Episode Summary
Executive Summary: Derek Thompson and economist David Deming examine why recent college graduates are facing weaker job prospects and whether generative AI is partly responsible. Deming argues the main drivers are business-cycle uncertainty and a slowdown concentrated among new grads, not a collapse in college value. He sees AI as a disruptive force that will reshape entry-level white-collar work, education, and career pathways, but not eliminate net employment.
Main Topics: Why recent college-grad unemployment is rising (Priority: 5/5): Deming says the most plausible explanation is cyclical uncertainty: firms hesitate to hire and train young workers when the economy feels unstable. He distinguishes this from a broader collapse in demand for college graduates. Whether college still pays off (Priority: 5/5): The conversation pushes back against claims that college has become a bad investment. Deming argues the degree still yields a strong lifetime wage premium, even if returns have plateaued from unusually high levels in the 1990s and 2000s. AI as a substitute for entry-level white-collar labor (Priority: 5/5): The hosts focus on how tools like ChatGPT can perform many tasks common in junior office roles—research, summarizing, drafting memos—raising the possibility that AI is eroding the bottom rung of the corporate ladder. AI adoption and labor-market disruption (Priority: 4/5): Deming notes generative AI has been adopted unusually quickly, at rates comparable to major technologies like personal computers. He expects broad disruption, but more in task reshuffling and productivity competition than mass unemployment. Education under AI pressure (Priority: 5/5): They discuss how schools and colleges may need to move away from written assignments that AI can easily complete and toward presentations, oral defenses, deeper mastery, and creative judgment. Who benefits from AI and how firms will use it (Priority: 4/5): Deming suggests AI helps people with uneven skill sets by filling gaps, especially in writing and translation. He expects the biggest gains for those who already have domain expertise and know how to leverage AI as a teammate. Why AI’s impact will be slow but real (Priority: 4/5): A Denmark study and Deming’s broader view suggest companies are still learning how to integrate AI, so the effects on hiring and output are gradual. The likely outcome is a reshuffling of winners and losers rather than immediate job destruction.
Key Arguments: The weak market for recent grads is more likely due to business uncertainty than a structural collapse in college demand. College remains a very good investment for individuals, even if its average premium is less exceptional than in prior decades. Generative AI is highly competitive with many tasks assigned to entry-level white-collar workers, especially reading, writing, and synthesis. AI adoption has been fast enough to plausibly matter for labor markets, though evidence is still early and incomplete. AI is more likely to change the composition of work and the skills firms demand than to eliminate employment broadly. Education should shift from easily outsourced written work toward oral, experiential, and deeper mastery-based learning. People with strong expertise plus AI augmentation will likely outperform those who rely on AI as a shortcut. The near-term productivity effects of AI may be limited because humans still must verify outputs and decide what to build.
Data Points: Unemployment rate for recent college grads: nearly 6% - Used to illustrate the worsening labor market for young graduates. Harvard MBAs still job searching three months after graduation: 23% - Wall Street Journal quote showing elite MBA placement weakening. Share of 25-29 year-olds with a bachelor’s degree before the Great Recession: 30% - Deming cites this to explain changes in the college wage premium. Share of 25-29 year-olds with a bachelor’s degree today: 40% - Shows a roughly one-third increase in degree attainment. Generative AI users at least once a month: 35% to 40% - From Deming’s survey on AI adoption. Workers who used generative AI at work in the last week: about 25% - Indicates substantial workplace penetration. Computer usage by workers in 1984: 25% - Used as a comparison point for AI adoption. Generative AI usage by workers in 2024: 27% - Deming compares this to 1984 PC adoption. Effect of Danish chatbot rollout on employment/earnings: precise zero impact - Study cited to show no immediate employment effect at adopting firms. ChatGPT age at time of discussion: about 2 to 2.5 years old - Used to explain why labor-market effects may not yet be visible. Top-tier MBA placement deterioration: more than a dozen programs affected - Harvard, Wharton, Stanford, NYU Stern, and others had weaker outcomes than recent memory.
Pivotal Quotes: "“This is going to be more of a slow roll.”" — David Deming: On how AI will affect jobs and firm organization over time rather than instantly displacing large numbers of workers. "“The college wage premium actually increases. It almost doubles over the course of somebody’s career.”" — David Deming: On why college still appears to be a strong long-run investment despite some flattening of returns. "“The best use of AI is to cheat.”" — David Deming: On why current education structures incentivize misuse of AI and need redesign.
Implications: Listeners should expect AI to reshape junior white-collar work, education, and hiring norms more than erase jobs outright. The winners will likely be people who combine expertise, judgment, and AI fluency; schools and employers must adapt quickly.