AI isn't just replacing jobs, it's rewriting the job description
On Monday, General Motors laid off somewhere between 500 and 600 IT workers. More than 10% of the IT department. People with years of tenure, people who had spent months learning AI tools to stay relevant, people who got a 15-minute virtual meeting with HR and a scripted message and then an abrupt goodbye.
The same day, GM had roughly 80 open IT positions on its careers page.
Not for the same roles. For AI-native developers. Data engineers. Prompt engineers. Agent and model builders. TechCrunch called it a “skills swap.” That framing is almost too tidy. But it’s accurate.
I want to talk about what this means. Not because I think it’s good, but because I think pretending the change isn’t coming is more dangerous than acknowledging how uncomfortable it is.
The pattern is real
GM isn’t a one-off. CNBC counted over 54,000 US layoffs explicitly attributed to AI in 2025, from companies including Amazon (14,000 corporate jobs), Microsoft (15,000+), Salesforce (thousands of customer support roles), and TCS (12,000+). In March 2026, AI was the single leading cause of announced US layoffs, accounting for roughly 25% of that month’s cuts.
Here’s the thing, though: most of these companies are also hiring. For different roles. That distinction matters.
This isn’t a story about AI eliminating tech work. It’s a story about AI changing what tech work looks like - and companies making brutal, sometimes premature, sometimes wrong decisions about how to get ahead of that change. A Harvard Business Review survey from January 2026 found that most executive-cited AI layoffs are anticipatory - based on expected AI productivity, not proven AI ROI. Companies are betting on a future that hasn’t fully materialized yet.
Some of those bets will be wrong. Klarna cut over 2,000 roles citing AI efficiency, then started hiring humans again when service quality fell apart. AI replacement at scale is not a one-way ratchet.
But the direction of travel is not in question.
Who’s actually getting hit
A Stanford study from August 2025 looked at ADP payroll data across millions of workers. What they found complicates the “AI replaces workers” narrative in an important way.
Workers aged 22–25 in AI-exposed roles like software engineering saw a 13% relative employment decline since late 2022. Software developer employment for that age cohort fell nearly 20% from its 2022 peak. Workers 30 and older in the same roles saw employment grow 6–12%.
The dividing line isn’t “tech worker vs. non-tech worker.” It’s “people who can wield AI effectively vs. people who can’t.” And right now, the people most at risk are early-career workers who haven’t had time to build the irreplaceable expertise that makes AI a collaborator rather than a substitute. The people who built that base? They’re doing fine. In some cases, better than fine.
PwC’s 2025 Global AI Jobs Barometer, based on nearly a billion job postings across six continents, found a 56% wage premium for workers with AI skills - up from 25% the year before. Skills in AI-exposed jobs are changing 66% faster than they were a year ago. The jobs are growing. The required skills are changing faster than a lot of people can keep up.
None of this is necessarily good
The environmental cost of AI infrastructure is staggering. Data centers are consuming water in drought-stricken regions and electricity at rates that are straining grids and accelerating emissions. We don’t have a good answer for this yet. The IEA projects that data centers could account for 4% of global electricity demand by 2026, up from 1–1.5% in 2022. That’s not a minor externality.
The labor disruption is real and it is not evenly distributed. The workers who got that 15-minute GM call on Monday - many of them will find other roles, eventually. Some won’t. The transition costs fall on individuals, not on the companies making the decisions. GM reported a 40% earnings beat the same quarter it cut those 600 people. The productivity gains are going somewhere, and it isn’t to the workers who got displaced.
And there are serious, unresolved questions about AI accuracy, about bias baked into training data, about what happens to creative and knowledge work when we offload it to systems we don’t fully understand. The Stack Overflow 2025 Developer Survey found that trust in AI accuracy among developers actually fell year over year - from 40% to 29%. Developers who use these tools every day are more skeptical of them than they were a year ago. That’s a signal worth taking seriously.
I’m not making the argument that tech workers should embrace AI because AI is good. I’m making the argument that the sea change is coming regardless of whether it’s good, and that understanding what’s happening - clearly, without either catastrophizing or cheerleading - is the only way to navigate it.
What I actually think you should do
The honest version of the “adapt or fall behind” message is less motivational poster and more risk management.
The World Economic Forum’s Future of Jobs Report 2025 projects 170 million new jobs created and 92 million displaced by 2030 - a net gain, with brutal distributional unevenness. The fastest-growing skills are AI and data literacy, yes, but also creative thinking, resilience, and analytical judgment. The skills AI doesn’t substitute for well are the ones that require context, stakes, and trust.
Get hands-on with AI tools in real work, not toy projects. Understand what they’re actually good at and where they fail. The DORA 2025 report is unambiguous that AI amplifies what’s already there - strong engineers get stronger, weak processes get faster chaos. The skill isn’t using AI. The skill is knowing when, and how, and whether.
And if you’re waiting to see if this is all really happening before you engage - I get it. It’s exhausting to keep up with a landscape that changes this fast. But if you’re actively avoiding these tools, you are making yourself easier to replace. Not because AI is better than you. Because someone who uses AI and has your experience is better than you. That person exists. They’re getting hired.
The ATM didn’t replace bank tellers. It changed what tellers did, and eventually bank branches multiplied and teller employment grew for decades before mobile banking actually did reshape the industry. The lesson there isn’t “don’t worry” - it’s that technology usually substitutes for tasks within jobs, not entire occupations, and the surviving job often looks meaningfully different.
I don’t know exactly how this transition ends. What I do know is this: the job description is being rewritten whether you participate or not. You can help write it, or you can find out later what it says. Those are genuinely different outcomes.