To do this, about 7,500 different professional profiles and tasks were defined, which were then assessed in terms of their potential for automation. For instance, taking a simplified approach, the occupation of primary school teachers was broken down into lesson preparation, dissemination of knowledge, creating a suitable learning environment, conducting parent meetings, etc. While some of these tasks clearly have less potential for augmentation or execution by LLM, some, such as preparing daily and longer-term lesson plans for example, can be supported and thus made more efficient.

Jobs with Potential for Automation

Occupations which can be strongly automated according to this classification include administrative and communication tasks, as well as client services such as making appointments, correspondence, translations, processing client inquiries, and data-based tasks such as accounting activities ranging from the preparation of invoices to monitoring payments.

By assessing the individual tasks it was possible to identify a large number of occupations in which these responsibilities are present to a limited degree as well as such which exhibit a high level of them. It is anticipated that these latter occupations will have significant potential for augmentation by LLMs. These include, for example, teaching, the medical, technical, and legal professions, and software development and analysis.

In another step, the ILO experts analysed the global share of these occupations according to various income groups and their distribution between women and men. The share of employment that is defined as being potentially completely subject to automation is seen as being very moderate, but there are significant differences between men and women.

Women are more strongly affected

In particular in the higher income ranges, women will be impacted disproportionately strongly by automation and thus by the possibility of job losses. This is due to the relatively high share of employment among women in occupations with accounting and secretarial responsibilities.

Regional differences

From a regional perspective, the possibility of deploying LLMs in the developed countries is significantly higher than in the rest of the world, due to the required energy consumption and access to the Internet. Individual countries, such as India or the Philippines, may be very strongly impacted by the disappearance of call centres.

So, it is not likely that we will run out of work. The potential for automation is never fully exploited and it is also impossible to foresee the development of new occupations, as it was for example in the field of web marketing.

That said, ChatGPT and other LLMs are here to stay and they will have disruptive impacts in some occupations. Accordingly, as a society, it is important for us to react to the changing technological conditions, limit negative impacts, and seize the opportunities.