ILO working paper 166

Disruption without dividend? - How the digital divide and task differences split GenAI’s global impact

This paper examines how Generative Artificial Intelligence (GenAI) may affect labour markets across 135 countries, covering around two-thirds of global employment. It focuses on how exposure differs between advanced and developing economies, and how digital infrastructure and task composition shape the balance between automation risks and productivity gains.

Abstract

This paper examines how generative artificial intelligence (GenAI) could affect labor markets globally, with particular attention to the uneven distribution of risks and opportunities between advanced and developing economies. Cross-country differences in occupational structure suggest that developing economies face lower aggregate automation exposure than advanced economies but comparable potential for task augmentation. However, disparities in digital infrastructure create an asymmetry: workers in positions vulnerable to automation typically maintain sufficient internet connectivity to experience displacement effects even in low-income settings, while those who could benefit from GenAI augmentation face substantial digital infrastructure gaps that may prevent them from realizing productivity gains. This finding suggests that developing countries may experience the disruptive effects of GenAI faster than its productivity benefits.

At the same time, conventional occupational exposure measures systematically overestimate the impact of GenAI in developing countries by assuming uniform task content across economies. Using data from skills surveys, the article demonstrates that workers in developing countries perform substantially fewer non-routine analytical tasks—the primary targets of GenAI—even within occupations classified as highly exposed. These findings highlight the importance of adapting GenAI exposure measures to reflect developing countries’ distance from the technology frontier.

Key findings

There is a clear positive relationship between GDP per capita and GenAI exposure.

Around 30–32 per cent of employment in high-income countries is exposed

In low-income countries, this figure is closer to 10–15 per cent.

Importantly, this difference is driven mainly by occupations facing higher automation exposure (clerical and certain professional roles).

In contrast, the share of jobs with augmentation potential is more evenly distributed across income levels.
Internet access is a precondition for using GenAI tools. When exposure is adjusted for connectivity, income gaps widen further.

The paper identifies a “small buffer, big bottlenecks” dynamic in developing countries:

Workers in jobs facing automation exposure are often already connected and therefore may experience displacement pressures relatively quickly.

Workers in jobs with augmentation potential frequently lack reliable internet access, limiting their ability to realize productivity gains.

Across the countries with detailed data, 441.8 million jobs fall into augmentation-oriented exposure gradients. Of these, around 66.9 million lack internet access, representing unrealized productivity potential.

This suggests that in some developing economies, disruptive effects may materialize before benefits.
In low- and lower-middle-income countries, the share of jobs exposed to automation is smaller overall. However, these jobs often represent:

Formal, higher-quality service-sector roles

Occupations disproportionately held by women and younger workers

Entry-level clerical and administrative positions that historically served as pathways into decent work.

This raises the risk of a “white-collar bypass”, where office-based jobs that supported upward mobility and women’s labour force participation in advanced economies may not fully materialize in today’s developing countries.
A key contribution of the paper is showing that occupational titles do not reflect identical task content across countries.

Using PIAAC and STEP data for 46 countries, the paper finds that:

Workers in developing countries perform fewer non-routine analytical tasks—even within occupations classified as highly exposed.

The same occupation (at ISCO level) can involve more routine or manual tasks in lower-income contexts.

When country-level exposure scores are adjusted to reflect actual task content, many developing countries move lower in the exposure ranking.

This suggests that global exposure measures based solely on occupational structure may overstate GenAI’s transformative potential in lower-income settings.
Across countries:

Exposure increases with education and household income.

Women and younger workers are overrepresented in occupations facing higher automation exposure, especially in upper-middle- and high-income countries.

In poorer countries, the digitally connected segment of skilled workers may face earlier labour-market pressures.

Sectorally, financial and business services show high exposure across all income levels, but connectivity gaps remain significant in low-income countries.
GenAI’s labour-market impact will depend not only on technological capability but on:

Digital infrastructure

Task organization within occupations

Education and skills systems

Labour-market institutions and social protection

For developing countries in particular:

Expanding digital connectivity is essential to enable augmentation and productivity gains.

Policies should support skills upgrading, especially in non-routine analytical and digital tasks.

Strengthened labour-market institutions and social protection systems will be critical to manage potential displacement pressures.

Gender-sensitive policies are needed to mitigate risks in clerical and administrative occupations.