AI, Employment, and Education (with Tyler Cowen)

Episode Summary

Executive Summary: Russ Roberts and Tyler Cowen discuss how AI may reshape jobs, governance, and higher education. Cowen rejects mass-unemployment fears, arguing AI will create new work, more wealth, and more leisure, while warning that politics and regulation may worsen. The conversation then turns to AI-enabled education, cheating, tutoring, certification, and how colleges should adapt.

Main Topics: AI and the labor market (Priority: 5/5): Cowen argues AI will displace some tasks but create new industries, especially in energy, biotech, law, and consulting, with no mass unemployment scenario. Governance as the biggest risk (Priority: 5/5): He repeatedly says his main concern is not economics per se but that AI could worsen governance, regulation, and political decision-making, which would then harm growth. AI and higher education (Priority: 5/5): The discussion centers on how colleges should teach AI use directly, possibly devoting a large share of curriculum time to AI-assisted learning and experimentation. Cheating, writing, and assessment (Priority: 4/5): Cowen argues AI cheating is overrated and can be managed through occasional in-room writing tests and better-designed assignments that combine AI use with human judgment. Human meaning, leisure, and work (Priority: 4/5): Both speakers explore whether AI will reduce work, increase leisure, and alter the meaning of jobs, with Cowen suggesting work will remain important and more varied. Certification and educational reform (Priority: 4/5): Cowen proposes AI could eventually outperform traditional credentials by testing, tutoring, and certifying mastery more effectively than grades and degrees.

Key Arguments: AI will not cause mass unemployment; it will shift labor demand into new sectors such as energy, biomedical trials, law, and AI integration services. Governance is the biggest uncertainty: if politics and regulation deteriorate, AI’s economic benefits could be blunted or reversed. AI will likely increase wealth, lower prices, and expand access to services, especially for people at the bottom of the income distribution. A large share of college instruction should teach students how to use AI well because that is becoming a universal workplace skill. Cheating can be reduced by sampling students’ work under controlled conditions and comparing in-room performance with take-home performance. Writing should be taught both with and without AI: AI-assisted writing can raise standards, while unaided writing still helps students learn to think. The future of education may be more personalized, conversational, and AI-driven, with small-group human interaction preserved where it matters. AI may eventually improve certification by testing knowledge, temperament, and strengths more effectively than conventional grades.

Data Points: Top episode share: 33% - Russ Roberts reports that Tyler Cowen’s favorite 2025 EconTalk episode, 'What is Capitalism with Mike Munger,' was chosen by 33% of listeners. Tyler Cowen appearances: 20th appearance - Roberts notes this is Cowen’s 20th time on EconTalk. Survey top 10: 10 episodes - Roberts lists the top 10 favorite EconTalk episodes of 2025. Curriculum share proposed for AI: One-third - Cowen suggests roughly a third of college course time should be devoted to learning how to use AI well. Controlled writing sample: 2% to 3% - Cowen proposes locking students in a room for a small sample of their output to detect AI-assisted cheating. Driverless vehicle timeline: Within 10 years - Cowen predicts Tesla may become a truly cheaper alternative to Waymo within about a decade. Class experiment frequency: One class - Cowen proposes each student take at least one AI-based class, or even one every two or four years, to test the model. AI tutoring duration: 15 weeks - Roberts and Cowen discuss a semester-long AI tutoring/class model as a transitional format. Small-group size: 25 or fewer - Roberts describes Shalem College’s core curriculum classes as groups of 25 or fewer students. Consulting pay example: $1.4 million to $300,000 - Cowen imagines displaced upper-middle-class consultants moving to Houston and earning far less than before.

Pivotal Quotes: "I think the biggest worry is governance." — Tyler Cowen: Cowen identifies political and regulatory deterioration as his main concern about AI’s broader effects. "The cheating problem with AI is much overrated." — Tyler Cowen: He argues that better assessment design, not panic, is the right response to AI-assisted cheating. "People come to Shalem not to study something, but to become something." — Russ Roberts: Roberts explains his college’s educational philosophy while discussing AI’s limits and the value of human formation.

Implications: AI is likely to raise productivity and reshape education, but institutions must adapt fast. The biggest danger is not joblessness but bad governance, weak assessment, and political backlash from displaced professionals.