Live: Anthropic co-founder on AI and jobs

Episode Summary

Executive Summary: This Planet Money live episode from the book tour pairs two conversations: one with Anthropic co-founder Jack Clark on AI’s economic, ethical, and security implications, and one with economist Daryl Fairweather on how economists shape markets and why housing reform matters. Together, they argue that AI will force a rethink of work, money, and education, while housing affordability depends on adding supply and considering future residents.

Main Topics: AI as an economic and moral disruptor (Priority: 5/5): Jack Clark describes AI as a force that could automate major tasks, reshape capitalism, and require society to rethink how money flows from machine-heavy production to people. AI security, hacking, and utility-like defenses (Priority: 5/5): The conversation focuses on Claude’s emerging hacking capabilities, Anthropic’s limited rollout to companies, and the idea that AI-powered cyber defense may need to be provided like a public utility. Preparing for an AI-shaped future (Priority: 4/5): Clark says education, curiosity, and distribution policy must change so people can adapt to a world where AI handles more complex work and humans spend more time with each other. How economists influence everyday products (Priority: 4/5): Daryl Fairweather explains that economists work inside firms like Amazon and affect pricing, employee engagement, ride-sharing, and other consumer systems, not just academia. Dynamic pricing and fairness (Priority: 4/5): The episode examines surge pricing, market-clearing prices, and why efficiency can conflict with fairness and public acceptance in contexts like Uber and groceries. Housing shortages, zoning, and YIMBY politics (Priority: 5/5): Fairweather argues that single-family zoning constrains supply and that increasing density is essential for affordability, even if new housing initially appears expensive.

Key Arguments: If AI systems can run production and innovation loops end-to-end, capitalism will need to be reconceptualized so people still have income to buy goods. AI companies and robots may need to be taxed heavily or otherwise redistribute value to the human economy. Anthropic’s Claude is already good enough at hacking to justify controlled, free deployment to many companies as a defensive measure. Cyber defense and other socially important AI capabilities should behave like utilities: broadly available, low-margin, and not restricted to paying customers only. AI will likely automate tasks that currently take highly paid humans many hours, such as coding, research, and circuit design. The future may require a new education model that preserves curiosity and teaches people to ask better questions. Economists increasingly shape real-world outcomes inside firms, from employee retention at Amazon to matching and pricing on platforms like ride-share apps and dating apps. Dynamic pricing can improve efficiency by signaling scarcity and recruiting supply, but it can also feel unfair or grotesque to consumers. Housing reform works through supply: adding new units lowers scarcity over time, even if the first wave of new housing is expensive. Local land-use decisions often privilege current homeowners over future residents and people who want to move in, which worsens long-term affordability.

Data Points: Cities on book tour: 6+ - The hosts mention stops in New York, D.C., Boston, San Francisco, Seattle, and Los Angeles. AI task horizon: By April 2027 - Jack Clark predicts AI systems may handle tasks that take a person 150 hours. Human work benchmark: 150 hours - Used as the size of complex tasks AI may soon complete. Code written by AI at Anthropic: majority - Clark says most code at Anthropic is already written by AI systems like Claude. Predicted code generation: 100% by April - Clark says the team speculated all code could be AI-generated by April. Companies testing defensive model: 40 to 50 companies - Anthropic has given the new model to dozens of companies to test security. Hack-testing clients: Apple, Google, Amazon - Examples of companies reportedly receiving access to the model. Human workers in Clark scenario: thousands of distributors - The fictional factory story includes an eBay-like site used by thousands of distributors. Children: 2 - Clark notes he has two children and is thinking about their future. Public events remaining: 3+ - The episode notes future shows in Pittsburgh, Chicago, and Toronto.

Pivotal Quotes: "You need to tax the robots and AI companies significantly, and you need to somehow find a way to reallocate money from this machine economy to the human economy." — Jack Clark: Clark explains how a future with machine-to-machine production would still need human purchasing power. "We think that business strategies which tend more towards mafia than anything else are probably not long-term resilient." — Jack Clark: He rejects using security tools as leverage to coerce companies into paying for defensive AI products. "I wish that communities had the mentality that it is their responsibility to grow because we know there's going to be future generations that are going to need housing." — Daryl Fairweather: Fairweather argues that housing policy should account for future residents, not just current homeowners.

Implications: The episode suggests AI will force new rules for work, income, security, and education, while housing affordability depends on expanding supply and prioritizing future residents. For listeners, it frames both AI and housing as policy problems, not just tech or market trends.