You’re watching a manager close a laptop after a layoff call and feel the air change in the room. I remember that same silence the first time a team of writers I edited got cut—no slaps on the back, only Slack pings. You feel the floor tilt beneath you—like a boat listing in fog.
I read the Brookings Institution study so you don’t have to. It measures how exposed different jobs are to AI and, more important, how likely workers are to find new paid work if those jobs disappear. The headline: millions are well positioned to move; a stubborn minority may not be.
At a Brookings desk, the researchers mapped exposure to AI and workers’ skills
The team matched how easily AI could perform task-level work with people’s age, education, union status, local labor markets, and previous job moves. I found the method blunt but useful: it doesn’t predict the future, it measures resilience.
The study flags 37.1 million U.S. workers in occupations with the highest exposure to AI. Of those, about 26.5 million have above-median adaptive capacity—so you, or a colleague, may be closer to a new role than it feels now.
In a downtown staffing center, the reality is uneven
Not every place has the same options. Large metro areas with dense tech and service markets offer more openings; midsize towns and university hubs often do not.
Which jobs are most at risk from AI?
Writers, customer service reps, translators, clerical and administrative roles show high exposure. But exposure doesn’t equal permanent unemployment. Web developers, marketers, and IT workers also face high exposure, yet their local networks and skill sets make it easier for them to find new gigs—especially in cities with strong labor markets and venture activity tied to OpenAI, GitHub Copilot, or cloud platforms from Amazon and Microsoft.
At a corporate all-hands where layoffs were announced, the human cost was obvious
Executives from Amazon to Pinterest and Block have pared teams this year, and CEOs like Jack Dorsey have framed cuts as an acceptance that AI will substitute for some human work. Those announcements ripple beyond PR: they hardwire fear into payroll spreadsheets.
Goldman Sachs told markets that AI investments had “basically zero” contribution to U.S. GDP growth in 2025, while the Federal Reserve Bank of Dallas says it doesn’t expect mass displacement over the next decade. Those statements give you two competing lenses: investment hype vs. measured macro evidence.
At a city workforce office, the most vulnerable clusters are obvious
The Brookings numbers place about 6.1 million workers in a troubling position: high AI exposure and low adaptive capacity. Most of these jobs are clerical and administrative, and roughly 86% of those workers are women.
Can workers retrain after AI layoffs?
Yes—many can. But retraining is not automatic. It depends on local employer demand, access to short, employer-recognized credentials, and social supports. I’ve seen successful reskilling programs that partner with community colleges, LinkedIn Learning, and local employers; I’ve also seen training that leaves people with certificates and no hires.
At a regional labor market meeting, solutions looked familiar and incomplete
Policymakers want to focus help where it will matter most. Brookings hopes the study pushes resources to the people and places that need them: women in clerical roles, workers outside major metros, union members who may face industry-wide shifts.
Workers who can reskill are like nimble chess pieces, able to slide into new squares if the board changes. But the board itself—hiring pipelines, employer demand, broadband, childcare—must be in play for those moves to actually happen.
At a startup product demo, the promise and the mismatch were both visible
Companies pour time and capital into AI tools—OpenAI’s ChatGPT and GitHub Copilot are now common in workflows—but surveys from the NBER and executives suggest productivity gains are mixed. You can adopt a tool; that doesn’t guarantee immediate growth or job creation.
What I want you to take away: the numbers are not destiny. If you’re in a high-exposure job, your odds of finding work are shaped by where you live, who hires locally, and whether your skills map to openings. If you run policy or hiring, the clearest lever is geographic and occupational targeting: move money and programs to places with the weakest markets before a shakeout widens.
How many US workers could be displaced by AI?
Estimates vary widely. Brookings focuses on exposure and adaptability—37.1 million in high-exposure jobs today, with roughly 26.5 million likely able to find other work and about 6.1 million at notable risk. Those are snapshots, not fate.
I’m not here to calm or alarm you—I want to point to what matters. Watch where hiring is concentrated (big cities, cloud services, venture-funded firms), who the affected workers are (clerical roles, many women), and where public dollars are scarce (midsize and Mountain West markets). The question now is less whether AI can replace work and more whether the institutions that support workers can move as quickly as the code—so what will you push for next?