The Jobs Apocalypse That Keeps Not Arriving

Ask most people what AI is doing to work and you will hear some version of the same answer: it is coming for the jobs, quietly and fast. The claim is intuitive, it gets repeated constantly, and it has launched a thousand anxious headlines. It is also, so far, not what the data shows.

Let me be clear about what is and is not settled. Nobody can prove what employment will look like in ten years, and I am not going to pretend otherwise. But the fear usually on offer is not a careful forecast about 2036. It is a claim about right now, that AI adoption is already shrinking payrolls. That claim we can actually check, and the early evidence cuts against it.

What the numbers say

A recent study by Kharazian and colleagues tracked companies through their first two years of AI adoption. The firms that adopted the technology intensively did not shed workers. They added them, with total employment up 10.2 percent against no meaningful change at the firms that stayed on the sidelines. The growth showed up across the board: sales, administration, engineering and customer service roles all expanded.

The most striking finding concerns the workers everyone assumed would be first out the door. Entry-level headcount rose 12.0 percent after adoption, faster than the growth for experienced staff or managers. If AI were hollowing out the bottom rung of the career ladder, this is not what it would look like.

Zoom out and the picture holds. The Information Technology and Innovation Foundation estimates that AI was tied to roughly 119,900 new jobs in 2024 against about 12,700 layoffs, the latter amounting to a tenth of a percent of all layoffs that year. That is not zero, and it is cold comfort if you are one of the 12,700. But it is nowhere near an extinction event.

Why the fear persists anyway

Part of it is that the grim story writes itself. A press release about a company swapping a call centre for software is vivid and quotable, while a diffuse gain of a few thousand jobs spread across an industry is neither. Part of it is that some displacement is genuinely happening. Roles built on repetitive, predictable tasks, like data entry and first-line support, really are being automated, and the people in them are not much consoled by growth somewhere else on the org chart.

The likeliest near-term pattern is not mass unemployment but reshuffling. Some tasks vanish, new roles appear, and the work rearranges itself around the tool. That has been the shape of every previous wave of automation, and the current evidence suggests AI is following that path rather than breaking it. Even the institutions bracing for trouble, like the Federal Reserve with its new task force on AI and jobs, are studying a transition, not certifying a collapse.

History is not a guarantee, and honest people can worry about where a much more capable model lands five years out. What they should not do is claim the collapse has already happened. The receipts, for now, say the opposite.