For the second straight month, artificial intelligence topped the list of reasons companies gave for cutting workers. Challenger, Gray & Christmas, the outplacement firm that tracks corporate layoffs, reported Thursday that AI was the leading cause employers cited for job reductions in April. Thousands of positions have already disappeared. And the firms doing the cutting, Cloudflare, Coinbase, and others, aren't hiding behind euphemisms anymore.
But the real story isn't the layoffs that make headlines. It's the slower, less visible transformation happening inside companies that keep their headcounts roughly intact while hollowing out the substance of what workers actually do. That shift deserves far more scrutiny than it's getting, especially from the workers living through it.
A CNN Business report published May 10 laid out the emerging consensus among consultants and tech executives: AI is not, in most cases, eliminating whole jobs. Instead, companies are automating specific tasks within a position and then "recalibrating" what's left around responsibilities that still require a human being. The framing is meant to be reassuring. Whether it should be is another question entirely.
Alexis Krivkovich, a senior partner at McKinsey & Company who leads the firm's People and Organizational Performance Practice, offered the soothing version of events. She told CNN that current AI and robotics technology automates very few jobs entirely.
"It's very few jobs that are actually entirely automated away by the current AI and robotics technology that's out there."
But Krivkovich herself cited McKinsey research showing AI is technically capable of automating 57 percent of work-related activities. Not 57 percent of jobs, 57 percent of the tasks that make up those jobs. She described what AI targets as "pieces and parts" of a role rather than the whole thing.
Read that number again. If a machine can handle more than half the tasks in a typical position, the employer doesn't need to fire you outright. It just needs fewer of you. Or it needs you to do something different, something the company may or may not train you for, and something that may or may not pay as well.
That distinction matters. When Cloudflare recently slashed more than a thousand jobs and pointed to AI, the market punished the company's stock. But the quieter version, trimming responsibilities, flattening roles, expecting more output from fewer people, doesn't generate the same backlash. It just erodes the bargaining power of the workers who remain.
Nitin Seth, cofounder of the digital services and consulting firm Incedo, claimed his company helps clients boost productivity using AI by at least 20 to 25 percent without reducing staff at the same scale. He framed the math as a practical barrier to mass layoffs.
"You can't take one quarter of Lisa, one quarter of Jessica, one quarter of Nitin and one quarter of somebody else and make it one person."
Fair enough. You can't literally Frankenstein four fractional workers into one. But you can decline to hire the next person. You can let attrition do the work. You can redefine a role so that one employee now covers ground that two used to share. Seth's point is technically correct and practically incomplete.
The productivity gains are real. The question is who captures them. If AI makes each worker 25 percent more productive, does the worker see a raise, or does the company pocket the margin and slow-walk its next round of hiring? Nothing in the current landscape suggests employers are rushing to share the windfall with labor.
Meanwhile, Coinbase's CEO cut hundreds of workers and said plainly that AI would replace them. That kind of candor is at least honest. The consultant class prefers softer language, "recalibrating," "reshaping," "augmenting." The result for the displaced worker is the same.
The adoption data tells its own story. A September survey from Google's research arm found that 90 percent of tech workers were already using AI in their jobs. Stack Overflow reported that 84 percent of respondents either use AI tools in the software development process or plan to.
Those are not fringe adoption rates. They describe a workforce that has already absorbed the technology into daily routines. Boris Cherny, head of Claude Code at Anthropic, suggested that companies may soon adjust job titles for software engineers to reflect broader responsibilities beyond coding, an acknowledgment that the role itself is being redefined around what AI can and cannot do.
Microsoft, in a report released last week on how AI is changing work, acknowledged the anxiety directly: "The anxiety around AI at work is real, from fears of job loss to the pressure to keep up with rapidly evolving technology." That sentence, buried in a corporate report, is more revealing than any reassurance from a consulting firm. The company building the tools admits the pressure is real.
And the pressure doesn't fall evenly. It lands hardest on workers in the middle, the ones whose jobs involve routine cognitive tasks that AI handles well, but who lack the capital, credentials, or connections to reinvent themselves overnight. Those are the people who built careers on reliable, repeatable skills and now find the floor shifting beneath them.
The CNN report, like most coverage of AI and employment, leans heavily on expert commentary from consultants and tech executives, people whose firms profit from AI adoption. McKinsey advises companies on how to deploy AI. Incedo sells AI-driven productivity services. Anthropic builds AI tools. Their reassurances are not disinterested.
None of this means they're wrong. Partial automation may indeed preserve more jobs than it destroys, at least in the near term. But the framing, "AI isn't taking your job, it's just changing it", obscures a harder truth. A job that's been stripped of its most skilled components, loaded with new expectations, and pays the same or less is not the same job. It's a demotion dressed up as adaptation.
The open questions are significant. Which companies beyond Cloudflare and Coinbase have cut workers for AI-related reasons? What are the actual headcount numbers? What happens to workers whose tasks get automated but who aren't retrained? The data on displacement is still thin, and the people collecting it have reasons to present the rosiest possible picture.
The broader economic consequences extend beyond any single layoff announcement. Goldman Sachs has warned that AI is already contributing to rising inflation and household costs, adding another dimension to the disruption that workers face, not just job instability, but higher prices at home.
Washington, for its part, has barely begun to grapple with the regulatory implications. Investor Paul Tudor Jones has warned that the federal government is already behind on AI regulation, a gap that leaves workers with few institutional backstops as the labor market shifts beneath them.
Concerns about AI replacing workers have simmered for over a year now, and the pattern is becoming clear. Companies adopt AI tools, extract productivity gains, and then make staffing decisions that reflect the new math. Sometimes that means layoffs. More often, it means hiring freezes, role consolidation, and a slow squeeze on the people who remain.
The consultants say not to worry, your job isn't disappearing, it's just evolving. But evolution in nature doesn't come with a severance package. It rewards the adaptable and discards the rest. The question for American workers isn't whether AI will take their jobs whole. It's whether the jobs that survive will still be worth having.
When the people selling the tools are the same ones telling you not to worry, it's probably time to worry.