Yes, AI will reshape American cities, but not equally

It is already conventional wisdom that artificial intelligence is rapidly affecting our lives and upending the way we work. But it’s also clear that the impact of AI will vary from place to place, community to community.
With the potential to create new job opportunities, and eliminate others, AI appears poised to have a major economic impact around the world, particularly regarding inequality and societal gaps.
A new comprehensive Nokia Bell Labs study indicates that, if left unchecked, AI’s impact risks deepening this divide in America, and it would have not only labor market implications but regional, geographical ones as well.
After analyzing more than 24,700 AI-related patents filed with the United States Patent and Trademark Office between 2015 and 2022, and linking them to nearly 18,000 occupational tasks, a team of our UK-based researchers developed a fine-grained map of where AI is likely to hit hardest.
The study found that it isn’t big metropolitan areas that will be most affected by AI. In fact, many of the most at-risk locations were mid-sized cities whose economies hinged on a single sector.
These findings come on the heels of another study Nokia Bell Labs completed on how AI innovations will impact American occupations. The research concluded that it was white-collar, highly technical jobs that would likely face more disruption than low-skill, blue-collar jobs, revealing that pile driver operators and butchers, for instance, were far less vulnerable to AI disruption than cardiovascular technicians and sound engineers.
Now, the researchers have been able to translate all that individual data into a U.S. map of the projected geographical AI disruption for various metropolitan areas.
“Rather than the range of job sophistication, we found that a far better indicator of susceptibility to AI changes was related to lack of economic diversification,” said Daniele Quercia, Department Head of Social Dynamics at Nokia Bell Labs Cambridge (UK).
Darker colors represent higher AI impact within metropolitan statistical areas. Grey areas indicate regions outside the boundaries of metropolitan statistical areas, which generally have lower population densities and are not included in the analysis due to limited data availability. Source: Springer EPJ Data Science.
It’s all about economic diversity
The new study that Quercia and his Cambridge-based colleagues published last month in Springer EPJ Data Science examined how AI was likely to affect various U.S. towns and cities.
The key indicator of AI impact was clearly the economic make up of each location.
Big cities with highly innovative hubs, like New York, Seattle and San Francisco, were also expected to face disruption, indicating that “creative class” jobs such as engineers, scientists and designers were not immune to automation and, in some cases, were on the front lines of impact.
But in these large metropolitan areas, workers have far more options to find other employment and the cities themselves can better adapt to specific disruptions because of their economic diversity.
Where it gets trickier is in small and medium-sized areas whose fortunes depend on how reliant they are on specific sectors and industries.
Dalton, Georgia, for example, is tightly bound to carpet manufacturing. Columbus, Indiana, depends on the automotive industry. Rochester, Minnesota, is renowned for its healthcare sector. All face an elevated risk of AI disruption because of their heavy dependence on routine-intensive industries that are exposed to automation. They have few alternative sectors to soften the blow.
On the other hand, some of the least-impacted metropolitan areas in the analysis, such as Daphne-Fairhope-Foley in Alabama, Coeur d’Alene in Idaho and Grants Pass in Oregon, shared a different trait: economic diversity. These cities spread employment across a broader range of industries, making them better equipped to absorb shocks and adjust as AI transforms specific sectors.
“The communities more likely to be impacted are those that are heavily specialized in impacted sectors, with little to no efforts at diversification,” said Quercia. “These dynamics suggest that AI could amplify existing divides, hitting hardest in areas where economic opportunities are already concentrated and leaving many behind in the race for innovation-driven growth.”
Adjusting to AI
However, there is still hope. The first step is to follow the advice often given by financial planners: diversify your portfolio.
Quercia said that smaller metropolitan areas should begin such efforts while also investing more heavily in reskilling workers and repackaging existing jobs. Local governments also need to adopt policies that more aggressively encourage such reskilling, he added.
“It’s not that these jobs are going to disappear. They are going to be different, and people need the new skills to do them,” he said. “And if this fails, legislation is required to prevent a further spike in inequality.”
From a government level, there are plenty of things that can be done to fix the current trajectory. Policymakers can consider specific strategies to support areas most at risk, while companies can prioritize strategic decisions about workforce development at their most exposed locations.
On an individual level, Quercia would recommend that residents of endangered communities do more research about the vulnerability of their occupations.
To that end, his team has developed and published an Artificial Intelligence Impact (AII) score that assesses how closely an occupation’s tasks align with recent AI innovations. They have also created an Atlas of AI Risks, an interactive tool that helps the public explore over 350 real-world AI applications and classifies them by risk levels under the EU AI Act.
Quercia insists it is still possible to change the current trajectory, but that, without intervention, AI will exacerbate existing trends of globalization that benefit larger urban centers while harming smaller communities.
“The impact of AI is going to be the same as the impact of other disruptive technologies. And we all know that disruptive technologies have deep economic ramifications,” he said. “If you rely on one industry, no matter how well it is currently doing, you may be disrupted in the future by a new technology. It’s common sense not to put all your eggs in one basket, but this is the first time we are demonstrating that in relation to AI.”