How Will AI Affect Jobs?

Episode Summary

Executive Summary: The episode examines whether AI will trigger a job apocalypse or a slower, uneven labor transition. Goldman Sachs’ Joseph Briggs expects meaningful but temporary displacement with substantial job creation over time, while MIT’s Neil Thompson and Darren Acemoglu argue that adoption will lag capability and likely produce smaller near-term losses, but potentially greater longer-term inequality if AI investment favors replacement over complementarity.

Main Topics: Near-term AI labor displacement (Priority: 5/5): Joseph Briggs says AI is already affecting a few sectors, but the current labor-market drag is narrow rather than economy-wide. Long-run job reallocation vs. permanent unemployment (Priority: 5/5): Briggs argues AI will reallocate workers over a decade, not permanently eliminate work, because technology historically creates new jobs as it destroys old ones. Capabilities vs. adoption constraints (Priority: 5/5): Neil Thompson emphasizes that AI capability alone does not equal job loss; adoption depends on data access, cost-effectiveness, and reliable applications. Partial automation and task composition (Priority: 4/5): Thompson explains that jobs contain many tasks, so automating only some tasks can raise wages, lower wages, or increase employment depending on whether AI replaces expert or inexpert work. Medium-term labor market risk and vulnerable occupations (Priority: 5/5): Darren Acemoglu expects some net job losses within five years, concentrated in cognitive routine jobs like customer service and back office work. Inequality and distributional effects (Priority: 4/5): Acemoglu warns that AI could widen labor-income inequality if it displaces routine middle jobs more than highly paid or blue-collar work. Broader wildcards: robotics, social tasks, and agentic AI (Priority: 3/5): The discussion highlights robotics, middle-management automation, and social acceptance of AI as major uncertainties that could expand or limit labor impact.

Key Arguments: Current AI labor impact is visible but limited to certain sectors; Briggs estimates only a modest drag on monthly job growth today. Over a 10-year transition, Briggs expects about 9% of U.S. workers to be reallocated, but unemployment should rise less than one percentage point in any given year if displacement is gradual. Historical evidence suggests technology drives the majority of job growth over time, so Briggs sees no reason to expect permanent mass unemployment. Thompson argues AI capabilities must be paired with usable data, workflow integration, and economic incentives before jobs are actually automated. He expects adoption to be slower than capability growth, with larger firms and high-value tasks automated first and a long tail of slower-to-change work. Automation affects workers differently depending on whether it removes expert or inexpert tasks: removing inexpert tasks can raise wages and make jobs more valuable, while removing expert tasks can depress wages but increase the number of workers able to perform the job. Acemoglu expects small net job losses in the next five years, particularly in cognitively routine roles, because current AI is not yet robust enough to complement workers at scale. He believes bigger net losses could emerge over 10–15 years if investment remains focused on replacing labor rather than augmenting it. AI’s impact on inequality may be substantial because the likely jobs at risk are not the highest-paid roles, while displaced workers may be pushed into lower-paid occupations. Major upside/downside uncertainty remains around AI plus robotics, which could expand displacement into physical tasks if breakthroughs occur.

Data Points: Current AI drag on job growth: 10,000 to 15,000 - Briggs estimates the current month-over-month drag from AI in sectors like tech, consulting, and graphic design Projected worker reallocation: 9% of all U.S. workers - Briggs’ baseline forecast under a 15% productivity uplift from full AI adoption Worker count equivalent of 9% displacement: 15 million workers - Briggs translates the 9% reallocation estimate into absolute labor-market terms Annual unemployment impact: less than 1 percentage point - Briggs says even with 9% displacement over 10 years, annual unemployment increases would likely stay below this level Share of job growth driven by technology: 85% - Briggs cites this as the historical share of job growth over the last 80 years Annual jobs created in the U.S.: around 30 million - Briggs highlights labor-market churn as evidence of dynamism Annual jobs destroyed in the U.S.: around 29 million - Briggs uses this to show that job creation and destruction constantly coexist Potential job creation needed to reabsorb displaced workers: 5% acceleration in new job creation - Briggs says this would be enough to absorb expected AI displacement Near-term net job losses forecast: less than 2% to 4% - Acemoglu’s estimate for the labor-market impact over the next five years Potential vulnerable worker pool: 8 million to 9 million workers - Acemoglu’s estimate for cognitive routine jobs such as customer service and back office roles Historical time frame referenced for labor trends: since 1940 / last 80 years - Both Briggs and Acemoglu discuss long-run technology and labor-market patterns Time horizon for major uncertainty: 10 to 15 years - Acemoglu says longer-term outcomes depend heavily on investment choices and application development

Pivotal Quotes: "I don't subscribe to the view that we are going to see a world in which a lot of people just don't end up with a job." — Joseph Briggs: Briggs rejects the idea of permanent mass unemployment from AI "AI capabilities are only one in a series of steps that lead to a change in jobs." — Neil Thompson: Thompson explains why capability gains do not directly translate into job losses "I think that people are right to look at the AI capabilities evolving and to say this does present a potential challenge to labor." — Darren Acemoglu: Acemoglu acknowledges labor risk but frames it as limited and conditional

Implications: Listeners should expect AI to reshape jobs unevenly rather than eliminate them outright. The biggest risks are slower hiring, task displacement, and rising inequality, while the biggest opportunity is productivity growth and new job creation if AI is deployed to complement workers.