Which occupations are most exposed to AI? The full ILO ranking
Office and clerical work at the top of the list, physical work at the bottom. What the International Labour Organization's index really measures, and what it does not say about layoffs.
According to the ILO's 2025 index, the groups most exposed to generative AI are general clerks, information technology professionals, numerical and material recording clerks, and customer service clerks. The least exposed groups are manual labour, agriculture and cleaning. Exposure means that some tasks can be performed or assisted by AI, not that the job will disappear.
What does the index measure?
The International Labour Organization (ILO) broke down thousands of tasks across different occupations and examined how many of them a generative AI model can perform or speed up. Each occupation received a score between 0 and 1. We calculated a simple average for each of the 40 occupational groups in the international ISCO-08 classification, so that occupations can be found by their Hebrew name.
Who is at the top of the list?
In first place: general and keyboard clerks, with an average score of 0.63. They are followed by information and communications technology professionals (0.54), numerical and material recording clerks (0.53), customer services clerks (0.52) and other clerical support workers (0.51). Four of the five most exposed groups are clerical and office jobs: work with text, numbers and documents.
Who is at the bottom?
The least exposed groups are agricultural labourers (0.11), cleaners (0.12), building workers (0.13) and labourers in construction, manufacturing and transport (0.14). These are physical jobs performed in the real world. It is important to remember: the index examines only generative AI. Robots, autonomous vehicles and other types of automation are not part of it.
What is surprising in the picture?
Previous waves of automation mainly affected routine manual work. This time, high exposure is found precisely in the office, and also among occupations that require education, such as computer professionals, business professionals, analysts and engineers. There are also large gaps within each group: among legal, social and cultural professionals, scores range from 0.13 to 0.59.
What does the ranking not say?
Exposure is not a forecast of layoffs. In many occupations, AI is expected to change some tasks and leave the rest, and sometimes to increase demand for workers. The scores are global and not adjusted to the Israeli labour market, and the average is not weighted by the number of workers. In our analysis of data from 13 European countries, no significant relationship has yet been found between exposure and AI adoption and changes in employment.
Sources and transparency
This publication explains the existing research in accessible language. Its accounting example and scenarios are neither official forecasts nor investment recommendations.
What do you think?
Are you already seeing AI change the tasks in your job? Which tasks remain human?