AI Exposure and Adaptability: Why Female-Dominated Administrative Roles Face Highest Risk

A new analysis from the Brookings Institution, summarizing research for the National Bureau of Economic Research, introduces a crucial dimension to the AI employment debate: not just how exposed a job is to AI disruption, but how well workers can adapt if displacement occurs. The findings reveal a stark divide where some professions face a double jeopardy of high exposure and low adaptability, creating a significant risk for a specific segment of the workforce.

The research moves beyond simple exposure metrics, which have previously focused on how susceptible different occupations are to automation by large language models (LLMs) and other AI systems. Instead, it evaluates "adaptive capacity" based on factors including age, financial security, union membership, geographic location, and the health of local labor markets. This approach provides a more nuanced picture of which workers are truly vulnerable to AI-driven job loss.

The Adaptability Divide

The analysis shows that of the 37.1 million U.S. workers in the top quartile of occupational AI exposure, 26.5 million—roughly 71%—also possess above-median adaptive capacity. These workers are generally well-positioned to weather potential disruption. Professions like lawyers, software developers, and financial managers fall into this category. Despite high exposure to AI tools, these roles typically offer strong pay, financial buffers, diverse skill sets, and deep professional networks that facilitate transitions.

Conversely, some jobs have low AI exposure and high adaptability, such as dentistry, firefighting, medicine, and flight attendant roles. Butchers were noted as having both low exposure and low adaptability.

The most concerning finding is the group of 6.1 million workers who face both high exposure to LLMs and low adaptive capacity to manage a job transition. These workers are concentrated in administrative and clerical jobs characterized by modest savings, limited skill transferability, and narrower reemployment prospects. The consequences are severe: longer job searches, lower chances of finding new employment, and more significant relative earnings losses compared to other displaced workers.

A Gendered Impact

The demographic profile of this vulnerable group is striking. The research indicates that 86 percent of these high-exposure, low-adaptability workers are women. This concentration reflects the persistent gender segregation in the U.S. labor market, where administrative and clerical roles have historically been and remain predominantly female-dominated.

The skills required in these positions—such as data entry, scheduling, and basic office management—are among the most susceptible to automation by current LLM capabilities. However, the workers in these roles often lack the financial resources or specialized training programs that would enable a smooth transition into other fields. The research suggests they are more likely to face prolonged unemployment or be forced into lower-wage positions outside their career trajectory.

Geographic Concentration and Labor Market Realities

The vulnerability is not evenly distributed across the country. The concentration of exposed and vulnerable workers is greatest in smaller metro areas and college towns, particularly in the Mountain West and Midwest regions. These areas have an elevated presence of administrative and clerical workers, often tied to local government, educational institutions, and small business services.

In these regions, the local labor markets may offer fewer alternative opportunities, especially for workers whose skills are not easily transferable to growing sectors. The combination of geographic isolation and occupational concentration creates a compounding risk, where a single AI-driven disruption could have outsized effects on local economies and communities.

What This Means for Compliance and Workforce Planning

For organizations and policymakers, this analysis underscores the need for targeted interventions. While broad upskilling initiatives are valuable, they must be designed with the specific constraints of vulnerable worker populations in mind. This includes:

1. Financial Support Programs: Addressing the modest savings and financial insecurity that limit workers' ability to retrain.
2. Skill Transferability Mapping: Identifying which adjacent roles or industries could absorb displaced workers and creating tailored pathways.
3. Geographic Considerations: Developing regional workforce strategies that account for local economic conditions and job availability.
4. Timing and Support: Recognizing that transitions for this group will likely be longer and require more intensive support services, including career counseling and job placement assistance.

The Brookings analysis provides a critical framework for understanding the differential impacts of AI on the workforce. It highlights that the conversation must move beyond which jobs will be automated to how we can support the workers in those roles—particularly those in historically marginalized and vulnerable positions—to navigate the coming changes. The data suggests that without proactive and targeted support, the transition to an AI-augmented economy risks exacerbating existing inequalities in the labor market.

For further reading on the methodology and detailed findings, the full Brookings Institution analysis is available on their official website. The underlying National Bureau of Economic Research study can be accessed through the NBER research portal.