UK AI Job-Risk Index · Tool
Your personal AI task score
Two people with the same job title can do very different work. Pick your occupation, choose the closest matching US O*NET occupation, then tick the tasks you actually do. You'll get an indicative score for how exposed your mix of tasks is to AI language tools.
Method and sources
- UK SOC 2010 to ISCO-08: ONS (2022), Truncated proportional conversion between ISCO-08 and UK SOC classifications (research conversion, not official guidance).
- ISCO-08 to US SOC 2010: US Bureau of Labor Statistics ISCO-08/SOC crosswalk (as packaged by the Institute for Structural Research, Hardy, Keister and Lewandowski 2018), then US SOC 2010 to O*NET-SOC 2019 via the O*NET Resource Center crosswalk. We show up to five US occupations whose crosswalk weight is at least a quarter of the strongest match.
- Task statements: O*NET, US Department of Labor/ETA, used under CC BY 4.0. O*NET® is a trademark of USDOL/ETA. We have not modified the task text.
- Task exposure ratings: human annotator ratings published by Eloundou, Manning, Mishkin and Rock (2023), "GPTs are GPTs: An early look at the labor market impact potential of large language models" (data released under the MIT licence). Each task is rated E0 (no exposure), E1 (a language model alone could cut the time by at least half) or E2 (it could with extra software built on top).
- Score = weighted average of the ticked tasks, counting E1 as 1, E2 as 0.5 and E0 as 0 (the paper's "β" measure), with core tasks weighted 1 and supplemental tasks 0.5, shown out of 100.
This is indicative only. The ratings are about US tasks and language-model-based tools as of 2023, the crosswalks are approximate, and exposure is not a forecast that a job will be cut. The headline Job-Risk Index uses a different measure (DfE 2023 AIOE), so the two numbers are not directly comparable.