6.1 Million US Workers Face High AI Exposure With Low Ability to Adapt

Not Everyone Can Pivot When AI Comes for Their Job

Most conversations about AI and employment treat workers as interchangeable. A report says your job is at risk, so you retrain. You learn to code, or you move to a growing city, or you pick up a new credential. Problem solved.

But what if you cannot? What if your savings are thin, your skills are specialized in ways that do not transfer easily, and the nearest growing job market is three hours away?

A new study from the Brookings Institution, published in January 2026 by Sam Manning, Tomas Aguirre, Mark Muro, and Shriya Methkupally, tries to answer exactly this question. Instead of just measuring which jobs AI can automate, the researchers measured which workers have the capacity to adapt when it does — and found that millions do not.

The Numbers Behind the Vulnerability

[Fact] The study identifies 37.1 million US workers in the top quartile of AI exposure — meaning their jobs overlap heavily with what current AI systems can do. That is roughly one in four American workers.

Here is where it gets interesting. Of those 37.1 million, about 26.5 million — or 70% — possess above-median adaptive capacity. [Fact] They have some combination of transferable skills, financial cushion, younger age, or proximity to diverse job markets. These workers are exposed to AI, yes, but they have realistic options to transition.

The remaining 6.1 million workers do not. [Fact] These workers sit at the dangerous intersection of high AI exposure and low adaptive capacity — representing about 4.2% of the total US workforce. They hold jobs that AI can increasingly perform, and they lack the resources to pivot.

To put that in perspective, 6.1 million is roughly the entire population of Missouri. It is more than the combined workforce of Wyoming, Vermont, Alaska, and the Dakotas.

Who Are These Workers?

The demographic profile is stark. [Fact] A full 86% of the 6.1 million vulnerable workers are women. This is not a small gender skew — it is an overwhelming one.

The occupations driving this concentration tell the story. The largest groups include office clerks at about 2.5 million workers, followed by administrative assistants and secretaries at roughly 1.7 million. Receptionists account for about 965,000, and medical secretaries add another 831,000. [Fact] These are historically female-dominated roles where the daily work — scheduling, data entry, document processing, customer routing — maps directly onto what large language models and AI assistants can now do.

The Brookings researchers measure adaptive capacity across four dimensions. [Fact] These are liquid financial resources (can you survive a period without income?), age (younger workers have more time to retrain), geographic labor market density (are there other jobs nearby?), and skill transferability (do your abilities apply to growing occupations?).

For many workers in office and admin roles, the answer across all four dimensions is discouraging. Median wages in these occupations tend to leave little room for savings. The skills involved — filing, phone management, basic data processing — are precisely the skills AI handles well, and they do not transfer easily to the occupations that are growing. [Claim — structural inference] When your core competency is the very thing being automated, the usual advice to "reskill" rings hollow without substantial support.

Geography Makes It Worse

[Fact] The study finds that vulnerability clusters geographically in ways that might surprise you. Rather than concentrating in major metro areas, high-exposure, low-adaptation workers are disproportionately found in college towns (like Laramie, Wyoming and Huntsville, Texas), state capitals (like Springfield, Illinois and Carson City, Nevada), and smaller cities across the Mountain West and Midwest.

[Claim — researcher analysis] The logic is intuitive once you see it: these places often have a few dominant employers — universities, government agencies, regional hospitals — that employ large numbers of administrative and clerical staff. The local job market is shallow, meaning there are fewer alternative employers if your position disappears. And these towns are often far from the kind of large, diverse metropolitan areas where displaced workers might find new roles.

This geographic pattern matters for policy. [Claim] Federal retraining programs designed around major metro areas may completely miss the workers who need them most. A displaced receptionist in Carson City faces a fundamentally different challenge than one in San Francisco.

What This Means for Workers in These Roles

If you work in one of these occupations, the Brookings findings are sobering but not hopeless. The study is not predicting that 6.1 million people will lose their jobs tomorrow — it is identifying who is least prepared if displacement accelerates.

The actionable takeaway is about building adaptive capacity along the four dimensions the researchers identified. Financial resilience — even modest emergency savings — provides crucial time to retrain. Skill adjacencies matter: an administrative assistant who also manages a department's project tracking or budget analysis has more transferable skills than one who primarily handles scheduling. Geographic awareness is important too — understanding the depth and breadth of your local job market can inform timing decisions about career changes.

For employers and policymakers, the study is a warning against one-size-fits-all approaches. [Claim — policy implication] A retraining grant that works for a 28-year-old data entry clerk in Austin will not work for a 55-year-old medical secretary in rural Wyoming. The 6.1 million are not a monolith — they are individuals whose specific combination of constraints requires targeted, localized support.

You can explore how AI affects specific occupations in more detail on our administrative assistants, office clerks, receptionists, and medical secretaries pages, including task-level automation rates and projected employment changes.

Update History

• 2026-03-21: Initial publication based on Brookings Institution report (January 2026)

Sources

• Manning, S., Aguirre, T., Muro, M., & Methkupally, S. (2026). Measuring US workers' capacity to adapt to AI-driven job displacement. Brookings Institution.

---

_This article was produced with AI-assisted analysis. All statistics are sourced from the referenced research. For our full methodology and AI disclosure, see our AI disclosure page._