Where do apprentices come from now?

Every expert you have ever met started out doing the boring bits.

The junior solicitor read ten thousand documents before they argued a case. The graduate accountant ticked and bashed spreadsheets before anyone let them near a client. The trainee programmer fixed small bugs in other people’s code for a year before they designed anything. The boring bits weren’t just cheap labour. They were how people learned what “right” looks like.

Here’s the uncomfortable thing: the boring bits are exactly what AI is best at.

The first rung is the exposed rung

You can see it in the task lists on this site. Look inside almost any professional job and the most exposed tasks tend to be the junior ones: drafting, summarising, checking, first-pass research, the paperwork around the real work.

Legal associate professionals, the paralegals and legal clerks who do a lot of the drafting and document work, score 61. Lawyers score 64. Accounting and bookkeeping clerks score 36, while accountants score 49. The pattern repeats across the site: the job that feeds a profession its new people is usually more exposed than the profession it feeds.

That’s not a coincidence. Junior work is junior because it’s well-defined, checkable and repetitive enough to learn from. Those are exactly the properties that make a task easy to hand to a machine.

Is it showing up yet?

Early signs say yes, a bit.

Researchers at Stanford’s Digital Economy Lab have been tracking US payroll data from ADP since late 2022. In their August 2026 update, workers aged 22 to 25 in the most AI-exposed occupations had employment 19% below similarly aged workers in less exposed jobs. Experienced workers in the same occupations showed no comparable gap. And it seems to be happening through hiring: fewer young people being taken on, rather than more being let go.

The authors call these young workers “canaries in the coal mine”, and it’s a good name. It’s early, it’s one country, and plenty else is going on in the economy. But it is exactly the pattern you’d expect if firms are asking “why hire a junior to do this when the tool does it?”

The problem nobody has priced in

Here’s the knot. A firm can save money today by not hiring juniors. But every senior it will need in 2040 is a junior it should be hiring now.

Expertise isn’t downloaded. It’s built, slowly, by doing lots of small things wrong and being corrected. If the small things are all done by a machine, where does the judgement come from? You can’t skip straight to the top of a ladder whose bottom rungs have been sawn off.

This isn’t a new worry. Trades have always had to protect the apprenticeship, because a master electrician is no use to anyone if nobody trained the next one. What’s new is that office professions, which never had to think about it much, suddenly do.

What might fix it

Nobody has the answer yet, but there are some sensible guesses.

• Redesign junior work rather than delete it. Juniors who check, question and correct the machine’s output may learn faster than juniors who did it all by hand. The work changes from “produce it” to “catch what’s wrong with it”, which is arguably closer to what seniors actually do.
• Pay for training on purpose. When learning was a free side effect of cheap junior labour, nobody had to budget for it. Now somebody does. Apprenticeship-style schemes, with time set aside to learn, start to make sense outside the trades.
• Use AI as the tutor. One of the most striking findings about workplace AI so far is that it helps novices far more than experts. That’s a problem for juniors’ job prospects and a gift for their learning, at the same time.

If you’re just starting out

Honestly, it’s harder than it was, and the data backs you up on that. A few things seem to help:

1. Aim for the tasks next to the exposed ones. On your job’s page, look at which tasks the machine likes, then get good at the ones beside them: talking to clients, deciding what matters, knowing when the output is wrong.
2. Learn the tool better than the people hiring you. “I can do the junior work in a tenth of the time and check it properly” is a strong pitch.
3. Look at the hands-on side of your field. Many professions have a practical wing (site work, patient contact, field work) where the junior tasks are still firmly human.

And if you run a team: please keep hiring juniors. Someone trained you.