A Nobel Prize-winning economist is warning that AI could triple U.S. unemployment within a decade, and the people building these systems know it. The music producer teaching AI to write songs thinks it'll never replace him because it's never had a bad breakup. The economist who studies labor markets for a living is a little less optimistic.
The Number That Should Be on Every Front Page
Here's what we know. According to a September report from McKinsey Global Institute, AI could force 11 million U.S. workers into entirely new careers by 2035. That's not a vague disruption forecast. That's 11 million people told their skills don't apply anymore.
And that's the moderate scenario. Nobel Prize-winning economist Daron Acemoglu told CBS News that in a worst-case outcome, unemployment could triple in the next decade if the country stays on its current course and does nothing. Triple. As in, if unemployment sits around 4 percent today, we're talking about something in the range of 12 percent. That's Great Recession territory, except permanent.
Nearly three-quarters of Americans already fear AI is coming for their jobs, according to a Pew Research study. So the public isn't asleep on this. They're watching it happen in real time and hoping someone with actual power is paying attention.
This Is Not Like Previous Tech Shifts. Stop Saying It Is.
Every time a new wave of automation hits, someone in a nice suit explains that this is just like the Industrial Revolution, and workers eventually adapted, and everything turned out fine. Acemoglu, who has spent his career studying exactly this, wants you to know that argument doesn't hold here.
"What we are living through with AI is unprecedented," he told CBS News. "The first phase of the Industrial Revolution took 80 years. Today it's taking place in one or two years and it's taking place across many different sectors at the same time."
Eighty years versus two. Read that again. Previous generations of displaced workers had time to retrain, to relocate, to have kids who grew up learning different trades. The AI disruption is happening faster than any workforce policy, retraining program, or social safety net is designed to handle. The usual reassurances don't scale at this speed.
Who's Getting Hit First
Young people and college graduates are already feeling it. Research from Stanford shows hiring is down for young workers in the most AI-exposed jobs, including software developers. A separate U.S. Census study found that hiring and wages among recent college graduates in AI-exposed majors have declined. CBS News reports both findings.
Clara Shih, who oversaw major AI divisions at both Salesforce and Meta before quitting her tech job in January, put it plainly. "Traditionally, the tasks that you might assign to a young person, maybe doing some market research or creating the first draft of a memo, AI is just really good at doing those now," she told CBS News.
She described a transformation she witnessed firsthand: products that used to require dozens of people to conceive, prototype, and deliver now get done by a handful of people using AI agents. She got so alarmed by what this means that she left her industry job and started a nonprofit focused on young workers navigating the shift. That's the kind of thing you do when you've seen the data and you can't unsee it.
The AI Training Gold Rush (For About 100,000 People)
Mercor, a company that trains AI models like ChatGPT and Claude across fields from finance to poetry, has assembled over 100,000 freelancers to teach AI what they know. CBS News reports that AI training is currently America's fourth-fastest-growing job category on LinkedIn, with gig pay ranging from $20 to $200 an hour. The company's CEO, Brendan Foody, is calling it the career of the future.
Acemoglu has a different read. "The number of people they're employing in these positions is small relative to the number of people who will be replaced," he said. "I mean, that's the whole point of automation."
One hundred thousand trainers versus eleven million displaced workers. The math is not ambiguous. The gold rush framing is not wrong exactly, but it's doing the same work it always does: giving everyone a reason to believe they'll be one of the miners, not one of the towns the mining company leaves behind.
What Can Actually Be Done
Acemoglu is not a Luddite. He told CBS News that companies could intentionally design AI to make workers more capable rather than expendable, and that the government could restructure the tax code to incentivize hiring actual human beings. Neither of those things is technically complicated. They require political will, which is a different and considerably scarcer resource.
Shih, drawing on her family's experience watching factory work disappear in Ohio after globalization, says the first thing that needs to happen is honesty. "I think, you know, sugarcoating that this AI utopia, maybe we'll get there, but there's a lot of hurt that could happen between now and there," she told CBS News.
She's right. The gap between the tech industry's utopian pitch and the economic reality of what automation does to workers and communities is not a misunderstanding. It's a choice. Someone is choosing to emphasize one story over the other. It's worth being clear about who that benefits.
The Dingo Take
Imagine if 11 million coal miners were told by 2035 their jobs were gone, and the government's response was: great news, a hundred thousand of you can teach robots to mine. Republicans would be holding hearings every other week. There would be emergency legislation. There would be rallies. The political energy around protecting those workers would be enormous and loud. But when the jobs evaporating are held by software engineers, paralegals, junior marketers, and recent college grads, the response from Washington is roughly: let the market sort it out. Interesting who gets the safety net and who gets the TED Talk.
The window Acemoglu is talking about is not metaphorical. Tax policy, labor law, and corporate incentive structures take years to change even when there's urgency. Right now there is no urgency in any branch of government that matters. The Trump administration has spent its energy cutting the federal agencies that would normally study and respond to labor market disruptions, not building new frameworks for a workforce crisis. This is not a coincidence. Regulatory paralysis during a period of rapid automation is itself a policy choice, and it favors the people who own the automation.
Acemoglu said the future is uncertain enough that human choices still shape it. That's true. But choices require choosers who are paying attention, who have the power to act, and who feel some accountability to the people who will bear the cost. Right now, the people building these systems are accountable primarily to their investors. That's the whole problem, and no one in a position to fix it is treating it like one.
Comments