Scaling Laws

A tense standoff between Anthropic and the Department of Defense with a hard deadline and high stakes. Debates over whether companies or governments should set limits on AI use. Possible government levers like contract cancellation, the Defense Production Act, or labeling firms as supply-chain risks. Concerns about mass surveillance, lethal autonomous weapons, and broader industry fallout.

Kevin Frazier, AI innovation and law fellow who studies AI governance, and Cullen O'Keefe, research director focused on frontier AI policy, discuss automating regulatory compliance. They explore how AI can compile reports, run model evaluations, and detect incidents. They debate limits like Goodhart's Law, the fairness of compute thresholds, and the idea of conditional "automatability triggers" to delay enforcement until cheap tools exist.

Amanda Askell, researcher leading Anthropic’s personality alignment team and primary author of Claude’s Constitution. She explains the Constitution as a training guide for values and behavior. Methods covered include supervised learning and RL signals. Discussion touches on enforcement, living-document updates, courageability vs. human judgment, cultural adaptation, instruction hierarchies, and ethics of personhood.

Gillian Hadfield, a law and policy scholar focused on adaptive regulation, and Andrew Freedman, strategist building AI governance frameworks at Fathom, explore novel models for overseeing fast-moving AI. They discuss regulatory markets, independent verification organizations, pitfalls of command-and-control rules, liability’s role, and how iterative, market-driven oversight could scale governance in practice.

David Rand, Cornell professor studying misinformation and AI influence. He discusses how accuracy nudges curb sharing and how conversational AI can durably reduce conspiracy beliefs. He explains experiments where chatbots shifted voter preferences, why factual-sounding claims persuade, and the risks of AI bot swarms, training data biases, and the need for transparency.

Nathan Labenz, host of the Cognitive Revolution, leads a wide-ranging talk about AI and the law. They debate whether models already outperform median lawyers and how AI could fill legal deserts. Conversation covers regulatory roadblocks, new rights like a right to compute, and how legal careers might split between routine work and high-level legal architects.

Most folks agree that AI is going to drastically change our economy, the nature of work, and the labor market. What's unclear is when those changes will take place and how best Americans can navigate the transition.  Brent Orrell, senior fellow at the American Enterprise Institute, joins Kevin Frazier, a Senior Fellow at the Abundance Institute, the Director of the AI Innovation and Law Fellow at the University of Texas School of Law, and a Senior Editor at Lawfare, to help tackle these and other weighty questions.Orrell has been studying the future of work since before it was cool. His two cents are very much worth a nickel in this important conversation. Send us your feedback (scalinglaws@lawfaremedia.org) and leave us a review! Hosted on Acast. See acast.com/privacy for more information.

Jakub Kraus engages Alan Z. Rozenshtein, an Associate Professor of Law, and Kevin Frazier, AI Innovation and Law Fellow, to dissect Anthropic's new constitution for Claude. They explore the philosophical underpinnings of AI governance and the document's guiding principles, aimed at ethical and safe AI development. The duo debates the concept of AI as a potential moral being and the implications of market dynamics on model choices. Frazier advocates for clear consumer metrics, while Rozenshtein emphasizes the need for virtue ethics in AI alignment.

Shlomo Klapper, founder of Learned Hand, brings a unique perspective on integrating AI into the judicial system. He discusses how the platform aids overworked judges by automating repetitive tasks, allowing them to focus on critical judgments. Shlomo highlights AI's potential to improve access to justice, especially for unrepresented litigants, while addressing concerns about biases and fact-checking. He also touches on the readiness of state trial courts for early adoption and the necessity of building trust in AI tools within the legal arena.

In this engaging discussion, Francis Shen, a Professor of Law at the University of Minnesota and director of the Shen Neurolaw Lab, explores the innovative intersection of AI and criminal justice. He highlights how AI can enhance criminal investigations and improve clearance rates. Shen advocates for precision sentencing to tailor justice to individual cases while addressing ethical concerns around AI bias. He urges for transparent implementation of AI in local governance, building public trust and institutional capacity for a more effective justice system.