The layer graduates are hired into is the layer a model absorbs first.
Across live entry-level specifications in the ten markets clearing the floor, as at 14 September 2026.
The layer graduates are hired into is the layer a model absorbs first, and the market is not yet hiring graduates into the layers that remain.
How many entry-level roles have AI at the centre
United States 46 of 891 (5.2%). Australia 8 of 608 (1.3%). United Kingdom 18 of 120 (15.0%). Every other market's cell is under 50 and prints as counts. Director and above in the same markets: United States 307 of 1,177 (26.1%), Australia 34 of 169 (20.1%), United Kingdom 74 of 297 (24.9%).
Every other market's cell is under 50 and prints as counts.
Where the apprenticeship used to be
The entry-level role has historically been execution-heavy by design: the layer where judgement is formed by doing the work under supervision. The market pages show execution load highest in the markets writing AI into fewest roles, and lowest where AI is written in most. The pathway page states the two facts together and stops: the layer graduates are hired into is the layer a model absorbs first, and the market is not yet hiring graduates into the layers that remain.
The ten marketsExecution load runs from 42.9% of the work in Singapore to 49.7% in Spain.
How this is measured
Live basis. Entry level is the seniority classification applied at extraction. Counts print beside shares for every cell under 150. This page serves two audiences from one finding: an early-careers lead reads a hiring problem, a university reads a formation problem, and both arrive at the same table.
Full method at /method