AI's Clearest Risk Is a Narrowing Entrance, Not a Vanishing Occupation

The short answer: entry-level roles in many professional fields exist partly to absorb routine preparation work, and that work is almost exactly what the exposure ratings rate highest. If it gets substantially cheaper, the economics of hiring a large junior class change while total employment holds up. Aggregate projections and exposure scores are both blind to this. The one place it is visible is accounting, where BLS happens to draw a classification line through the middle of the field.

The mechanism

Senior people in professional fields have priced routine preparation out of their day: the reconciliation, the document assembly, the first-pass research, the formatting. That work flows downward to whoever is newest. It is how the junior role is defined in practice, and it is how new entrants learn the field, by doing the preparation until they are trusted with the judgement.

That preparation work is also, almost exactly, what the exposure ratings rate highest. Structured inputs, a rule set applied, a document out. If a model can produce the first pass at a fraction of the cost, the case for hiring six juniors to produce first passes weakens, even while the occupation as a whole is growing.

Why the usual numbers cannot show it

An aggregate projection will not show you this, because BLS reports employment for an occupation rather than the shape of its hiring pyramid. An occupation can be projected up 5% with a junior class half its former size, and the headline number will look fine.

An exposure score will not show it either. The score describes a week, not a career ladder. It can tell a junior analyst that most of their current hours are exposed. It cannot tell them that the rung they are standing on is being removed from the ladder, because the ladder is not in the data.

The one place the trace is visible

Accounting is the exception, because the two ends of the same field are classified separately. BLS projects accountants and auditors up about 5% from 2024 to 2034, and bookkeeping, accounting, and auditing clerks down about 6% across the same decade, citing software automation for the decline.

That is the shape: the accountable, licensed end holds and the processing end shrinks. It is legible only because someone drew a classification line through the middle of the field. In law, marketing, finance, consulting, and most other professional occupations, nobody has. The junior analyst and the partner share a code, and the projection averages them. I go into the accounting split in more detail in accountants up, clerks down.

What I am and am not claiming

I am not claiming this is happening at scale. The evidence that entry-level hiring is contracting because of AI is suggestive but early, and it is tangled up with interest rates, post-pandemic over-hiring, and ordinary cycles. What I am claiming is narrower: this is the version of the story the available numbers are least equipped to detect, and the mechanism that would produce it is already priced into the exposure ratings.

If you are early in a career

This is a more useful question than your own score. Your score will be high, because junior weeks are full of preparation work. That is expected and it is not, by itself, a problem. The question is what the junior roles in your field spent their time on five years ago, and whether that work is still the path into the field or has become the work the field is trying to stop paying for.

If the preparation work is going away, the path in has to run through something else: the client contact, the site work, the review step, the parts of the job with the three floor properties. Getting hours in those earlier than the ladder used to allow is the practical response, and it is easier to ask for while you are new than to reconstruct later.

What the numbers cannot yet see

When better data arrives, it will most likely come from hiring pyramids rather than employment totals: the ratio of junior to senior postings, the size of graduate intakes, the age distribution within an occupation. Until then, the accounting split is the clearest trace we have, and it points in a direction worth taking seriously.

Frequently asked questions

Is AI eliminating entry-level jobs?

The evidence is early. The mechanism is clear: junior roles absorb routine preparation work, which is the most exposed category in the ratings. Whether that has already contracted junior hiring at scale is not yet settled in the data.

Why do BLS projections not show a narrowing entrance?

BLS reports employment totals for an occupation, not the shape of its hiring pyramid. An occupation can grow overall while its junior class shrinks, and the total will not reveal it.

Where is the pattern visible?

Accounting, because BLS classifies accountants and auditors separately from bookkeeping, accounting, and auditing clerks. The first is projected up about 5% and the second down about 6% over 2024 to 2034.