The ILO AI Exposure Index (Gmyrek et al., 2025) estimates the technical potential for Generative AI to perform the tasks that make up existing occupations. It is designed to assess occupational exposure to current Generative AI capabilities and should not be interpreted as a prediction of job loss or employment outcomes.
The methodology begins by evaluating individual work tasks. Each task receives a potential automation score between 0 and 1, where 0 indicates that current Generative AI cannot perform the task and 1 indicates that it can be performed entirely using current Generative AI capabilities.
Occupation-level exposure is then calculated by combining the task-level scores using two complementary indicators:
Together, these indicators are used to classify occupations into six exposure categories that reflect both the overall level and consistency of Generative AI exposure across their task profiles.
Interpreting the results. The exposure gradients indicate the technical potential for Generative AI to automate or support tasks within occupations. They should not be interpreted as measures of employment risk or estimates of future job losses.
The findings suggest that very few occupations consist entirely of automatable tasks. Instead, most occupations combine tasks with different levels of exposure, implying that Generative AI is more likely to transform how work is performed than to replace occupations in their entirety.
Poland level. The Polish national classification of occupations (KZiS) lists 2,541 detailed occupations, each mapped to an ISCO-08 group. Its tasks are scored with the same method and the same category rules as the ISCO-08 index.
Source: Gmyrek, Berg, Kamiński, Konopczyński, Ładna, Nafradi, Rosłaniec and Troszyński (2025), Generative AI and Jobs: A Refined Global Index of Occupational Exposure, ILO Working Paper 140. Data: github.com/pgmyrek/2025_GenAI_scores_ISCO08 and github.com/pgmyrek/POLAND_2025_GenAI_scores_6digit_occupations. See also the ILO-NASK index for Poland. Prototype for review.