New research from the Brookings Institution reveals a critical gap in how policymakers assess artificial intelligence’s impact on the workforce. While 37.1 million American workers face high AI exposure, 6.1 million lack the resources to adapt if job loss occurs.
Sam Manning and Tomás Aguirre report for the National Bureau of Economic Research. The researchers developed an “adaptive capacity” measure that examines workers’ ability to navigate job transitions based on savings, age, local job markets, and skill transferability.
The findings challenge common assumptions about AI’s workforce impact. Many highly exposed workers, including software developers, financial managers, and lawyers, possess strong financial buffers and transferable skills. About 26.5 million workers in high-exposure roles have above-median adaptive capacity.
The 6.1 million vulnerable workers tell a different story. They concentrate in clerical and administrative positions where modest savings meet limited skill transferability. Women comprise 86% of this group. Office clerks, secretaries, receptionists, and medical administrative assistants represent the largest affected occupations.
“By bridging the two literatures, we move beyond identifying which jobs face potential AI exposure to understanding which workers might face the greatest or least adjustment costs if disruption leads to displacement,” Manning and Aguirre write.
The research combines six datasets to create the adaptive capacity index. Financial security emerges as a key factor. Workers with greater liquid savings experience less financial distress after job loss and secure better job matches. Age matters significantly. Workers aged 55 to 64 show 16 percentage points lower reemployment rates than those aged 35 to 44 after job loss.
Geographic location shapes outcomes. Workers in densely populated areas face lower transition costs than those in rural regions. Skill transferability provides occupational mobility that specialized skills cannot match.
Geographically, vulnerable workers cluster in smaller metropolitan areas, particularly university towns and midsized markets in the Mountain West and Midwest. College towns like Laramie, Wyoming, and Huntsville, Texas, show elevated concentrations. State capitals including Springfield, Illinois, and Carson City, Nevada, face similar patterns.
Tech hubs like San Jose and Seattle show the highest concentrations of exposed but adaptive workers. The share of high-exposure, low-capacity workers ranges from 2.4% to 6.9% across metro areas, averaging 3.9% nationally.
The researchers acknowledge limitations. The index operates at occupation level, but individual circumstances vary widely within occupations. A 30-year-old computer network architect in San Francisco differs significantly from a 56-year-old counterpart in a small market.
The analysis draws primarily from localized displacement events rather than large-scale employment shifts. If AI affects multiple related occupations simultaneously, structural job availability may matter more than individual characteristics.
The research offers practical applications for workforce development. Understanding resilience patterns could inform public funding allocation for adjustment programs. Policymakers could use adaptive capacity measures to target data collection and streamline eligibility for transition assistance.
Manning and Aguirre emphasize that exposure measures alone fail to capture which workers would experience the most severe welfare costs from AI-driven displacement. Their approach provides a framework for identifying those most in need of support.
