Live: Anthropic Co-Founder on AI and Housing – A Planet Money Book Tour Special

This episode of Planet Money brings together two live interviews from the show's book tour, tackling two of the most pressing economic questions of our time: what happens when AI systems become capable of doing work that currently employs millions of humans, and how to fix a housing market that locks out first-time buyers. The conversation moves from the speculative (a co-founder of Anthropic reading his own science fiction as a kind of warning) to the practical (a behavioral economist explaining why zoning reform matters and how surge pricing actually works), all while maintaining the show's signature blend of curiosity and skepticism. The stakes are personal: Jack Clark has two young children whose future he is actively shaping, and Daryl Fairweather is trying to convince communities to care about residents who haven't been born yet.

5:05Jack Clark's Science Fiction as Blueprint and Warning

The interview with Anthropic co-founder Jack Clark begins with an unusual artifact: a collection of science fiction short stories he published online years before starting the company, available for free at jackclark.net. Host Kenny Malone asks Clark to read a passage imagining a factory in 2025 where AI oversees robots on the production line, designs new products, simulates them in high-fidelity environments, airlifts prototypes by drone to human focus groups, and then sells the successful ones through an eBay-like auction system. The factory's "improvisation line" then begins exploring the "latent space" of new products, iterating without human intervention.

Malone presses Clark on the obvious question: where are the humans in this story, and how do they have money to buy things they didn't help make? Clark's answer is direct: "If AI goes as far as people think, you actually need to reconceptualize how capitalism in the largest possible sense works." In a world of closed-loop machine production, he argues, you need to tax the robots and AI companies significantly and find a way to reallocate money from the machine economy to the human economy. This is not a throwaway line—it becomes the central tension of the entire conversation.

Clark describes his fiction as "messages in a bottle" he is trying to throw out of "this semi-frightening AI lab, which I'm a principal character in." When Malone asks whether the stories should be read as a blueprint, instructions, warning, or advice, Clark answers simply: "Yes." He quotes a Jawbreaker lyric—"my fiction beats the hell out of my truck truth"—and explains that the newsletter he writes is about AI research papers, but the fiction is where he grapples with his moral and ethical responsibility. The audience's response to this admission is described as "eerie applause," capturing the ambivalence that runs through the entire episode.

10:10Concrete Predictions: 150-Hour Tasks and 100% AI-Generated Code

Clark reveals that he had just come from making a new set of predictions, having been pleased with how well his previous ones performed. His first prediction: by April 2027, AI systems should be able to do tasks that might take a person 150 hours—nearly a month of work. When Malone asks what kind of task that would be, Clark offers examples: designing a complicated piece of circuitry, doing a very involved research project involving reading and cross-referencing many things, building a piece of software. "The kinds of tasks we currently pay extremely talented people large amounts of money to do are exactly the type of tasks which AI systems are kind of creeping into," he says.

His second prediction is even more striking for what it implies about Anthropic itself. The majority of code at Anthropic is already written by AI systems like Claude. Clark predicts that by April, maybe 100% of the code will be. This led the team to imagine a future where their job becomes "sitting around analyzing and critiquing the code that Claude writes and verifying that it's correct." Clark admits he can't work out whether this sounds like "a really fun country club style job" or something more troubling. Malone responds that he waffles between horrified and optimistic, and Clark agrees, adding that he finds himself "staring into the face of change that we haven't experienced in perhaps a century."

To give historical perspective, Clark references C.S. Lewis's *The Screwtape Letters*, in which a demon celebrates that his human subject has acquired a telephone—because whenever the human would otherwise be alone and introspect, they will instead call friends. Clark argues that Lewis got the shape of the change right: phones massively altered how people relate to one another. But he rejects the implication that we are doomed because of technology. The point is that we are about to experience immense change in how we relate to each other, and we are only beginning to think through what that means.

