OIT - Gen AI, occupational segregation and gender equality in the world of work

 ILO Brief 1 Gen AI, occupational segregation and gender equality in the world of work  Research Brief March 2026 Gen AI, occupational segregation and gender equality in the world of work

 Generative artificial intelligence (Gen AI) is evolving at an unprecedented pace, creating both opportunities and challenges for employment, productivity and working conditions. Its impacts are not gender-neutral, often shaped by persistent inequalities between women and men in access to decent work, leadership and economic opportunities.  In countries with available data, female-dominated occupations, such as business administration and clerical support, are almost twice as likely to be exposed to Gen AI as male-dominated ones such as construction, manufacturing and trade (29 versus 16 per cent). They also face much higher automation risk (16 per cent for female vs. 3 per cent of male-dominated occupations).  Exposure to Gen AI varies widely across regions and income levels. In high-income countries, 41 per cent of jobs are exposed, compared to 11 per cent in lowincome countries. These gaps reflect differences in occupational structures and sectoral composition, digital readiness and skills.  Women are more exposed to Gen AI than men in 88 per cent of countries in the sample (see footnote 2). The highest levels of exposure (over 40 per cent of female workers) is found in small island countries in the Pacific and the Caribbean, as well as in European countries such as Switzerland and the United Kingdom, and in the Philippines. This can be likely attributed to a higher share of women in the services sector and the rapid expansion of AI in these economies.

 The higher exposure of women is closely linked to entrenched occupational segregation and the systemic barriers that sustain it. Discriminatory social and legal norms and biases in recruitment, promotion and workplace practices, and macroeconomic and sectoral policies often shape labour markets in ways that have implications for

 The higher exposure of women is closely linked to entrenched occupational segregation and the systemic barriers that sustain it. Discriminatory social and legal norms and biases in recruitment, promotion and workplace practices, and macroeconomic and sectoral policies often shape labour markets in ways that have implications for women's equality of opportunities and treatment.  Gen AI is expected to drive job growth in tech-intensive sectors, yet women remain underrepresented in STEM and AI, making up only 30 per cent of the AI workforce globally. Gaps in access, skills and use are compounded for women facing intersecting inequalities while underrepresentation in AI development, risks perpetuating gender-bias in technologies and deepening the digital divide.  The more widespread impact of Gen AI lies in the quality of employment rather than quantity through its reshaping of tasks, work organisation and skills. It can intensify workloads, reduce autonomy and introduce bias. Yet Gen AI also has the potential to improve job quality by easing physical demands, supporting well-being and enhancing workplace safety and equality, including at enterprise level. This requires Gen AI be designed inclusively and supported by strong labour market institutions and social dialogue.  The policy choices made now will determine whether GenAI drives greater equality or entrenches disparities in the world of work, and whether opportunities are seized or lost. Embedding gender equality in the design, deployment and governance of GenAI, tackling the drivers of occupational segregation, and ensuring women’s representation in AIrelated roles are essential. Social dialogue is fundamental to ensuring that technological transformations enhance working conditions and advance an inclusive world of work. Key points ILO Brief 2 Gen AI, occupational segregation and gender equality in the world of work Why a focus on Gen AI and gender equality? Generative artificial intelligence (Gen AI) has entered the lives of many workers and impacted enterprise operations across the globe, evolving at an unprecedented pace, both in its technological capabilities and its adoption across occupations

Why a focus on Gen AI and gender equality? Generative artificial intelligence (Gen AI) has entered the lives of many workers and impacted enterprise operations across the globe, evolving at an unprecedented pace, both in its technological capabilities and its adoption across occupations and sectors. This transformation brings significant opportunities for productivity gains, innovation and job creation while also raising questions about its effects on employment, task shifts, and working conditions.

Early studies have revealed that the impacts of Gen AI will not be uniform. They will differ across economies based on their income levels and employment structures as well as between groups of workers, often reflecting existing inequalities in labour markets and societies (Gmyrek et al., 2023, 2024 & 2025; Lewandowski et al., 2025). Women and men often work in different occupations, with unequal access to decent work, leadership positions and economic opportunities. These disparities shape how they are likely to be affected by technological change—both in terms of opportunities and risks.

This research brief seeks to provide a global and regional overview of how Gen AI may affect gender equality in the world of work. It draws on new evidence from the ILO’s harmonized microdata collection and applies an ILO index of occupational exposure to Gen AI. By analysing occupational structures through a gender lens, it provides a more detailed and nuanced picture of the potential impacts of Gen AI on women and men around the world.

