The Worst-Case Future for White-Collar Workers

This article was featured in the One Story to Read Today newsletter. Sign up for it here.

White-collar workers are getting nervous, with good reason. Sure, 98 percent of college graduates who want a job still have one, and wages are ticking up. Sure, some companies that cite the labor-saving, efficiency-promoting effects of ChatGPT and Claude as they let employees go are just “AI washing”—talking about algorithms to distract from poor managerial decisions.

But the labor market for office workers is beginning to shift. Americans with a bachelor’s degree account for a quarter of the unemployed, a record. High-school graduates are finding jobs quicker than college graduates, an unprecedented trend. Occupations susceptible to AI automation have seen sharp spikes in joblessness. Businesses really are shrinking payroll and cutting costs as they deploy AI. In recent weeks, Baker McKenzie, a white-shoe law firm, axed 700 employees, Salesforce sacked hundreds of workers, and the auditing firm KPMG negotiated lower fees with its own auditor. Two CNBC reporters with no engineering experience “vibe-coded” a clone of Monday.com’s workflow-management platform in less than an hour. When they released their story, Monday.com’s stock tanked.

Maybe algorithm-driven changes will happen slowly, giving workers plenty of time to adjust. Maybe white-collar types have 12 to 18 months left. Maybe the AI-related job carnage will be contained to a sliver of the economy. Maybe we should be more worried about a stock-market bubble than an AI-driven labor revolution.

I don’t think anyone knows what will happen, or even what is happening now. AI technology is changing at an exponential pace, and changing the workforce in a thousand hard-to-parse ways. But if AI quickly eliminates white-collar work, the country is going to end up in something much stranger than a downturn, and something much harder to recover from too.

The United States is adept enough at handling the labor-market damage caused by recessions. Congress slashes taxes, writes stimulus checks, and fattens unemployment-insurance payouts. Washington amps up infrastructure spending and patches holes in the budgets of state and local governments. The Federal Reserve drops interest rates down to zero and purchases hundreds of billions of dollars of safe assets, making borrowing cheaper for families and encouraging businesses to invest. Demand increases, pushing the unemployment rate down and GDP up.

But if white-collar layoffs cause a downturn, Washington might not be able to restore hiring and lift consumer spending as it has done before. Businesses wouldn’t need the skills workers possess. Firms wouldn’t want to hire the legions of accountants, engineers, lawyers, middle managers, human-resources executives, financial analysts, PR types, and customer-service agents they just laid off. (Writers would be fine, I choose to believe.) The United States would have a “structural” unemployment problem, as economists put it, not a “cyclical” demand problem.

The country has struggled with structural joblessness in the past, but the problem has tended to afflict blue-collar workers, not white-collar ones. The labor market is normally womblike in the security it delivers the college-educated. The downsizing trend of the 1990s did not actually result in the pool of office jobs shrinking. Even during the Great Recession, the unemployment rate for workers with a bachelor’s degree never went higher than 5.3 percent. For workers with only an associate’s degree or some college credits, it hit 9 percent; for workers with only a high-school diploma, 11.9 percent.

In this new economic paradigm, the educated and well-to-do would fare worse than their less-educated and lower-income neighbors. The main bulwark against joblessness—the unemployment-insurance system—wouldn’t meet the challenge. Payments last for only so long: six months at most currently, 18 months during the coronavirus pandemic. If AI eliminated the need for office work, many people would be unemployed for years; their earnings would potentially fall, their mental health would deteriorate, and their chance of finding a new position would diminish with every month that passed. Plus, the unemployment system isn’t designed to support six-figure earners. Most state payments max out at $500 or $600 a week, a quarter of what many upper-middle-class employees earn. Young workers would face their own challenges: The pool of entry-level white-collar jobs would shrink, as it already is, pushing recent graduates’ income down for years, even decades.

If rich households cut back on spending, many businesses that have nothing to do with AI—grocery stores, gas stations, retail shops, hairdressers, restaurants—would suffer and the labor market would keep deteriorating. The housing market would begin to falter; home prices in many regions would fall and fewer properties would go to market. Tax revenue might fall. Bond yields might go up as the country’s deficits increased, making it harder for the Fed to stimulate investment. Rank-and-file workers’ incomes would drop at the same time that executives using AI to cut costs would see their wealth increase. Inequality would rise to even higher heights. The 0.01 percent would pull away from the 0.1 percent, which would pull away from the 1 percent.

White-collar workers would go through what blue-collar workers went through beginning in the 1970s. Advances in machine technology improved productivity and depressed employment in Detroit; Pittsburgh; Gary, Indiana; and Worcester, Massachusetts. Rust Belt communities fell apart and never recovered. Then China joined the World Trade Organization, and globalization spurred another round of job losses, causing even more permanent damage. Affected workers ended up poorer, less happy, and less healthy. They died sooner. Their kids were worse off too.

To get the economy going again during the AI transition, the country would need to figure out how to get white-collar workers back to work. And I really mean figure out—essentially from scratch. Existing workforce-training initiatives have yielded “muted” and “inconclusive” results. Existing displaced-worker initiatives have “dubious” value. These kinds of programs offer net-negative value to participants, taxpayers, and “society as a whole,” studies have found. Community-college programs have the strongest track record, but the overwhelming majority of office workers have a two-year or four-year degree already. The country doesn’t have answers that don’t involve upskilling, education, or trying and failing to compensate the losers.

Silicon Valley leaders are enamored of one policy Hail Mary: the establishment of a universal basic income, in which the government would provide all adults with $1,500 a month or so, no strings attached and in perpetuity. It’s not as crazy an idea as it sounds; think of it as Social Security for everyone, or an extension of the earned-income tax credit to families without any earned income. The cash would ensure that every family kept its head above water, and redistribute the wealth generated by rising productivity. “People will be freed up to spend more time with people they care about, care for people, appreciate art and nature, or work toward social good,” Sam Altman of OpenAI has argued.

But UBI is a dystopian outcome, not a utopian one. For families to thrive in this new post-work paradigm, the government would need to redistribute a lot more than $1,500 per person per month, necessitating confiscatory taxes on corporations—taxes they would fight tooth and nail. The bigger problem would be that Americans would hate a world without work, where the jobless rate floats at 30 percent instead of 4 percent. Many Americans like working. They like having somewhere to go during the day. They like trading watercooler stories with their colleagues. They like getting promoted and starting their own firms. They are proud of earning a living. Long-term unemployment destroys people’s mental and physical health, and generates toxic societal unrest. Politicians love saying that Americans want a hand up, not a handout—and they aren’t wrong.

Perhaps Americans would adjust, and American culture would adjust in turn. Maybe people would become more interested in and fulfilled by leisure, art, and other activities. But it’s hard to imagine social capital being divorced from actual capital. It’s even harder to imagine the country becoming more egalitarian and its politics more participatory, instead of developing a hyper-wealthy techno-oligarchy and a dispossessed and teed-off underclass. Some people would spend their time gardening, volunteering, painting, exercising, or learning. Others would spend hours a day watching short-form videos and never leave their apartments.

I don’t really think society will come to this. Throughout history, people have used technology to become more productive and prosperous without reducing the demand for human labor overall. Then again, maybe I’m just in denial.