13:31The Weapon and the Shield: Claude's Hacking Capabilities

The conversation takes a more immediate turn when Malone brings up Anthropic's new model, Mythos, which has proven exceptionally good at hacking. Clark confirms that this is not a special hacking model—it is "our regular Claude, and it just turns out to be good at hacking now." The company has not made the model public but has given it to approximately 40 to 50 companies, including Apple, Google, and Amazon, to help them defend against the very capabilities the model possesses. The only financial institution confirmed to have access is JP Morgan, and Treasury Secretary Scott Bessent has personally called Clark requesting access.

Malone presses Clark on the phone calls with government officials: "How have these phone calls been?" Clark responds that he has "a high tolerance for weird phone calls at this point in time." The nature of the conversations, he says, is that Anthropic has been warning for years that AI systems would keep getting better until they surpassed humans at most tasks. "That doesn't go down super well on the call," he admits. "So then you frame it as, well, what do we do about this?"

The most pointed question comes when Malone observes that Anthropic appears to have invented both the weapon and the shield, and asks what happens when the company starts charging for the defensive tool. "It does feel a little bit like, hey, you know, be a shame if something happens to that company of yours if you don't pay for our product," Malone says. Clark's response is firm: "We think that business strategies which tend more towards mafia than anything else are probably not long-term resilient." He argues that parts of AI need to become a true utility, where cyber defense capabilities are provided at cost or even at no margin. "I'm sorry if my financial officer is listening to me say that," he adds, "but it's the shape of where you end up in the future."

17:12Raising Children in the Future They Are Creating

Malone asks Clark directly about his two young children, who just returned from parental leave: "What is the future you're imagining for them? How do you think they will fit into the world that you are creating?" Clark's optimistic vision is that people will get to spend more time with one another and less time "being kind of atomized and driven away from their families and friends." But he immediately acknowledges the condition: "This requires you to actually solve the distribution problem." When Malone says, "Me personally?" Clark responds, "Yes, you. Congratulations, we're solving economics together."

The solution, Clark argues, requires completely rethinking education. "You're going to need to upend how education works in pretty fundamental ways, and you're going to have to teach people to be much more curious." He describes watching his child ask questions constantly—"why do camels have humps?"—and notes that this skill of asking questions gets "beaten out of you by rote learning and working in regular jobs." AI, he believes, can answer these questions and allow adults to maintain childlike curiosity. "I think that's going to lead the world to really exciting places where everyone has the inventiveness that you have as a child, but you can still use it and are encouraged to use it more as an adult."

The interview ends with Malone reading the final line of Clark's preface: "Good luck to us all." Clark nods, and the audience is left with that ambivalent benediction hanging in the air.

20:52The Behavioral Economist Inside Amazon

The second half of the episode shifts to Daryl Fairweather, chief economist at Redfin and former senior economist at Amazon. Hosts Kenny Malone and Alex Mayasi ask her about her unusual title: "behavioral economist." Fairweather explains that Amazon hired her specifically with that title because "they didn't want people at the part of the company that I was working at to get too nervous that they were hiring some economists in and it was going to cut roles or make things harder for people." The "behavioral" modifier was a signal that she would think about things "more holistically than maybe the typical economist."

Why did Amazon need a behavioral economist? Fairweather says it was during a period when the company was receiving bad press about how it treated employees—specifically, the New York Times exposé about working conditions in fulfillment centers. Her job was to work on employee engagement for frontline call center employees: figuring out how to keep them from quitting, how to keep them happy, and how to give them career opportunities. She had actual data to measure these outcomes.

Mayasi observes that economists are secretly influencing daily life in ways most people don't realize. He points out that dating app algorithms are based on Nobel Prize-winning economics research. Fairweather adds that rideshare apps are a famous example: dynamic pricing set by supply and demand replaced the old taxi meter system. She notes that this is now moving into grocery stores with electronic price tags that can change in real time. "I was tracking the price of my book," she says, "and my book is now cheaper than it was a couple of weeks ago."

24:29The Fairness Problem with Surge Pricing

The conversation turns to the backlash against surge pricing. Fairweather acknowledges that while dynamic pricing prevents shortages—if it starts raining and Uber prices don't go up, there's no incentive for more drivers to come out—people often prefer allocation by waitlist because it feels more fair, even if it's less efficient. "This is where my behavioral economist brain starts working," she says. "I feel like there has to be some nuance. When is surge pricing a bit grotesque?"

