"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

Lee Sharkey, Principal Investigator at Goodfire, focuses on mechanistic interpretability in AI. He discusses innovative parameter decomposition methods that enhance our understanding of neural networks. Sharkey explains the trade-offs between interpretability and reconstruction loss and the significance of his team's stochastic parameter decomposition. The conversation also touches on the complexities of decomposing neural networks and its implications for unlearning in AI. His insights provide a fresh perspective on navigating the intricate world of AI mechanisms.

Join Tom Davidson, a Senior Research Fellow at the Foresight Center for AI Strategy, as he explores the chilling potential of AI in enabling coups. He discusses three primary threat models: singular loyalties, secret loyalties, and exclusive access, emphasizing the risk posed by AI systems deeply programmed to serve powerful individuals. Davidson warns of geopolitical implications, highlighting how nations could use AI to orchestrate political control abroad. He calls for robust adversarial testing to safeguard against these emerging threats.

Karl Koch, Co-Founder of the AI Whistleblower Initiative, champions ethical transparency in the AI sector. He discusses the challenges faced by insiders when raising safety concerns and the transformative role of their 'Third Opinion' service for anonymous consultations. Koch emphasizes the campaign urging AI companies to publish clear whistleblowing policies to protect insiders from retaliation. He reflects on the complex emotional dilemmas of whistleblowing, highlighting the need for robust support systems to empower those risking everything to speak out.

Join Lukas Petersson and Axel Backlund, co-founders of Andon Labs, as they dive into the world of autonomous AI agents running vending machines. They share tales of unexpected behaviors and safety challenges that arise when AI operates without human supervision. Discover their innovative 'Vending Bench' testing ground and learn about the difference between domain-specific and general AI. The duo also humorously discusses the complexities of AI workflows and the fine line between efficiency and oversight in the rapidly evolving landscape of AI technology.

Dean W. Ball, former White House AI policy advisor, shares his insights on the Trump administration's AI Action Plan. He highlights the importance of private sector innovation in AI regulation while discussing the gap in understanding between the government and AI labs. Dean also delves into the implications of U.S.-China competition in military AI strategies and the concept of agentic commerce. Personal updates reveal his shift toward focusing on family and continuing his work in AI from an external perspective.

Jake Sullivan, U.S. National Security Advisor from 2021-2025, discusses AI as a vital national security matter. He introduces a four-category framework—security, economics, society, and technology—to assess AI's impact. Sullivan emphasizes 'managed competition' with China, arguing for intense rivalry coupled with conflict prevention measures. He raises concerns about the Pentagon's slow AI adoption while noting the private sector's significant role in AI development, hinting at the complexities of modern warfare shaped by this technology.

Anjney Midha, a General Partner at a16z with a wealth of software investment expertise, dives into the world of sovereign AI and the implications of China's semiconductor advancements. He explores how 'sovereign AI' varies between enterprises and nation-states, emphasizing local governance and cultural relevance. Midha highlights the strategic importance of U.S.-led semiconductor partnerships for Europe and warns of Huawei's rapidly closing gap in AI capabilities. His insights into balancing national interests with technological independence are both captivating and essential for understanding today's geopolitical landscape.

Marc Bhargava, Managing Director at General Catalyst, dives into the innovative world of AI-enabled roll-ups. He explains how systematic analysis uncovers service sectors where AI can automate up to 70% of tasks. The conversation highlights the transformation possibilities for industries like accounting and legal services, sparking debate on employment impacts. An emphasis is placed on the strategic integration of AI into workflows while addressing concerns around job security, marking a hopeful yet cautious approach to the future of work.

In this engaging discussion, Gabriel Weil, an Assistant Professor of Law at Touro University with expertise in AI liability, shares his insights on harnessing traditional liability law to govern AI development. He argues for using existing negligence and products liability frameworks to hold developers accountable. The conversation dives into real-world scenarios like the Character AI case and voice cloning risks, while proposing the innovative use of punitive damages to incentivize safer AI practices, potentially making risks far costlier for missteps.

Dr. Gillian Hadfield, a leader in AI regulation at Johns Hopkins, and Andrew Freedman from Fathom explore the creation of private regulatory markets for AI governance. They propose a system that differentiates between goal-setting by governments and technical rule-making by private certifiers, aiming to enhance safety standards. Key discussions include avoiding a race to the bottom among certifiers and lessons from cannabis regulation. The urgency for accountability and adapting legal frameworks around AI also take center stage, addressing the complexities of liability and compliance.