Twenty-eight per cent of scientific work is reachable with a chat window today, and only 30 per cent cannot be sped up. Both figures are near the top of the report.
Exposure means published research judged an AI could halve the time a task takes, at the same quality. It is not a prediction about any job. How to read this
Not one task. At the other end of the same group, Agricultural Technicians are 81 per cent protected, because the work involves being in a field with equipment.
Eighty-one points of spread inside a single label. The pattern is now familiar: what matters is the medium, not the subject.
Survey Researchers sit at the top of the whole economy for exposure. Reading literature, designing instruments, analysing responses and writing them up is text work from end to end.
This matters beyond the group, because a great deal of what other professions call "research" is the same activity under a different name.
Solid is reachable with a chat window today, pale needs software building first, grey cannot be sped up. Task counts are shown because several of these rest on fewer than twenty.
Designing the instrument, analysing free text, writing up the findings. That work sits at the most exposed end of the entire report, done by the most qualified people in it.
If the most exposed occupation in the economy is somebody with a doctorate designing surveys, the idea that exposure tracks seniority does not survive.
There is no data here on what anybody actually does with AI. A dataset that appears to measure it was removed from this report because its meaning could not be stated. The reasoning is on the limits page.
The exposure ratings cover all 1,236 tasks in these 60 occupations with no gaps, and were made in 2023 against the model capabilities of that year. Read every figure as a floor.
Exposure ratings from Eloundou, Manning, Mishkin and Rock, GPTs are GPTs: Labor market impact potential of LLMs, Science 384, 1306–1308, 2024, used under the MIT licence. This page includes information from the O*NET Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. People Team AI has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications. Employment figures from the US Bureau of Labor Statistics.