For White-Collar Workers, AI Also Stands for “Apocalyptic Insecurity”

“It doesn’t feel good to build a flow with my labor that could potentially replace me—not just another person, but me.”In Greenville, South Carolina, photographer Celina Odeh, also 30, feels it too. As a digital tech on commercial shoots, her work has become a parlor game of elimination: Which skills will AI kill off next? She’s taken up knitting to stay sane. As humans fade into background and AI enlists workers in their own demise, it feels like it’s Soylent Green meets The Hunger Games: Middle-class labor is forced to eat itself. For white-collar workers across the country, the upheaval is psychological and existential as much as technological. They are grappling with what we call apocalyptic insecurity: the realization that something massive is underway but there’s no clear timeline or playbook. Everything moves at incomprehensible speed. It’s made work itself into an uncertainty, with dark impacts on our behavior, careers, and health of mind and body. A massive 71 percent of Americans are now scared that AI will steal livelihoods. Tech leaders issue Magic 8 Ball musings: white-collar jobs gone in months; half of entry-level jobs wiped out in five years; or, depending on who’s talking, jobs will simply “transform.” But how? When? What, if anything, is the plan? Of course, the narrative that AI is destabilizing middle-class work is both very real and hype. But that’s exactly what makes it maddening—it’s close enough to be terrifying but too ill defined for people to act on. Knowledge workers drift in a twilight of not knowing. That can hurt even more than jobs. One researcher in business psychology we spoke to framed job insecurity as a diffuse condition that makes us feel less in control, which is ultimately paralyzing. A growing body of research concurs, linking chronic workplace uncertainty to anxiety, depression, burnout, and even physical symptoms. A new report on AI-driven insecurity establishes a novel work hazard: “AI replacement dysfunction.” Brooklyn-based brand copywriting expert Lexi, 56, is living that insecurity. After 30 years perfecting her craft, she now watches AI spit out this kind of work in minutes, albeit with near-zero personality. She’s burning through her savings, unsure what comes next. “I made my living for decades helping people figure out what to say and how to say it,” says Lexi. “But now it kind of doesn’t matter if things are well written. It’s just AI writing things for AI to read. It frees me up to do my own writing, although I’m not sure who’s reading that, either.”Labor sociologist Victor Chen of Virginia Commonwealth University says that part of why the current automation wave is so unnerving is that there’s no clear answer for what workers should do. “There’s no obvious solution like ‘Get a college degree’ or ‘Get a STEM degree,’” Chen says. “That makes it difficult to plan your career, much less your life and your children’s futures.” Every point in a career adds a new layer of AI-infused insecurity, a bot-ified version of the stages of man. The youngest should be better able to adapt but are struggling to land entry-level jobs and are often the first fired. Older employees worry it’s too late for a second act—a dubious concept to begin with. And for someone in midcareer, the pressure can be suffocating: kids to support, bills piling up, and a future that keeps erasing itself. Ria Julien, a literary agent and lawyer representing everyone from blue-collar workers to tech employees, sees the same mood everywhere. “It’s an absolute climate of fear,” she says, one driven by a growing sense that the middle-class script is fraying. When some of her more well-off clients’ earnings drop to zero, and the idea that they’ll find new work is far from guaranteed, “it is absolutely devastating.” Of course, in America’s brand of pitiless capitalism, job insecurity has long been a feature not a bug. Over the last decades, we have seen globalization ship work overseas, financialization siphon profits from those who do the work, and waves of technology set the labor market spinning again and again. AI is simply the latest vector. What’s different, though, is its speed and scale, and the epochal dread it provokes about human relevance.Some economists are optimistic about automation, averring that AI could boost wages and job quality. (We find them cheery to the point of delusional.) But many others are profoundly skeptical. Nobel laureate Joseph Stiglitz observes to us that “a lot depends on how we manage this technological transition,” and he worries that history offers us little reassurance. “Looking back, we didn’t manage industrialization well, and we don’t appear to be managing this well, either,” he warns.The market can’t be relied on to retrain workers or prepare them for this shift. William Lazonick, president of the Academic-Industry Research Network, calls out American corporations structured to extract value for shareholders, not invest in workers: “Companies want to use AI as a way to reduce costs, increase profits, pay massive dividends, and do stock buybacks,” he warns. “What does any of that do for human workers?” Thomas Ferguson, research director at the Institute of New Economic Thinking, is blunt about the power imbalance: “The problem is that workers don’t run American companies: Business does. Conventional economic academic accounts don’t recognize this enough.”In many ways, this is a story that has been on repeat for over a century. In the early twentieth century, Taylorism—Frederick Taylor’s so-called scientific management—recast skilled machinists as extensions of the assembly line. Thinking belonged to the boss; the worker was there to obey. Autonomy and morale withered. Today, this same logic is migrating to office towers. Shortsighted companies roll out AI to surveil and dictate and, at the same time, de-skill and diminish. Some white-collar workers tell us it’s an assault on their minds.Take Claire, 34, a data scientist in New York City, watching her role at a security camera start-up blur as AI agents take over most of her coding work. “Even three months ago, I was doing a completely different job,” she says. “Now I’m not even sure what to call myself. AI engineer? Manager of AI agents? I don’t know.” Overseeing multiple agents has created what some are calling “AI brain fry”—a mental