The brief is organised in two sections. The first section presents key findings on the interaction between occupational segregation, Gen AI exposure and impact on gender equality at the global and regional levels. The second section unpacks these findings and discusses the wider implications of Gen AI for gender equality in the labour market.1

1 Throughout the report ‘employment’ refers to all working modalities – wage workers and non-wage workers. A brief description of the methodology

unpacks these findings and discusses the wider implications of Gen AI for gender equality in the labour market.1

1 Throughout the report ‘employment’ refers to all working modalities – wage workers and non-wage workers. A brief description of the methodology This brief uses the ILO-NASK global index of occupational exposure to Gen AI (Gmyrek et al., 2025) to analyse the potential impacts of Gen AI on gender equality in the world of work. The index is constructed at the task and occupational level: tasks within occupations are scored for their potential exposure to Gen AI, drawing on a combination of worker surveys, expert validation, and AI modelling. These scores are then aggregated to the level of occupations (ISCO-08, 4-digit) and classified along a continuous gradient of exposure, from low to high. The gradient categories can be summarized as follows:  Minimal/no occupational exposure. These are occupations in which tasks cannot be done at all, or only minimally, with Gen AI technology. Many of these occupations consist of manual work.  Gradient 1 represents occupations with low overall Gen AI exposure and significant variability across tasks. While some tasks within these roles may be automatable, the occupation has many tasks that require humans, and thus these occupations are more likely to be “augmented” by AI.  Gradient 2 includes occupations with moderate Gen AI exposure and a mix of highly exposed and minimally exposed tasks, resulting in uneven impacts where some tasks may be disrupted while others remain unaffected, potentially allowing for augmentation.  Gradient 3 captures occupations where a significant portion of tasks are exposed to Gen AI, signalling growing automation risks and requiring adaptation strategies for workers.  Gradient 4 concerns occupations with a concentration of tasks that have high potential automation scores, thus making them most likely to face potential redundancy. ILO Brief 3

portion of tasks are exposed to Gen AI, signalling growing automation risks and requiring adaptation strategies for workers.  Gradient 4 concerns occupations with a concentration of tasks that have high potential automation scores, thus making them most likely to face potential redundancy. ILO Brief 3 Gen AI, occupational segregation and gender equality in the world of work Exposure to Gen AI (gradients 1-4) does not mean necessarily a technologically induced job redundancy in the near future. Risk of full automation is mostly associated with gradient 4 and even for these occupations, the process should not be assumed as automatic. To understand how this occupational exposure translates into differences between women and men, the index is applied to employment data from the ILO’s harmonized microdata collection, which provides the gender composition of employment across occupations in 84 countries2. This allows a comparison of exposure across: • Occupational categories: female-dominated, male-dominated, and mixed occupations (those with important shares of both women and men in the occupation); • Employment outcomes: the share of women and men employed in occupations with different exposure levels; • Regional and country patterns: differences arising from sectoral and occupational structures, levels of digital readiness, and labour market composition. In this way, the analysis first establishes exposure at the level of occupations, then examines how the distribution of women and men across these occupations results in different levels of potential exposure across countries and regions (Gmyrek et al., 2025). Female-dominated occupations are defined as those with a female share in employment of at least 75 per cent in at least 25 per cent of countries. Male-dominated occupations are defined as having a female share in employment of less than 25 per cent across the same proportion of countries with reliable data, (Hegewisch and Liepmann, 2013).

2 A minimum sample size requirement was applied at the occupational

Male-dominated occupations are defined as having a female share in employment of less than 25 per cent across the same proportion of countries with reliable data, (Hegewisch and Liepmann, 2013).

2 A minimum sample size requirement was applied at the occupational level to ensure that results were not driven by individual country contexts. Specifically, for an occupation to be included in the analysis, data needed to be available for at least nine countries—representing over 10 per cent of the 84 countries with reliable 4-digit ISCO data. Main findings Occupational segregation persists in the labour market

Understanding which occupations are dominated by women and men in the labour market is essential to anticipating and addressing the differential impacts of Gen AI. According to the criteria identified in Hegewisch and Liepmann (2013), and for countries where data are available, among the 436 occupations (unit groups at the 4-digit ISCO-08), 82 occupations (19 per cent) are female-dominated, 89 occupations (20 per cent) are male-dominated, while the remaining 266 occupations (61 per cent) are mixed (important presence of both women and men).

Grouped in six clusters (See Figure 1), femaledominated occupations are prominent in the following sectors: health and care, teaching, social work and culture, business administration and clerical support, personal services, sales and food preparation, and finally, textiles and wearing apparel manufacturing (see Figure 1).

