The canaries in the AI mine
Entry-level jobs are disappearing. That is not just a problem for the young. It is a problem for everyone who depends on the pipeline they feed.
The promise of artificial intelligence was productivity. For experienced workers, that promise is being kept. For those just starting out, something different is happening. Their jobs are vanishing.
A Stanford analysis of ADP payroll data found a 13 per cent decline in early-career employment within occupations most exposed to AI. Workers aged 22 to 25 have seen double-digit relative drops in AI-affected tasks. Their older colleagues, meanwhile, are more likely to benefit from augmentation — AI as a tool, not a replacement.
The pattern is not hard to explain. Entry-level work is often routine, structured and rules-based. It is precisely the kind of work that large language models and automation tools handle well. Scheduling, data entry, first-pass research, basic copywriting, junior bookkeeping — these were never glamorous tasks, but they were how people learned. They were the bottom rung. (You can see how roles like data entry clerk and bookkeeper score in our AI risk rankings — both are among the highest-risk occupations.)
Venture capitalists have begun to say the quiet part aloud. Companies are citing AI as justification for cutting entry-level headcount. Not restructuring it. Cutting it.
A ladder with missing rungs
This creates a problem that extends well beyond the young workers it displaces. Entry-level roles are not just cheap labour. They are training grounds. The junior analyst who spends two years cleaning data learns what good data looks like. The graduate copywriter who drafts fifty headlines a week develops editorial judgement. Remove those roles and you do not just lose output. You lose the pipeline that produces senior talent.
Firms that eliminate the bottom of the ladder will, in five years, find they have no middle. The cost will not appear on this quarter’s balance sheet. It will appear later, as a shortage of competent mid-career professionals and a growing dependence on external hires who cost more and know less about the business.
The effect is not spread evenly. Places with large service economies and high concentrations of administrative work — think London, Washington DC, or any city with a dense financial sector — are more exposed than rural areas where the workforce skews toward trades, healthcare and agriculture.
What competent employers will do
The answer is not to preserve pointless work for sentimental reasons. If a task can be automated, it probably should be. But the training function that task once served still needs to exist somewhere.
Some firms are already building internal academies — structured programmes that expose junior staff to real work alongside AI tools, rather than in place of them. Others are redesigning roles so that early-career workers use AI to do more, not to be replaced by it. The distinction matters.
The firms that get this right will have a genuine advantage. They will have people who understand both the work and the tools. The firms that get it wrong will have a gap in their org chart and no obvious way to fill it.
This is not a future problem. It is already showing up in the data. The question is whether anyone will act on it before the damage becomes structural.
Check how exposed your own role is with the AI job risk quiz, or see how your area compares in the location rankings. We explain how we score each occupation using 10 research sources.