Measuring US workers' capacity to adapt to AI-driven job displacement

New research pairs estimates of occupational exposure to artificial intelligence with measures of workers' adaptive capacity to assess who may be best positioned to handle potential job displacement. The analysis, detailed in a National Bureau of Economic Research paper by Sam Manning and Tomás Aguirre, incorporates factors such as liquid savings, age, local labor market density and skill transferability alongside exposure estimates from prior studies including Eloundou and others. Of the 37.1 million U.S. workers in the top quartile of occupational AI exposure, 26.5 million also have above-median adaptive capacity. The analysis identifies 6.1 million workers, or 4.2 percent of the sample workforce, who face both high exposure and low adaptive capacity. These workers are concentrated in clerical and administrative roles, with about 86 percent women according to Lightcast data. The affected workers are geographically concentrated in smaller metropolitan areas, including university towns and midsized markets in the Mountain West and Midwest. The study combines six primary datasets to build a composite measure of adaptive capacity by occupation, including the Survey of Income and Program Participation for net liquid wealth, the American Community Survey for age distributions, Occupational Employment and Wage Statistics for wage and employment figures, Lightcast for county and metro employment shares, Bureau of Labor Statistics employment projections and O*NET for skill requirements. The approach aims to help policymakers better target resources by distinguishing workers with strong means to adjust from those with more limited options.