Work & Economy
By Roa Editorial | June 2026 | 9 min read
A few months ago, AI job losses stopped being a headline I scrolled past and became something personal. Two friends of mine, both in tech, one in software and one in IT support, lost their jobs within three weeks of each other. Neither company used the word AI anywhere in the layoff announcement. But in our group chat, nobody needed to say it out loud. We all knew.
Around the same time, I noticed something about my own habits. I genuinely cannot remember the last time I looked something up the old way. Fixing an error message, planning a trip, even drafting an awkward email to a landlord: it all goes through AI first now. There is almost nowhere left, work, errands, even casual curiosity, where AI is not somehow in the picture.
So when people ask me whether AI job losses are real or just a scary headline, I do not have a tidy answer. The truth sits somewhere uncomfortable in the middle. I spent a few weeks going through the actual research instead of the summaries, and what I found was more interesting and more nuanced than either the doom posts or the optimistic press releases suggested.
Here is what I came away with. I will give you the short version first.
The short version: AI job losses in 2026 are concentrated, not universal. They are hitting entry level workers in tech, customer support, and other AI-exposed roles the hardest, while overall employment has not collapsed. At the same time, most of the companies pouring money into AI are not actually seeing a financial return from it yet. Both of those things are true simultaneously, and that gap between hype and reality is exactly what most coverage skips over.
Why AI Job Losses Are Concentrated, Not Universal
AI is good at a specific kind of work. Tasks that are structured, repetitive, and easy to check are exactly where today’s tools shine. Writing boilerplate code, answering common support questions, drafting a first version of a marketing email: this is the work AI does well, fast, and cheaply.
That happens to be, not coincidentally, the kind of work that junior employees usually handle. The first two or three years of a career in software development, customer service, or marketing are often spent on exactly the structured, repeatable tasks AI now does instead. So when a company adopts AI tools, the jobs that shrink first tend to be the entry level ones, not the senior ones.
I keep thinking about my two friends who got laid off. One was three years into his first software role. The other was in a support function that had been gradually automated for two years before the final headcount reduction. Neither of them was doing bad work. They were doing exactly the work their companies had started automating.
The Numbers Behind AI Job Losses in North America
Stanford’s 2026 AI Index Report, one of the most thorough annual studies on this question, found that employment for software developers ages 22 to 25 has fallen nearly 20 percent since 2024, even as older developers at the same companies kept their jobs or grew in number. The same report notes that roughly a third of employers expect further workforce reductions over the coming year.
If you graduated with a computer science degree in the last two years and you are sending out dozens of applications with little to show for it, this is not just bad luck. The data backs up what you are feeling.
Stanford’s report also measured productivity gains from AI in specific functions: 14 to 15 percent in customer support, roughly 26 percent in software development, and an estimated 73 percent faster output in marketing content. Those numbers explain why companies are excited about AI even while the entry level job market tightens around them.
| Function | Reported AI productivity gain |
|---|---|
| Customer support | 14% to 15% |
| Software development | ~26% |
| Marketing output | ~73% |
Source: Stanford HAI, 2026 AI Index Report
On the adoption side, McKinsey’s 2025 Global AI Survey found that 88 percent of companies now use AI in at least one business function, up from 78 percent the year before. As for what that means for headcount, McKinsey found opinions split: 32 percent of respondents expect their overall workforce to shrink in the year ahead, 43 percent expect little change, and only 13 percent expect it to grow.
Nobody fully agrees on where this is headed, including the people running the companies doing the adopting.
The Part the Hype Machine Leaves Out
Here is where I think most coverage of AI job losses gets it wrong. It treats AI adoption as if it automatically translates into massive profits for the companies making these cuts. The actual numbers tell a messier story, and this is the part that surprised me most once I started looking properly.
Despite that 88 percent adoption rate, McKinsey found that only about 6 percent of companies, those they categorize as AI high performers, report a meaningful bottom-line financial impact from their AI investments. Most companies are using AI. Most are not seeing it show up in their profits in any significant way yet.
It gets more striking. MIT’s Project NANDA studied around 300 enterprise AI deployments and found that 95 percent of generative AI pilots delivered no measurable impact on profit and loss at all. Only about 5 percent produced real, fast revenue gains. Meanwhile, Gartner projects that global AI spending will reach roughly 2.5 trillion dollars in 2026, a 44 percent jump from the year before.
