Who Will Adapt Best to AI Disruption?

As artificial intelligence begins to fundamentally reshape the American labor market, a new working paper from the National Bureau of Economic Research (NBER) argues that policymakers are asking the wrong question regarding job displacement. Rather than simply calculating which roles will be automated, the study suggests the critical metric is "adaptive capacity"—the financial, geographic, and skill-based resources that allow workers to survive professional dislocation.

Key Takeaways

• The "Adaptive Capacity" Metric: The NBER study introduces a new index measuring a worker's ability to transition jobs based on savings, age, geography, and skill transferability.
• The Vulnerable 6.1 Million: While many high-tech workers are exposed to AI, a specific cohort of 6.1 million workers faces the "double jeopardy" of high exposure and low adaptability.
• Gender Disparity: The study indicates that 86% of workers in the most vulnerable, low-adaptability category are women, primarily in administrative and clerical roles.
• Geographic Concentration: Vulnerability is highest in mid-sized institutional hubs, such as state capitals and college towns, rather than major metropolitan tech centers.

Redefining AI Risk: Exposure vs. Adaptability

Previous economic analyses have largely focused on "exposure"—identifying which professions involve tasks that Large Language Models (LLMs) can perform. However, the NBER researchers argue that exposure does not guarantee economic ruin. Instead, they propose that the outcome of disruption depends on the worker's ability to pivot.

The study constructs a composite measure of "adaptive capacity" based on four distinct economic factors:

• Liquid Financial Resources: Workers with substantial savings can endure longer job searches and wait for better employment matches. Low-wealth individuals are often forced to accept lower-quality employment immediately.
• Age: Younger workers historically find reemployment faster. A 2017 study cited by the researchers notes that workers aged 55–64 are significantly less likely to find new employment after displacement compared to their 35–44 counterparts.
• Geographic Density: Workers in densely populated cities have access to a thicker labor market with more "destination jobs" compared to those in rural or low-density areas.
• Skill Transferability: Generalist skills allow for occupational mobility, whereas highly specialized skills can become liabilities if a specific industry contracts.

The Tale of Two Workforces

By cross-referencing AI exposure with adaptive capacity, the researchers identified a stark divide in the American workforce.

The first group consists of approximately 26.5 million workers who possess high exposure to AI but also high adaptive capacity. This cohort includes software developers, financial managers, and lawyers. While their daily tasks may change significantly due to automation, they benefit from strong pay, financial buffers, and diverse professional networks.

"These well-positioned workers who observers often cite as being highly threatened by AI automation likely possess relatively strong means to adjust to AI-driven dislocation if it were to occur."

The second, more concerning group consists of 6.1 million workers who face high AI exposure combined with low adaptive capacity. These workers predominantly occupy administrative, secretarial, and clerical positions. They often lack significant savings, possess skills with limited transferability, and face narrower reemployment prospects.

The demographic implications of this finding are significant. The data reveals that 86% of this high-risk group are women. Furthermore, the risk is geographically concentrated in smaller institutional centers that rely on administrative support roles, such as Laramie, Wyoming; Stillwater, Oklahoma; Springfield, Illinois; and Carson City, Nevada. In these areas, 5% to 7% of the local workforce falls into the high-vulnerability category.

Limitations and Structural Concerns

While the NBER framework provides a granular look at worker vulnerability, analysts note a potential flaw in the methodology: it presumes the labor market will continue to operate on historical precedents.

The study’s reliance on historical data—such as the impact of trade shocks or plant closures—models AI disruption as a localized event where workers eventually transition into an otherwise stable economy. However, AI represents a potential systemic shock across all cognitive task categories simultaneously.

If AI fundamentally reduces the aggregate demand for human cognitive labor, the traditional "destination jobs" that displaced workers would move into may no longer exist. For example, if secretaries, customer service representatives, and insurance claims processors are all displaced simultaneously, they cannot absorb each other's labor supply. In this scenario, factors like skill transferability may become irrelevant if the skills themselves are devalued across the board.

Policy Implications: A Triage Approach

Despite these structural uncertainties, the NBER study offers a roadmap for immediate "triage policy." Regardless of whether the long-term impact of AI is a transformation of roles or a structural reduction in work, the sequencing of disruption is likely to follow the vulnerability gradient identified by the researchers.

The data suggests that policymakers should prioritize resources for the 6.1 million workers in the low-adaptability quadrant. Because these workers have the least capacity to self-insure and are geographically identifiable, they represent the most urgent sector for government intervention, retraining programs, and financial safety nets.

As the economy heads toward 2026, the focus must shift from theoretical discussions of automation to discrete policy decisions aimed at the specific demographics and geographies most at risk of immediate income disruption.