The New Geography of AI: Urban Gains, Rural Displacement

A new study on regional labor markets argues that the biggest divide in the age of AI may not be between skilled and unskilled workers alone, but between cities and the countryside. Automation tends to reduce employment and wages, especially in rural regions, while AI is associated with higher wages and is concentrated in urban labor markets. The finding implies that technology is not simply widening inequality in a generic sense, but reorganizing it, rewarding some places while putting others under sharper strain.

The paper, The Urban–Rural Divide in the Age of Artificial Intelligence: Assessing the Effects of Technology and Automation on Regional Labor Markets, separates two technologies that are often lumped together in policy debates. Automation mainly substitutes for routine tasks, such as those found in manufacturing, processing, and other repetitive work. AI, by contrast, is more likely to complement cognitive and analytical tasks, especially in information-intensive occupations.

The paper warns that a country that treats "technology" as a single force will miss the fact that one shock is pulling jobs out of rural areas while another is raising wages mostly in cities.

Automation's rural footprint

According to the study, automation has a negative impact on employment and wages. In the preferred estimates, higher automation exposure is associated with a lower employment-to-population ratio and weaker earnings. The rural employment effect is the most pronounced: a one-standard-deviation increase in automation exposure is linked to roughly a 3-percentage-point decline in employment. For regions where labor markets are already thinner and adjustment options are limited, that is a substantial hit.

Rural labor markets typically depend more heavily on routine and manual work, whether in agriculture-linked processing, light manufacturing, logistics, or low-skill services. Those are precisely the kinds of jobs automation can replace most easily. When the shock arrives, workers in those places do not just lose tasks; they may lose the nearest alternative jobs as well. The results suggest that automation does not merely reduce labor demand in the abstract. It hits hardest where people have the fewest ways to move into something else.

The paper also finds that the automation shock is less damaging in cities, which means they are better able to absorb the disruption. In urban labor markets, workers can more easily shift into new roles, firms can reorganize tasks more quickly, and the broader economy can support job reallocation.

AI's very different geography

The second major finding is that AI looks nothing like automation. Instead of depressing wages, AI exposure is associated with higher wages. The study estimates that a one-standard-deviation increase in AI exposure corresponds to roughly a 2.6% to 2.8% wage gain. The effect is especially visible in urban regions, where AI exposure is already higher to begin with. In the descriptive data, urban regions show noticeably greater AI exposure than rural regions, reflecting the concentration of cognitive and information-heavy jobs in cities.

The pattern challenges the idea that all digital technologies function as job destroyers. AI, at least in the study's framework and results, seems to complement workers in occupations that rely on judgment, analysis, coordination, and communication. These jobs are more common in cities, where firms cluster and where digital infrastructure is stronger. So while automation strips away work from routine-heavy regions, AI may be reinforcing earnings in places already positioned to benefit.

A positive average wage effect does not tell us who within the labor market is benefiting. Higher average wages can coexist with widening inequality inside occupations, between firms, or across education levels. AI is not a mirror image of automation. It is a different kind of shock, with different winners, different losers, and a different spatial pattern.

The divide is changing shape

The research shows how the urban–rural divide is being reshaped rather than simply widened. For years, debates about technological change have often used broad language about inequality, polarization, and displacement. This study suggests that those labels are too blunt. The more useful question is not whether technology raises inequality in general, but which technology affects which places through which mechanisms.

Automation is concentrated in routine work, which is more common outside cities. AI is concentrated in cognitive work, which is more common in cities. Rural regions thus face the displacement side of technological change more strongly, while cities are better placed to capture the complementary side. This is why the divide becomes more pronounced in some dimensions and less in others. Cities may see stronger wage gains, while rural areas suffer more job losses. The net result is not a simple national average but a patchwork of local outcomes.

National labor policy often assumes that a single set of training programs or digital policies can serve everyone. The study shows why that is unlikely to work. A region facing routine-based job losses needs a very different response from one experiencing wage gains in AI-intensive occupations. In practical terms, one-size-fits-all policy risks under-serving both groups. Rural workers need support to transition out of vulnerable jobs, while urban workers may need protections against the unequal distribution of AI's benefits.

The paper also resonates with a larger global trend: technology is increasingly interacting with local capacity. Places with stronger institutions, better infrastructure, and deeper labor markets tend to adjust more quickly. Places with weaker systems absorb more of the pain. The pattern is not unique to AI, but AI may intensify it because its benefits depend so heavily on digital access, education, and the ability to use new tools productively. In that sense, the study is not only about labor markets. It is about the geography of resilience.

What policymakers should do

The study argues for place-sensitive responses rather than uniform national programs, which means targeted reskilling in rural regions exposed to automation, stronger digital infrastructure outside major cities, and labor-market support that helps displaced workers move into new tasks or new locations. It also means investing in foundational skills that make AI-complementary work more accessible beyond big urban centers.

Industrial policy, education policy, and regional development policy need to be coordinated. If automation is reducing demand for routine labor in rural areas, then training should focus on transferable skills, digital literacy, and pathways into non-routine work. If AI is generating wage gains in cities, then policy should ensure that those gains do not remain locked inside a small set of firms or occupations. Broadband, data systems, and access to digital tools are no longer secondary concerns; they are part of labor-market inclusion.

In emerging economies, rural regions may be far less equipped to handle displacement while also being less able to capture AI's upside. It could deepen spatial inequality even when national growth looks healthy. The lesson for donors and multilaterals is to think beyond aggregate GDP and employment figures. They should ask where jobs are being created, where they are being lost, and which regions have the capacity to adapt.

Businesses should also pay attention. Firms that expand automation into rural supply chains or regional manufacturing networks may face workforce disruption that is not visible in national statistics. Companies deploying AI in urban centers may see productivity gains that do not automatically translate into broad-based wage growth. Corporate strategy should include workforce transition planning, not just technology adoption. Civil society organizations, meanwhile, can use these findings to advocate for stronger retraining systems and more equitable digital access.

Rural communities in the Global South often have weaker digital infrastructure, lower education levels, and fewer formal jobs to fall back on. If technology policy ignores those constraints, it risks deepening existing divides. If it responds strategically, however, AI and automation could be handled in a way that broadens opportunity rather than concentrating it.