So companies are spending an enormous amount of money, cutting entry level jobs along the way, and most of them still cannot prove it is paying off. That is not a conspiracy theory. That is what McKinsey, MIT, and Gartner each found independently of one another. I personally find that combination more unsettling than the idea of AI cleanly and efficiently replacing jobs. At least that version would make some kind of logical sense.
The MIT research also revealed a misalignment in where companies put their AI budgets. More than half of generative AI spending goes into sales and marketing tools, yet MIT found the biggest actual returns in back office automation: eliminating outsourced business processes, cutting external agency costs, streamlining repetitive admin work. The flashy customer facing projects are where the money goes. The boring back office is where it pays off.
What This Actually Means, Depending on Who You Are
If you are early in your tech career
The honest advice is to stop competing on the tasks AI already does well. Writing routine code or answering basic support tickets is not where your value comes from anymore. Building real depth in system design, AI tooling, or a specific industry niche puts you in the smaller group of roles that are actually growing instead of shrinking. I have watched friends who pivoted toward this approach find their footing much faster than those who kept applying for the same roles that were quietly disappearing.
If you work in customer service, admin, or a structured role
It is worth asking your employer directly how they plan to use AI in your function over the next year. You are not being paranoid by asking. Stanford’s own data shows employers expect more change ahead than what has already happened. Getting ahead of that conversation beats being surprised by it. I know this is easier said than done, but the alternative is finding out through a calendar invite to an all-hands meeting you did not expect.
If you work in marketing or content
The productivity numbers above explain why your team’s workload is changing even if nobody has said the word automation out loud. That does not mean your job disappears. It likely means less time on first drafts and more time on the judgment calls AI still cannot make well: strategy, tone, audience, and context. The people I see thriving in this space are not fighting AI. They are deciding what to do with what it gives them.
If you run a small business in North America
The failure data is actually useful here. MIT’s research found the most reliable returns came from back office automation: cutting outside agency costs, streamlining repetitive admin work, eliminating outsourced tasks that no longer need to be outsourced. Most companies pour their AI budgets into sales and marketing, which is exactly where MIT found the weakest returns. If you want better odds than the 95 percent of companies that saw nothing, start in the boring, unglamorous corners of your business. That is where I would start if I were building something right now.
Practical Tips Worth Actually Acting On
- Treat AI fluency as a basic professional skill now, the same way spreadsheets or email became standard practice years ago. You do not need to become an engineer, but being comfortable using these tools well is no longer optional in most fields. I see this gap in hiring conversations more often than I expected.
- Pay close attention to which parts of your job are easy to describe in a short, structured prompt. Those are the tasks most exposed to AI job losses going forward, regardless of your title or industry. If you can write out what you do in three sentences, someone has probably already tried to automate it.
- Do not assume your company’s AI strategy is working just because they are spending money on it. The data says most companies have not figured this out yet either. That uncertainty cuts both ways. It is not all doom, but it is not the clean efficiency story in the press releases either.
- If you are rebuilding a resume right now, lead with outcomes and judgment, not just tool use. Plenty of applicants can say they used AI. Far fewer can show what they decided to do with what it gave them. That gap is where the actual value lives, and it is the part that is hardest to automate.
Where I Land on All of This
I think about my friends who lost their jobs often, and I do not think it is fair to tell them this was somehow avoidable if they had just picked a different path. AI job losses right now are real, structural, and concentrated in ways that have very little to do with individual effort or ability.
But I also do not think panic is the right response. The data shows a job market that is shifting, not collapsing, and a technology that companies themselves are still struggling to make pay off. There is something almost grimly reassuring about the MIT finding that 95 percent of enterprise AI pilots produced nothing measurable. It suggests the transformation is slower and messier than the loudest voices on both sides are willing to admit.
If there is one thing I would want someone reading this to walk away with, it is that the loudest stories about AI, whether utopian or apocalyptic, tend to skip the messy middle where the real data actually lives. That middle is where the useful decisions get made. It is also where most people I know are actually spending their working lives right now, trying to figure it out one week at a time.
Leave a Reply