TL;DR
Goldman Sachs research finds AI already weighing on hiring across developed economies, though the effect is concentrated in a narrow band of industries and workers. Call centre employment has fallen furthest below its historical trend. Across more than 800 occupations, the sharpest drag lands on people trying to start a career.
Where the effect is visible
Industries more exposed to automation have posted slower growth in job openings since mid-2022, a relationship Goldman found strongest in Australia, the United States and Germany. Information and communication services, which ranks among the most exposed, has slowed almost everywhere; outside the US it nonetheless still tracks at or above its longer-run path.
The clearest damage is narrower and easier to name. Advertising, management consulting, software publishing and call centre work have all sunk well beneath their historical baseline throughout the developed world. Call centres are the extreme case: 39% under trend in the US, 33% in Canada, 27% in Germany. These are the roles where usable automation actually exists today, which is rather the point.
The entry-level squeeze
The occupational analysis is where the finding gets uncomfortable. Taking a 10% increase in exposure to AI, the effect on annual headcount growth across the whole workforce is around 0.1 percentage points in Canada, France and the US — statistically real, practically small.
For workers at the start of their careers the same exposure produces a drag ranging from more than 0.2 points in the US to over 0.6 in Australia. A secondary, smaller effect shows up in occupations judged at high displacement risk. Goldman’s overall read is deliberately measured: the pressure is plainly visible in the data, and it remains confined to particular industries and particular workers.
Looking forward
On adoption, Goldman pooled 11 surveys and put developed economies at roughly 15% to 20%, with Britain among the leaders alongside the Netherlands, France and the US; New Zealand, Japan and Italy trail. Emerging markets sit around 10% to 15%.
The UK-specific tension is worth stating plainly. Lloyds research yesterday found more than half of British firms saying AI had created jobs — a genuine finding, but one that measures net positions inside existing employers rather than the door into work. Goldman is measuring the door. Britain is also answering youth unemployment with AI boot camps, which assumes entry-level roles still exist to be trained into. This research suggests that assumption deserves testing before more money follows it.