How might generative AI impact different occupations?

The figure below shows the results of the analysis, which was based on a global assessment of the 436 detailed occupations that comprise the International Standard Classification of Occupations. 

Select one of the occupations major groups to filter the occupations.

Each occupation is made up of tasks and each of the tasks with an occupation received a 0-1 score on the ability to automate that particular task, with 0 indicating it is not possible and 1 indicating it was entirely possible.

The figure plots these occupations according to its average score (mean of all tasks within an occupation) and its standard deviation (the dispersion of task-level automation scores within an occupation). 

The intersection of these scores was used to elaborate a framework for classification based on four categories:

• Automation potential: Jobs where the majority of tasks today could theoretically be performed with GenAI – such jobs could potentially be automated.
• Augmentation potential: Jobs where some tasks can be performed using GenAI, but the majority need to be done by humans. Such jobs can be augmented by GenAI, speeding up some tasks and allowing more space for creative human work and new tasks.
• Big unknown: A category between automation and augmentation potential, representing jobs where the balance of tasks today is between those that can be done with GenAI and those that cannot. This balance might shift over time as technology improves and occupations evolve, moving some jobs closer to automation and others to augmentation potential.
• Not affected: Occupations where the majority of tasks cannot be performed with GenAI (e.g. physical tasks).

What is the possible effect of generative AI on employment?

To assess the potential impact on existing employment from generative AI, we apply the scores of occupational exposure to generative AI to labour force survey data for more than 140 countries. This allows us to derive both global estimates and estimates among countries in different income brackets.

With respect to automation, the share of employment that is exposed is highest in high-income countries, reflecting their greater economic and labour market diversification. In low- income countries, the share of employment potentially exposed to automation is much smaller, due to the greater share of workers employed in occupations that would not be exposed to generative AI technology such as in agriculture, transport or food vending. 

A larger share of jobs falling into the augmentation category suggests that, at least in the near future, generative AI systems similar to GPT are more likely to become productivity tools, supporting and speeding up the execution of some tasks within certain occupations. 

The digital divide will, nonetheless, influence how the benefits of such productivity tools are distributed among societies and countries, with high-income countries and privileged groups likely to reap the biggest rewards. Low-income countries, in particular, are at risk of falling behind. While up to 10.4 per cent of employment in these countries is found in the potential augmentation category, in practice potential benefits of GenAI technologies are likely to be limited, as the lack of reliable infrastructure will constrain their application.

NOTE: “The Big Unknown” refers to those professions that are at the junction of automation and augmentation, meaning they could either be transformed through GenAI-enabled augmentation or suffer displacement through automated substitution.