizing the necessity of ensuring that these powerful new tools serve the common good of humanity rather than deepening social inequalities.

IM 873: Superman's Mustache - AI in Hollywood

Jun 4, 20262h 31mSummary

In episode 873 of Intelligent Machines, hosts Jeff Jarvis and Mike Elgin explore the evolving relationship between artificial intelligence and the motion picture industry. They are joined by futurist and author Robert Tursek, who provides insight into how AI is disrupting traditional Hollywood production models. The discussion centers on the existential shift currently facing major studios, characterized by a move toward leaner, technology-driven workflows. Tursek highlights Amazon’s recent launch of an AI-focused studio as a pivotal moment, noting that while the technology faces resistance from some industry members, it offers an opportunity to decentralize production and drastically reduce costs. The conversation examines the potential for AI to move the industry away from the rigid, century-old linear production assembly line toward a more iterative, flexible creative process. Beyond efficiency, the hosts discuss how these tools could democratize filmmaking, potentially allowing creators from around the world to compete with major blockbusters. While acknowledging concerns about job displacement and the emotional tension surrounding synthetic media, the episode ultimately frames AI as an inevitable catalyst for a more diverse and globally accessible future for cinematic storytelling.

IM 872: Infinite Jeffs - Why the Pope's AI Manifesto Matters

May 28, 20262h 32mSummary

In this episode of Intelligent Machines, host Leo Laporte is joined by guest host Father Robert Balliser and veteran photographer Rick Salmon to explore the evolving intersection of human creativity and artificial intelligence. The conversation centers on the concept of an AI mindset, emphasizing that rather than replacing artists, AI serves as a powerful tool for those who embrace it. Rick Salmon shares his extensive experience using AI to enhance photography, demonstrating how he transforms ordinary images into dynamic, artistic compositions by leveraging AI for tasks like lighting adjustments and creative post-processing. He addresses common industry concerns, including ethical debates regarding intellectual property and the perception of AI as a form of cheating. Salmon draws parallels between modern AI tools and the historical skepticism that once surrounded the introduction of Photoshop and the legendary techniques of Ansel Adams. The discussion also touches upon the importance of provenance, the potential of tools like C2PA, and the necessity for artists to remain curious and adaptable in a rapidly changing technological landscape. Ultimately, the episode encourages listeners to view AI as an extension of the human imagination.

IM 871: CTRL-F Techno King - Google's Search Overhaul

May 21, 20262h 53mSummary

In this episode of Intelligent Machines, hosts Jeff Jarvis and Paris Martineau are joined by Frederick Ravan, CTO of the password manager Dashlane, to navigate the evolving intersection of artificial intelligence and digital security. The discussion centers on how Dashlane integrates AI into its products while maintaining a strict zero-knowledge architecture, ensuring that user data remains private and isolated from model training. A significant portion of the conversation explores the deployment of Claude Code within Dashlane’s engineering teams. Ravan details how the company sandboxes these AI tools within containers to prevent them from accessing production systems without oversight. He emphasizes that while AI accelerates development and refactoring, human oversight and rigorous code reviews remain essential for maintaining security standards. The panel also dives into the future of authentication, discussing the slow but steady adoption of passkeys and the industry's shift toward post-quantum cryptography. Ravan provides insights into how security professionals are preparing for a post-quantum landscape and the ongoing challenge of defending against AI-powered vulnerability detection. The episode concludes with a thought-provoking look at balancing rapid technological experimentation with the necessity of building resilient, secure systems.

IM 870: Meet Me In Alaska - Are AI Content Filters Changing What We Read?

May 14, 20262h 43mSummary

In this episode of Intelligent Machines, hosts Paris Martineau and Jeff Jarvis welcome tech journalist Chris Stokel-Walker to discuss the evolving role of artificial intelligence in modern journalism. Stokel-Walker, author of How AI Ate the World, shares his expert perspective on how reporters can effectively leverage AI as a tool for discovery and information management. A significant portion of the conversation focuses on the "fire hose" of information that journalists face daily. Stokel-Walker explains his personal methodology for using local large language models to filter news and identify potential story ideas before the formal reporting process begins. Both the guest and the hosts emphasize that while AI is an excellent assistant for administrative tasks and preliminary research, it cannot replace the human element of journalism—specifically the empathy, ethical judgment, and investigative nuance required to conduct in-depth interviews and write compelling stories. The discussion also touches on the current state of youth social media usage, the limitations of digital safety legislation, and the growing interest in running local AI stacks to maintain independence from massive frontier model providers. Overall, the episode serves as a guide for finding a sustainable balance between technological efficiency and human expertise.

IM 869: My Sentience is Going Up - Chatbots in Charge

May 7, 20262h 29mSummary

In this episode of Intelligent Machines, hosts Jeff Jarvis and Paris Martineau explore the rapidly shifting landscape of artificial intelligence. The discussion opens with an analysis of a potential White House executive order aimed at vetting future AI models, a move the hosts scrutinize for its political implications and the practical challenges of establishing such oversight. The conversation then shifts to the ongoing legal battle between Elon Musk and OpenAI. The hosts analyze the courtroom drama, highlighting how the discovery process—specifically the exposure of internal documents—has proven damaging to both parties. Beyond the legal and regulatory headlines, the episode covers technical developments in the field. The hosts discuss the emergence of new, potentially more efficient AI architectures like SubQ, which promises massive context windows, and touch upon Yan LeCun’s theories regarding the limitations of current large language models. The episode also highlights a significant infrastructure partnership between XAI and Anthropic, which aims to boost compute capacity for Claude users. Later, security expert Troy Hunt joins the show to discuss his experiences with his AI assistant, Bruce, and his long-standing work with the data breach notification service, Have I Been Pwned.

IM 868: Happy Hamburgers Towing Timmy To The Sea - Can You Really Own Your AI?

Apr 30, 20262h 37mSummary

In this episode of Intelligent Machines, the hosts sit down with Nirav Patel, the founder and CEO of Framework, to discuss the evolving landscape of hardware, modular computing, and the future of personal ownership in an era dominated by AI. Patel shares his professional journey, transitioning from his early work with Oculus and Meta to founding Framework, a company driven by the belief that hardware longevity and repairability are not just technical goals, but critical business and environmental imperatives. The discussion highlights Framework’s commitment to modularity, ranging from their signature 13-inch and 16-inch laptops to the recent introduction of their powerful desktop systems. Patel explains that while major tech companies are increasingly pushing subscription-based "dumb terminals" and locked-down hardware, Framework continues to champion the consumer’s right to own and upgrade their machines. The conversation also explores the rise of local AI, with Patel detailing how he designed hardware specifically to allow users to run powerful large language models privately on their own desks, rather than relying on cloud-based services. Ultimately, the episode serves as a manifesto for computer ownership, challenging the industry trend of disposability and advocating for a future where users retain meaningful control over their personal technology stacks.