Men are instead concentrated in construction, building, manufacturing and trade as well as in occupations relating to science and engineering, information and communication technologies, protective services, drivers, and agriculture workers. Male-dominated occupations also include some chief executives, senior officials and armed forces (See Figure 2). The remaining occupations (266) are considered “mixed”, in that there is an important presence of both women and men. These include occupations across all exposure gradients ranging from hotel

workers. Male-dominated occupations also include some chief executives, senior officials and armed forces (See Figure 2). The remaining occupations (266) are considered “mixed”, in that there is an important presence of both women and men. These include occupations across all exposure gradients ranging from hotel managers (gradient 1) to call centre workers (gradient 4), as well as occupations that are not exposed to Gen AI technology.

Consequently, while the overall dataset covers 84 countries, not all occupations met this minimum coverage criterion, particularly once data were disaggregated by sex. ILO Brief 4 Gen AI, occupational segregation and gender equality in the world of work  Figure 1: Female-dominated occupations

Note: Sample size (number of countries with reliable data for the occupation) in parenthesis Source: ILO harmonized microdata collection  Figure 2: Male-dominated occupations

Note: Sample size (number of countries with reliable data for the occupation) in parenthesis Source: ILO harmonized microdata collection ILO Brief 5

Gen AI, occupational segregation and gender equality in the world of work Female-dominated occupations comprise an important share of female employment in high -income countries

Female-dominated occupations account for important shares of female employment in highincome countries, given the importance of occupations in health, education, social work, business administration and clerical support, in these economies (figure 3).

 Figure 3: Female concentration in femaledominated occupations by region and income level

Source: ILO harmonized microdata collection.

In lower-income countries, where larger shares of women are employed in agriculture or in lowproductivity service industries (e.g., retail trade, food and accommodation, other services), the female-dominated clusters account for lower shares of female employment. This pattern reflects both lower overall female labour force participation and the earlier stage of structural transformation in

women are employed in agriculture or in lowproductivity service industries (e.g., retail trade, food and accommodation, other services), the female-dominated clusters account for lower shares of female employment. This pattern reflects both lower overall female labour force participation and the earlier stage of structural transformation in these economies. As many women remain concentrated in subsistence farming or informal activities, they are less likely to be employed in the formal service sectorssuch as health, education, or public administrationand occupations that are traditionally highly feminized in upper-middle and high-income countries. Countries where femaledominated occupations account for lower shares of female employment are primarily in Africa and in the Asia and Pacific region.

While male concentration in male employment is also important (accounting for 40 to 60 percent of total male employment), income and regional differences across countries are less striking (figure 4). This is because male-dominated occupations are spread more widely across clusters within all three broad economic sectors (agriculture, industry, and services) and male labour force participation is high regardless of country income level.

 Figure 4 : Male concentration in maledominated occupations by region and income level

Source: ILO harmonized microdata collection.

Female-dominated occupations have higher exposure to Gen AI A significantly higher proportion of female-dominated occupations (29 per cent) and mixed occupations (28 per cent) are exposed to Gen AI, compared to just 16 per cent of male-dominated occupations (see Figure 5). In other words, female-dominated occupations are almost twice as likely to be exposed to Gen AI as male-dominated ones. This figure, however, captures only part of total employment exposed to Gen AI (as discussed below) and should be interpreted in the ILO Brief 6 Gen AI, occupational segregation and gender equality in the world of work wider context of labour market exposure across countries.

only part of total employment exposed to Gen AI (as discussed below) and should be interpreted in the ILO Brief 6 Gen AI, occupational segregation and gender equality in the world of work wider context of labour market exposure across countries.

 Figure 5: Exposure to Gen AI for female-dominated, male-dominated and mixed occupations

Source: ILO harmonised microdata collection

Female-dominated occupations are also more likely to have a higher degree of exposure to Gen AI with 16 per cent of occupations in gradient 3 and gradient 4, compared to 13 per cent for mixed occupations, and only 3 per cent for maledominated occupations. These occupations have a significant portion of tasks that are at greater risk of exposure to AI-driven automation and require adaptation strategies that support both workers and enterprises in addressing these risks while leveraging technology for productivity and skills upgrading.

The female-dominated occupations that belong to these high-exposure categories are primarily the ones under the business administration and clerical support cluster (see Annex table 1). Specifically, female-dominated occupations in gradient 4 include for example typists and word processing operators, accounting and bookkeeping clerks and payroll clerks. Gradient 3 ones include secretaries, receptionists, librarians, translators and interpreters.

