Generative AI, the technology powering innovations like ChatGPT, presents a double-edged sword for the American workforce. While it promises to enhance productivity and create new opportunities, it also threatens to devalue skills and jeopardize job security, according to a Brookings Institution report.
The impact of this transformative technology, however, is not uniform across the United States. Brookings' analysis reveals that different regions face varying levels of AI exposure, with some areas poised to experience more pronounced effects—both positive and negative—on their local economies and labor markets.
San Jose and San Francisco in California; Durham, North Carolina; New York City; and Washington, D.C. are examples of metro areas with high office-oriented exposure to generative AI. Las Vegas, Nevada; Toledo, Ohio; and Fort Wayne, Indiana are less exposed. For a sense of range, 43% of San Jose workers could see effects on at least half of their work tasks, while for Las Vegas, the amount is 31%.
However, most of the data that the Brookings report mentioned was by county. Santa Clara County, California; King County, Washington; and New York County had worker exposures of at least 40%. Small and rural heartland counties varied downward from the 30% to the 20% range.
Continuing on the county basis, states (including D.C.) typically had some wide dispersal between counties exhibiting the most and least exposure. Here are the five regions with the highest average exposure: District of Columbia (about 43%); Massachusetts (average, 37.5%; high is Suffolk County at 40%; low is Nantucket County, 28%); Maryland (average, 37%; high is St. Mary’s County, 41%; low is Worcester County, 30%); Colorado (average, 36.5%; high is Broomfield County, 42.5%; low is Morgan County, 27%); and New York (average, 36%; high is New York County, 47%; low is Lewis County, 28%).
The five stated with the lowest averages are Indiana (average, 32%; high is Martin County, 40%; low is Carroll County, 23%); Wyoming (average is 31.5%; high is Albany County, 35%; low is Crook County, 28%); North Dakota (average 31%; high is Sioux County, 31.5%); Nevada (average 30.7%; high is Carson City County, 35.5%; low is Eureka County, 20.5%); and Mississippi (average 30.5%; high is Clairborne County, 36%; low is Scott County, 22%).
The approach that Brookings used started with data from OpenAI, the ChatGPT vendor, that forecast the exposure for an education or occupational pay level. Again, exposure doesn’t immediately mean a negative or positive impact. Different people might experience it differently.
What makes generative AI different from technological disruptions of the past is that the people most affected tended to be less-educated working lower-wage jobs. That has turned upside down. These new artificial intelligence systems excel at the cognitive work typically performed by programmers, writers, financial analysts, engineers, lawyers, doctors, designers, and others. Those dealing with customer service, skilled trades, service industries, and other work needing emotional or physical presence are much less susceptible to displacement — at least for now.
Brookings matched local shares of different occupations combined with the OpenAI data to calculate overall exposure. In some cases, the exposure might be negative with loss of employment. In others, it could mean greater productivity and capability.
As for CRE, the threats in high-impact areas are likely less need for real estate if businesses manage with fewer people for whatever reason. Also, more people out of work means economic pressures on renters and less business, so greater financial pressure, for those who depend on consumers as customers.