AI-related layoffs keep coming. But there’s more to the story
We’re barely into the new year, and companies are shedding more jobs due to AI. But experts say that’s not the omen you think it is.
If you scanned the headlines in 2025, it was hard not to feel a sense of dread. Between big tech giants and major banks, the narrative was clear: Artificial Intelligence was coming for our jobs.
We saw nearly 55,000 layoffs attributed to AI at companies like Amazon and Microsoft. As the new year rolled in, the cuts didn’t stop. Citigroup announced over a thousand job cuts, with CEO Jane Fraser warning in a memo that automation and AI would fundamentally reshape how work gets done. Meanwhile, Meta sliced another 1,500 jobs from its virtual reality division to double down on its AI investments.
It feels like a robot takeover, right?
The Reality Check
Despite the scary memos and the splashy headlines, experts say we might be getting ahead of ourselves. If you look past the press releases, there is actually very little evidence that AI is displacing workers en masse or drastically changing how business operates—yet.
A recent analysis by the Brookings Institution and the Budget Lab at Yale University found something surprising: The proportion of workers in jobs that are "ripe" for AI disruption hasn’t actually budged since ChatGPT launched back in 2022.
So, what is going on?
According to Martha Gimbel, executive director of the Budget Lab, what we are seeing is a labor market confused by a lot more than just algorithms.
"What we’re seeing overall right now is consistent with a labor market that has been hit with a lot of uncertainty in the macroeconomic environment," Gimbel explains. She points out that changes in immigration policy are curbing employment growth, making it difficult to interpret the raw numbers. "If you look for any signs of changes that seem to be due to AI, those are not yet showing up."
The Productivity Puzzle
There is one argument that *does* suggest AI is quietly working behind the scenes: productivity.
In the third quarter of 2025, labor productivity climbed by 4.9%—the highest increase in two years. To many economists, this looks like proof that AI is finally boosting efficiency. But Gimbel isn't buying it.
She argues that productivity is a "noisy" metric, especially when looking at just a single quarter. "Productivity growth will be really high or really low in one quarter," she says. "And if it fits their preferred narrative, people will jump on that."
Furthermore, the lingering effects of the pandemic are still skewing the data. When companies fired scores of lower-wage (and statistically less productive) workers during the pandemic, productivity numbers artificially spiked. As those workers were rehired, numbers leveled out. It wasn't a technological revolution; it was a statistical quirk.
The "Rework" Problem
There is also the messy reality of actually using AI. While the promise is efficiency, the practice often involves cleanup.
A new report from Workday found that nearly 40% of the time saved by using AI is immediately lost to "rework." On average, workers spend about 1.5 weeks every year simply correcting or fixing content generated by AI. That’s a full work and a half lost just to babysitting the robots.
When Will the Shift Happen?
This doesn’t mean AI is harmless—it just means the job apocalypse isn't here yet. Unemployment figures remain stable, and while sectors like professional and business services have seen declines, it’s too early to pin that solely on automation.
Gimbel suggests that the true test won't be a healthy economy, but a sick one. "The place to start looking for the impacts of AI is when we have a recession," she says. "That is usually when technological change really takes off."
For now, Gimbel is keeping a close eye on high-adoption sectors like tech, the arts, and education. While the current panic may be overblown, the impact is coming.
"It would be unusual for a new technology to have no impact on the labor market," Gimbel concludes. "We just still need to find out how fast, and where."