30% of global jobs exposed to GenAI; developing countries face ‘white-collar bypass’ risk, says ILO

Generative Artificial Intelligence (GenAI) is expected to transform labour markets worldwide, but its impact will vary significantly between advanced and developing economies, according to new cross-country research analysing employment data across 135 countries.

The study finds that while AI exposure is higher in advanced economies, developing countries may face a different challenge: limited digital infrastructure could delay productivity gains while still leaving some workers vulnerable to automation.

The research shows a strong correlation between economic development and exposure to generative AI. In high-income countries, around 30–32% of jobs are exposed to GenAI, compared to about 15% in low-income countries.

Clerical workers, administrative roles, and some professional occupations were found to be among the most exposed to automation risks from generative AI.

One of the most important findings relates to internet access. The study found that millions of workers whose jobs could benefit from AI tools cannot use them due to lack of internet connectivity.

In some low-income countries, the gap is even more severe. For example, in Mali, only 12% of jobs are exposed to GenAI, and just 10% of those workers have internet access, meaning only about 1.2% of workers are in jobs that are both exposed to AI and able to use it.

By contrast, in countries such as Switzerland and Singapore, roughly one in three workers is employed in AI-exposed jobs, and nearly all of them have internet access.

The research also suggests that generative AI could disproportionately affect middle-income and office-based jobs, particularly clerical and administrative roles that historically provided stable employment and career mobility.

These jobs have historically played a major role in expanding the middle class and enabling women’s participation in the workforce. 

In high- and upper-middle-income countries, 17% of female workers face automation exposure, compared to about 11% of male workers. Younger workers aged 16–35 also face significantly higher exposure compared to older workers.

At the same time, AI exposure increases with education levels. College graduates face exposure rates between 32% and 38%, regardless of whether they are in high- or low-income countries, showing that generative AI primarily affects knowledge-based work.

Another key insight from the research is that AI exposure depends not only on job titles but also on the actual tasks performed within jobs. Workers in the same occupation often perform different tasks depending on the country, which affects how exposed they are to AI.

In developing countries, jobs often include more routine and manual tasks and fewer analytical and digital tasks, meaning AI may have less immediate impact than global estimates suggest.

When researchers adjusted AI exposure based on actual job tasks rather than occupational titles, many developing countries showed significantly lower exposure levels than previously estimated.

Overall, the research concludes that generative AI will reshape labour markets globally, but the effects will be uneven. Advanced economies are likely to see faster productivity gains due to higher digital adoption, while developing economies may face slower benefits and potential job disruptions in key sectors.

The study warns that without investment in digital infrastructure, skills development, and labour market policies, generative AI could increase global inequality, with productivity gains concentrated in wealthier economies while some workers in poorer countries face displacement risks without corresponding opportunities.