Two numbers landed this month that are supposed to tell the same story and do not. Close to 90,000 job cuts this year have been blamed on AI. At the same time, the companies spending the most on AI are growing their headcount, not shrinking it. Both are true. That is the whole problem.
The AI jobs debate has become a place where people arrive with a conclusion and collect the number that fits it. Let me try the opposite.
The Data Refuses to Pick a Side
Start with the case for fear. Layoffs attributed to AI are real and rising. One projection has up to 15 percent of US jobs eliminated within five years. Separate analysis points to roughly 16,000 net jobs erased each month in parts of the economy. If you only read those figures, the future looks like a smaller workforce arriving fast.
Now the other column. Companies that spend heavily on AI, on the order of 30 dollars per employee per month, grew headcount 10.2 percent. Entry-level roles, the ones everyone said would vanish first, rose 12 percent at the most AI-intensive tech firms. Where AI is used most aggressively, more people are being hired, including juniors.
These are not contradictory by accident. They are measuring two different things. The cuts are concentrated where AI is an excuse or a cost story. The growth is concentrated where AI is a production tool that makes the whole operation worth expanding. Same technology, opposite outcomes, decided entirely by how the company uses it.
Why Heavy Adopters Hire
This is the part the doom framing misses. When AI genuinely lowers the cost of producing something, software, documentation, analysis, the return on that work goes up. Higher return justifies doing more of it, which justifies more people, not fewer. The tool does not replace the team. It makes the team's output worth enough to grow the team.
I have seen this directly. The places using AI well are not running skeleton crews. They are taking on work that did not pencil out before, because the cost to deliver it dropped. The junior who was supposedly obsolete is now productive on day one instead of month three, which makes hiring juniors a better deal, not a worse one.
There is a measurement trap underneath all of this. A cut is easy to count and easy to attribute, so it gets a headline. A job that exists only because AI made the work profitable enough to hire for is invisible, because it shows up as ordinary growth and nobody issues a press release blaming a new hire on AI. The displacement gets a number. The creation gets none. That asymmetry is worth remembering every time a layoff cites the technology, because you are reading the half of the ledger that is easy to see.
The companies cutting and blaming AI are usually telling on themselves. The cut was coming anyway. AI is the line on the press release that sounds like strategy instead of a bad year. I have written before about why AI agents will not replace whole jobs, and the labor data is starting to agree. Tasks move. Whole roles mostly do not, at least not yet.
The Divide That Actually Matters
If there is a real threat hiding in these numbers, it is not human versus machine. It is one company versus another.
The firms with capital, technical depth, and the management bandwidth to turn AI pilots into actual gains are pulling ahead. The firms that bought a few subscriptions and called it a strategy are falling behind, and they are the ones most likely to frame their struggle as AI taking jobs. The technology did not do that to them. The gap in how they used it did.
I keep saying that AI is the excuse, not the reason, and the jobs debate is the clearest case. When a healthy, well-run company adopts AI, it tends to grow. When a struggling one does, it tends to cut and point at the robot. The technology is the same. The operational reality is not.
So the honest answer to whether AI is cutting your team is that it depends on you. If you treat it as a reason to do less with fewer people, it will deliver exactly that, and you will call it inevitable. If you treat it as a way to make your people's work worth more, it tends to grow what you can take on.
The machines are not deciding this. The operators are. The data is messy because the operators are making opposite choices, and the technology is loyal to whichever choice you feed it. Before you blame the tool for what happens to your team, check which of those two companies you are running.