Sections

Commentary

Interactive: How AI will affect American cities

Featured Image - AI in Workforce TechStream

As artificial intelligence is increasingly adopted at American workplaces, its impact won’t be equally spread across the United States. Instead, its geography will be diffuse, with greater impacts on large metropolitan areas.

Using a novel technique developed by Stanford University Ph.D. candidate Michael Webb that uses AI-related patents and Labor Department job descriptions, we sought to understand in a recent report what occupational areas the technology is poised to affect and where in the country those effects will be felt most dramatically.

As a young, complex technology, artificial intelligence seems ready to disrupt the large metro areas with a high concentration of workers in high-tech and white-collar industries. That includes places such as San Jose, Calif., Seattle, and Salt Lake City, as well as the Boston-Washington, D.C. corridor.

In the interactive graphic below, mouse over metropolitan areas to see how different U.S. cities are affected.

Larger, denser urban communities—as well as Heartland metros—are more exposed to AI

Average standardized AI exposure by metro or NECTA, 2017

Source: Brookings analysis of Webb (2019)

But it’s not just America’s techies who will feel the effects. There is plenty of overlap between AI patents and jobs in manufacturing as well, putting large swaths of America’s heartland into the high-exposure category. That includes mostly the eastern heartland states of Wisconsin, Michigan, Indiana, Kentucky, before sweeping south into Alabama and Georgia. On a metro area level, you can see agriculture and logistics hub Bakersfield, Calif. with a high AI exposure score, as well as manufacturing centers Greenville, S.C., Detroit, and Louisville, Ky.

The map of AI exposure stands in contrast to our January 2019 report that focused more broadly on automation’s workforce impact. The takeaway there was a big impact on smaller, rural communities with lower educational attainment levels and high concentrations of workers in accommodation and food services, manufacturing, transportation, agriculture, retail, and mining. We found in that report on automation’s impacts that less than one-fifth of the workforce in these areas are in tech-intensive jobs, exposing them more to automation—a dynamic that is flipped when talking about AI.

An interesting case study for the AI versus automation outlook is Nevada. A hub of the accommodation and food services industry, the automation report placed Nevada in its top tier of most-exposed states. But artificial intelligence is a much different technology, and our analysis didn’t foresee it as a major disruptor of those occupations. Thus, looking at Nevada’s AI-exposure score, the state becomes one of the least at-risk of disruption.

Read more: The full report is here.

Mark Muro is a senior fellow and policy director in the Metropolitan Policy Program at The Brookings Institution.
Jacob Whiton is a research analyst in the Metropolitan Policy Program at The Brookings Institution.
Robert Maxim is a research associate in the Metropolitan Policy Program at The Brookings Institution.

Michael Gaynor is a staff writer and editor at the Metropolitan Policy Program at Brookings.

Authors