Mayasi introduces the concept of "freeness," which he describes as optimizing for whether people will be cool with being left out. "Surge pricing probably does not do that," Fairweather agrees. She notes that the standard economic argument for surge pricing—that high prices signal drivers to enter the market, which then brings prices down—works well in theory but doesn't always apply. "I try to think of that when I grumble about Uber prices," she says, "but I mostly just grumble anyway."

26:37Housing, Zoning, and the Musical Chairs Metaphor

Fairweather identifies the movement to eliminate single-family zoning as one of the biggest developments in housing over the last five years. Single-family zoning—the rule that only one house can be built on one plot of land in a neighborhood—is behind the scarcity of housing in cities like Seattle and across the country. The YIMBY movement ("Yes In My Backyard") has been winning the ideological fight, she says, because it is based on basic economic principles of supply and demand. "It's finally starting to click in politicians' minds and voters' minds that this is how you get more affordable housing."

But she acknowledges a common objection: people see new expensive townhouses being built and wonder how that helps affordability. Her response is the musical chairs metaphor. People who already own houses are sitting in chairs. When the spring housing market comes, some people get up and move to new chairs. If you're not adding more chairs, first-time home buyers can't get in. "If you were to add some luxury, plush, nice chair, it's probably going to go to a very wealthy person," she explains, "but then that wealthy person is not going to be sitting in the folding chair anymore. And that chair opens up for a first-time home buyer."

The interview closes with a quote from the Planet Money book: "I wish the 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." Fairweather explains that housing decisions are usually made by current residents—homeowners who go to town halls and advocate for less housing because they don't want their neighborhoods to change. The people who don't get a vote are the ones who don't live there yet but would like to, or the children who will one day need homes of their own. "I wish that the focus was less on what's best for current residents and what's best for the residents who might live here 5, 10, 15 years from now."

Mayasi observes that this is not about incentives or policies but about empathy. Fairweather agrees: "Empathy in economics. It happens."

Conclusion

What stays with the listener is the uncomfortable recognition that the people building the future are also the ones warning us about it. Jack Clark writes science fiction as a message in a bottle from inside an AI lab, predicting a world where his own children will need a completely rethought economic system to thrive. Daryl Fairweather argues that solving the housing crisis requires not just zoning reform but a fundamental shift in empathy—caring about people who don't yet exist. Both interviews circle the same truth: the hardest economic problems are not technical but moral. The episode matters because it refuses to let the audience off the hook. Clark's "good luck to us all" is not a throwaway line; it is the honest conclusion of someone who knows that technology will force choices we are not ready to make.

Key takeaways

• Jack Clark predicts that by April 2027, AI systems will be able to perform tasks that take a human 150 hours, and that Anthropic's code may soon be 100% AI-generated, with humans shifting to verification roles.
• Clark argues that a closed-loop machine economy requires taxing AI companies and redistributing money from the machine economy to the human economy—a fundamental reconceptualization of capitalism.
• Anthropic's Claude model is now good enough at hacking that the company is giving it to 40-50 companies (including Apple, Google, Amazon, and JP Morgan) for free as a defensive tool, with Clark arguing that cyber defense should be provided at cost as a utility.
• Clark believes education must be completely rethought to preserve childlike curiosity into adulthood, since AI can answer the questions that rote learning and conventional jobs currently beat out of people.
• Daryl Fairweather worked at Amazon as a "behavioral economist" specifically to signal a softer approach than traditional economics, focusing on employee retention and happiness for call center workers.
• Surge pricing prevents shortages by signaling supply and demand, but Fairweather acknowledges it feels unfair and that people often prefer waitlists, even if they are less efficient.
• The YIMBY movement is winning the ideological fight against single-family zoning, but Fairweather's "musical chairs" metaphor explains why building expensive housing still helps affordability by freeing up cheaper units for first-time buyers.
• Fairweather argues that the housing crisis is fundamentally a failure of empathy: current residents make decisions at town halls, but the people who don't get a vote are future residents who haven't arrived yet.