The jobs question, and what the data actually says

The aggregate. Anthropic economists Maxim Massenkoff and Peter McCrory found no detectable rise in aggregate unemployment for AI-exposed workers since late 2022. An ECB survey of around 5,000 firms, published in March 2026 by Laura Lebastard and David Sondermann, found no overall employment gap between firms using AI and firms that do not.

The entry level. This is where it shows. The job-finding rate for young workers entering high-exposure occupations has fallen by more than three percentage points since its peak in November 2023. Junior analyst, support agent, copywriter, data entry: the roles going first are the ones people used to start in. And AI-exposed entry-level jobs are now around seven times more likely than other entry-level jobs to demand senior-level skills like judgement and leadership, which is a polite way of saying the first rung has been sawn off.

The announcements. Challenger, Gray & Christmas counted 116,175 job cuts attributed to AI in the first eight months of 2026, about 22 percent of all announced cuts. Treat that number carefully. It counts what companies say, and saying AI is doing the cutting is a better story for a share price than saying demand fell.

The other direction. In 2024, the AI sector directly generated roughly 119,900 jobs in the United States, against roughly 12,700 tracked as lost to AI automation. That ratio will not hold forever and it is also not nothing.

This is the part of the AI question that touches the most people in any room, and it is the part where the loudest claims on both sides are least supported. The mass-unemployment story is not in the data yet. Neither is the nothing-is-happening story.

The distributional point is the one worth holding: an economy can keep total employment flat while making it much harder to enter. That is not a smaller problem, it is a differently shaped one, and it lands on exactly the people who are least able to wait it out.

Note also who is doing the measuring. Two of the most-cited reassuring findings come from economists employed by an AI lab. That does not make them wrong, and it is worth saying out loud.

There is no Flemish or Belgian equivalent of these studies linked here yet. The High Council for Employment report is the obvious place to look and it has not been read into this dossier.

Every figure here comes from summaries of the studies rather than the studies themselves. Read the originals before quoting a number on stage.