TL;DR

UK vacancies have nearly halved since their 2022 peak, and Bank of England research published today finds the decline is steepest precisely where AI can do the work. Occupations in the top third by AI exposure lost 15% of their online adverts. The middle third lost 10%, the bottom third 6%. Customer service roles fell 23% and administrative roles 22%.

The author, Haley Schlicht of the Bank’s Data and Statistics Division, is careful not to declare a culprit. The correlation between exposure and advert contraction is strong across 26 occupational groups, and the exposure measure combines five separate academic and industry indices reweighted to British employment patterns rather than borrowed from American data.

The competing explanation

Against the reading that AI is displacing these roles, the post sets a rival account: remote working, not AI, snapped the lower steps of the career ladder by making junior staff costlier to train. The two are genuinely hard to separate, because finance, professional services and ICT adopted both at once. Those three show the deepest vacancy reversals, and roles of that kind make up 84%, 81% and 63% of their respective workforces.

There is also a measurement trap worth noting. At industry level the correlation disappears entirely — professional services file accountants and building staff under one code — and only reappears when industries are broken into their occupational make-up. Sector-level dashboards will miss this.

Pandemic over-hiring, the subsequent correction, National Living Wage rises and employer National Insurance changes all sit in the same window.

Looking forward

The Bank suggests the traditional employment pyramid may be flattening into a diamond shape — thin at the junior end, weighted towards experienced staff, with AI absorbing the routine processing that once justified graduate intake. That squares with the two-speed UK jobs market reported here on 3 August. Three things will settle the argument: whether headcount follows adverts downward, whether entry-level hiring keeps weakening, and whether measured exposure turns into actual adoption.