overload from multitasking and AI babysitting. In handing off the work, Claire senses some essential part of herself slipping away. “I miss the flow of coding, the creative problem-solving, the thrill of wrestling with abstract ideas,” she admits quietly. “I’m afraid of losing my dreams to AI.” Work has started to feel oddly alien. Employees tell us about a tech-weirding of office jobs as emails, meetings, and even casual interactions are machine-filtered. Jade gets chirpy emails from management at her insurance tech firm insisting that AI is there to “help,” not replace. “Ironically, those emails are the most AI-sounding writing,” she says. “It’s creepy.”What economists call information asymmetry only amplifies the unease. Ferguson says it’s one thing to know that what we call the “robot gaze” (which monitors worker data to maximize profits) exists. But it’s another to have no clue what’s being tracked and whether it might be used against you. “And you might want to learn AI on your own computer,” he advises. “Lest you teach your employer how to eliminate you.”Invisible algorithmic eyes can watch every keystroke, spy emails patterns, and predict who might be “at risk” of underperforming. Executives say it’s for our own good; that’s what they always say.People understand that their work—and their worth—is being dictated by AI, and they’re losing the parameters in which to succeed. Sociologist Janet Vertesi, who studies AI and robotics at Princeton, puts it like this: “We are effacing expertise instead of enabling expertise.” Giorgio Ascoli, a neuroscientist at George Mason University, says that in his field, the formative years of learning by doing are disappearing, and without that, “you’re cutting your own roots,” leaving a workforce that never gains the experience needed for the part of the scientific method that demands human capability.Lazonick calls this both shortsighted and backward. By cutting loose the employees who carry institutional memory, judgment, and hard-won skills, companies are tossing out the very knowledge that makes innovation actually work. “For creativity and innovation to happen, you need a workforce that’s equipped, engaged, and actually has a stake in what comes next,” he says. Bots can’t provide that spark.Diana Enriquez, a sociologist who studies large-scale automation, warns of companies following a “tech playbook” pressuring workers to trust technology, even when the algorithm is, well, wrong. Middle managers are forced to claim successes the system hasn’t actually delivered. Why? Because the C-suite mindset in tech companies, says Enriquez, holds that “workers are a problem that needs to be solved.”In offices, the impact is obvious. An AI bot handles Jade’s meeting minutes—and half the time, they don’t even make sense. “You end up spending more time fixing them—a person could have done it better and faster.” In this example, AI becomes the problem to solve. Some, like Joanna Popper, CEO of the film and AI content company Laurel Beach, see a split screen for workers. On one side, she finds opportunities. On the other side, these opportunities aren’t evenly spread. “AI tools can act as a lever, letting creators move faster and cheaper, which could help those historically sidelined,” Popper says. “But it also shrinks how many workers are needed.” Natasha Lennard, author of a forthcoming book on the philosophy of uncertainty, warns that we can’t let large language models put humans in a subordinate seat. “We’re told, ‘You don’t understand this; it’s beyond what you can imagine.’ Or, ‘AI will doom us all; AI will save us all,’” says Lennard. The real danger, she points out, is what she thinks of as “AI determinism”—unquestioned assumptions about how technology will develop. The real story is about power. Darrick Hamilton, chief economist of the AFL-CIO, says that dealing with AI labor uncertainty starts with asking the right questions. Who does AI actually serve? What’s it for? Who benefits?Collective action offers a way out of the fog. Hamilton points to a 2022 Gallup poll showing union approval at its highest level since 1965. It’s true that white-collar workers remain far less unionized in the United States than their blue-collar counterparts, but with apocalyptic uncertainty knocking on their doors, they might want to join the club.Which brings us to a potential plot twist. For years, office workers were told they were safer from the turmoil that marked many blue-collar jobs. But with the “middle precariat” feeling the AI squeeze, there’s a chance for something truly powerful—a cross-class bloc, united in its need for stability and a say in the future.It wouldn’t be the first time. In the New Deal era, shared economic shock pushed white- and blue-collar workers into a broader labor coalition. It was imperfect and incomplete, but it showed that collective angst can bloom into collective power. Early signs of AI-driven organizing are already visible: U.S. entertainment striking over AI, Germany’s Verdi negotiations, and the Communications Workers of America, or CWA, setting up principles for AI. Across industries, workers have wrested victories from resistant bosses that limit AI’s impact on their jobs, including advance notice, human oversight, and safeguards against automatic replacement.One thing we can be certain about: Human relatedness, expertise, insight, and imagination can’t be substituted. Protecting them will require smart policies, retraining programs, and worker participation. The tech overlords may rattle on about AI and the billions they are investing and making. Few are talking about what’s actually required for human beings and for shoring up America’s middle class: education, dignified and well-paying jobs, and robust social safety nets.If the middle class is going to thrive or even persist, our government has to step up to set rules, enforce protections, and put in place bottom-up AI policies that reflect human needs, making sure that the benefits of AI don’t just go to corporate boards. And if the federal government drags its feet, then states have to take action. Many are already doing just that, despite the murky political climate.We don’t have to stand back and let AI write the script for the middle class. We still get to choose. For now. This story was co-published and supported by the journalism nonprofit the Economic Hardship Reporting Project.