By Rob Wile
Gareth Patterson spent nearly a year shutting out friends, missing gym sessions, and grinding through coding tutorials just to transition from sales into a software engineering job in New York City.
“I gave up everything to make this career change,” he said. “Now I’m hearing that AI could do my job — or at least that’s what people are saying. It’s scary.”
Gareth’s story is far from unique. Across the tech world, there’s growing anxiety — not just about mass layoffs, but about what’s driving them. And many fingers are pointing toward artificial intelligence.
Recent reports and layoffs are fueling fears that AI may be replacing people faster than it’s helping them. At Microsoft, where CEO Satya Nadella says AI now writes 30% of the company’s code, recent layoffs in Washington state hit software engineers hardest, making up more than 40% of the 2,000 job cuts.
But is AI really making workers redundant — or is it just a convenient scapegoat during a rocky economic period?
A new study from METR, a nonprofit that evaluates AI models, suggests the truth may be more complicated. The research found that AI tools slowed down software engineers, making them 19% less efficient on average. Strikingly, even after experiencing the slowdown, developers still believed AI had made them 20% faster — a perception gap that stunned the study’s authors.
“This contradiction is telling,” said Ben Hilton, a researcher at METR. “People want AI to work for them. They believe in the promise — even when the evidence shows it’s not helping yet.”
Still, the tech industry can’t seem to get enough of AI. Just last week, Google signed a $2.4 billion licensing deal with AI coding startup Windsurf, even poaching its CEO. Cursor, another AI-focused company, was recently valued at $10 billion after a massive funding round.
Meanwhile, developer job postings are at a five-year low, and on platforms like Blind — an anonymous forum popular with tech workers — anxiety is palpable. “Some engineers feel like they’re being replaced by something that doesn’t even work as well as they do,” one poster wrote. “That’s demoralizing.”
MIT researchers back up some of that frustration. In a new paper, they outlined major challenges AI still faces in software development — especially when it comes to writing large-scale or logically complex code.
“We’ve made a lot of progress,” said Armando Solar-Lezama, an MIT professor and senior author of the study. “But AI still struggles with nuance, context, and creativity — the things great engineers bring to the table every day.”
The perception that AI is to blame for job cuts may also be masking broader business pressures.
“Teams are getting leaner, yes,” said Heather Doshay, a partner at venture capital firm SignalFire. “But that’s more about companies watching their spending and less about AI replacing people wholesale — at least for now.”
Still, that’s cold comfort to workers on the ground. One senior engineer at a tax and audit firm described the current environment as “brutal.”
“You’re expected to know AI tools and be twice as good as before,” he said. “Only the top talent is getting hired. It’s intimidating — even for people already in.”
For Gareth, the pressure isn’t just professional — it’s personal. “I finally have a job that pays me well enough to live decently in this city,” he said. “But every week, there’s a new tool that can ‘code better than a human.’ It feels like we’re constantly trying to prove our worth.”
Despite the gloom, he’s not giving up. Like many others in tech, Gareth is adapting, learning AI tools, and hoping to stay ahead of the curve.
“I didn’t fight this hard just to be replaced by a machine,” he said. “If AI is coming for my job, I plan to be the one helping build it — not watching from the sidelines.
