Geography of AI Disruption

In the AI Story

Autor's China shock research demonstrated that national-average measurements conceal enormous geographic heterogeneity in adjustment costs. Some communities absorbed the shock smoothly; others experienced decades of decline. The determining factors were local industry composition, the education and mobility of affected workers, the health of local institutions, and the responsiveness of public policy. These same factors will determine which communities absorb AI smoothly and which experience cascading decline.

The exposure geography of AI is in some ways inverted relative to prior shocks. Manufacturing automation disproportionately affected Rust Belt cities; the China shock hit the same regions and extended to furniture-making areas of North Carolina and textile regions of the Southeast. AI's initial exposure concentrates in knowledge-economy hubs — San Francisco, New York, Boston, London, Bangalore — precisely the places that previously benefited from automation and trade. This reversal has political implications: the workers most affected by AI may be those whose political coalitions were built around the assumption that they were automation's winners.

The international geography adds another layer. Segal's developer in Lagos and Trivandrum engineers illustrate how AI redistributes capability across national boundaries. The outsourcing arrangements that built economic pathways for emerging-economy knowledge workers over two decades are being disrupted by AI's ability to perform remotely-executable tasks directly. Whether emerging economies can capture the new value AI creates depends on institutional factors — infrastructure, education, regulatory frameworks — that operate on slower time scales than the technology's diffusion.

Origin

The spatial analysis of technology shocks has been a continuous thread in Autor's research, from his early work with David Dorn on the geographic distribution of routine-task employment through the China shock papers to his 2024 'New Frontiers' paper which includes a spatial dimension of new work creation.

Key Ideas

Exposure is geographic. AI's labor-market effects are distributed unevenly across space, concentrated in regions whose occupational composition is heavy in AI-exposed tasks — which is not the same geography as prior technology shocks.

The knowledge-economy hubs are exposed. Unlike manufacturing automation, AI's initial exposure concentrates in cities that previously benefited from technological change, inverting familiar political coalitions.

International patterns are complex. AI simultaneously empowers developing-country workers (by lowering thresholds to high-value tasks) and threatens them (by automating outsourced work); the net effect depends on institutional factors that vary by country.

Local adjustment capacity varies. Whether a region absorbs the AI shock smoothly depends on local institutions, worker mobility, education, and policy response — factors that the China shock research shows produce divergent outcomes from identical shocks.