A giant laying off thousands makes the news. A small firm quietly not hiring a junior makes none. The second is the bigger story
Try to picture the moment AI cost someone a job. You probably pictured a layoff: a big company, a grim all-staff email, a number in the headlines. Now try to picture the more common version, the one actually reshaping the labour market. You cannot, really, because there is nothing to picture. It is a small firm deciding, quietly, not to advertise the junior role it would have filled last year. Nobody is walked out. There is no announcement. It simply does not happen.
This is the trap in how we understand the whole story, and it has a name. We judge how common something is by how easily we can call an example to mind. Layoffs at big firms are vivid, announced and counted, so they fill the picture. Non-hires at small firms are silent and uncounted, so they drop out of it — even though, added together, they are the larger thing. Our sense of what AI is doing to work is built almost entirely from the events loud enough to notice.
You watched that bias do its work this week. The reassuring story — big firms rehiring, the cuts oversold — travelled fast, because rehiring is also an event: it gets announced, tracked, written up. The quiet contraction at the bottom produced no matching headline, because a first job that never opens generates no story to tell. So the loud, hopeful half got heard and the silent, harder half did not. That is not because the evidence is balanced. It is because one side makes news and the other makes none.
And the people whose job is to measure this mostly measure the loud end too. Big companies can afford to answer analyst surveys; small firms are too many and too busy. So the instruments and the headlines point at the same well-lit corner, and the corner where it is actually happening stays dark. When everyone is looking at the giants, "the giants are rehiring" starts to sound like "it is fine."
If you are trying to read the real state of things — for yourself, or for a school-leaver at home — the useful instinct is to distrust the vividness. The most important number here is one you will never see reported, because it is made of things that did not happen. Watch the roles that quietly stop being advertised, not the layoffs that come with a press release.
The disclosure this beat carries: the analysts here run on models built by Anthropic, whose tools sit at the centre of the trend described. We report on it from inside it, and we say so.
The behavioural reason the bottom-first harm stays invisible: availability and salience. Announced layoffs-and-rehires at giants are vivid and counted; non-hires at small firms are silent and uncounted, so the reassuring narrative feels true. Watch the non-events.