The bottom rung is thinning

A snapshot published this past week by the government's AI and the Future of Work unit, with LinkedIn, put UK hiring at 14 per cent below a year earlier as of April. Every tracked industry was down. Of 38 entry-level occupations, 30 were shrinking. The fastest falls were in exactly the roles you would expect if you believed the simplest story: trainee accountants, graphic designers, junior software engineers — the desk jobs that mostly move information around, the kind a capable model now does a passable version of in seconds.

That is the headline, and it is frightening if you are 22, or if you are 43 and watching your own employer quietly stop taking on graduates. The official numbers underneath it are real: ONS data show vacancies down 7.1 per cent on the year in February to April, with about 2.5 unemployed people now chasing each opening, up from 1.9 a year earlier.

Here is the complication, and I think it matters more than the headline. The same government snapshot is careful to say what it cannot say. Entry-level hiring is falling broadly in line with the wider market, not racing ahead of it. The hiring rate turned negative back in late 2022, before the current wave of AI tools existed. And while some of the steepest declines are in the roles where machines have improved most, the report states plainly that this is not yet causal evidence. Correlation is not a robot in the chair. A weak economy thins graduate hiring on its own; it always has.

So which is it — the machines, or the slump? I find the honest position is that it is both, tangled, and that we cannot yet cleanly pull them apart. That is an uncomfortable answer, because it offers no one to blame and nothing simple to fix. But pretending we can already prove AI is eating the bottom rung does the reader a disservice in the other direction. We would be manufacturing a certainty we don't have.

What I keep coming back to is the human shape of it. The first rung of a career was never only about the work. It was where you learned the unwritten things — how an office actually runs, how to be wrong in front of people and recover, who to ask. If that rung thins, whether the cause is software or sluggish growth, a generation loses the place where competence is built, not just the place where the money starts. You cannot intern under an algorithm.

The government's own assessment, published in January, reckons AI could lift UK productivity growth by up to 1.2 percentage points a year over the next decade. That is a real prize. The unanswered question — the one the snapshot is honest enough to leave open — is whether the gains arrive as better work for people or simply as fewer of them. We don't yet know. Watch the entry-level numbers, not because they prove the case, but because they are where the answer shows first.