In contrast, the male-dominated occupations that are exposed in gradients 3 and 4 mostly fall under the ICT cluster. Such jobs comprise for example web and multimedia developers, software developers, and application programmers.

In terms of employment, exposure to Gen AI differs across the world Countries in different regions and income brackets differ widely in their levels of exposure to Gen AI, largely due to differences in occupational structures and sectoral compositions, as well as access to digital infrastructure, firm-level readiness and foundational skills of individuals. Overall, 41 per cent of employment is potentially exposed to Gen AI technology in high-income

widely in their levels of exposure to Gen AI, largely due to differences in occupational structures and sectoral compositions, as well as access to digital infrastructure, firm-level readiness and foundational skills of individuals. Overall, 41 per cent of employment is potentially exposed to Gen AI technology in high-income countries compared with just 11 per cent in low-income countries. Furthermore, 9.6 per cent of employment in high-income countries is in the category of highest exposure to automation by Gen AI (Gradient 4), compared to just 0.3 per cent in low-income countries (Gmyrek et al., 2025). In general, advanced economies have higher exposure to Gen AI as a result of greater diversification of their economies and thus occupations. Specifically, low exposure to Gen AI in lowand middle-income countries primarily reflects the sectoral composition of their economies, where employment remains concentrated in agriculture and lower productivity services, rather than an absence of digital connectivity. In addition, task content tends to differ within same occupations located across highand low-income countries (Lewandowski, 2025, Gmyrek et al., forthcoming).

Women in employment are more exposed than men to Gen AI

Considering all four exposure gradients, women are more exposed overall to Gen AI than men in most (88 per cent) of countries in the sample. Small island countries in the Pacific and in the Caribbean, followed by European countries (Bosnia and Herzegovina, Switzerland, United Kingdom) and the Philippines (figure 6) is where women have the highest levels of exposure to Gen AI (over 40 per cent of total female employment). Overall, Europe and Central Asia and Latin America and the Caribbean are the two regions with the highest levels of exposure for female workers on average, while Africa and Asia have the lowest exposure.

The few countries where women are less exposed ILO Brief 7

Europe and Central Asia and Latin America and the Caribbean are the two regions with the highest levels of exposure for female workers on average, while Africa and Asia have the lowest exposure.

The few countries where women are less exposed ILO Brief 7 Gen AI, occupational segregation and gender equality in the world of work to Gen AI than men include countries with higher prevalence of agricultural employment (e.g., Burundi, Tanzania) sometimes also combined with low female labour force participation (e.g., Afghanistan, Bangladesh, Iraq, Pakistan, Sudan). Countries where men have relatively high exposure (although still lower than among women) include some countries in Europe (Switzerland and the United Kingdom), Asia (Maldives) and some Arab States (Lebanon, United Arab Emirates).

For most occupations, the more widespread impact of Gen AI lies in the transformation of work and working conditions. In many cases, Gen AI is likely to reshape tasks within occupations, change work organisation and processes, or modify supervision and performance management. It is also likely to alter the skills required to perform certain jobs, as it gets integrated into digital tools that workers are required to use to perform their duties. For most jobs, these technological changes are more likely to reshape responsibilities and affect the quality of employment, rather than eliminate their current tasks altogether.

 Figure 6: Percentage of female and male employment in exposure gradients 1-4 by country

Note: Latest year available in parenthesis Source: ILO harmonized microdata collection ILO Brief 8

Gen AI, occupational segregation and gender equality in the world of work Unpacking the findings: occupational segregation, systemic barriers, and AI underrepresentation

The higher exposure of women to Gen AI reflects the persistent and gendered patterns of occupational segregation that continue to shape labour markets globally. Technologies, including

Unpacking the findings: occupational segregation, systemic barriers, and AI underrepresentation

The higher exposure of women to Gen AI reflects the persistent and gendered patterns of occupational segregation that continue to shape labour markets globally. Technologies, including Gen AI, are not inherently neutral, but are embedded within and shaped by societal structures and relations (Spencer, 2018). Past waves of technological change have shown how, rather than disrupting gendered patterns and divisions of labour, technology can reproduce them in new forms (Howcroft and Rubery, 2019).

This section unpacks how these patterns are reinforced by structural and institutional factors, including discriminatory social norms, intensive and unequal unpaid care and domestic work, and economic and labour market policies that do not fully address the different needs of both women and men. The under-representation of women in Gen AI employment and development can also influence how technologies are designed and deployed, with direct consequences for both the quality and quantity of their employment.

Drivers of occupational segregation While there has been an increase of women in professional roles in the last two decades, they continue to be overrepresented in clerical and administrative roles, which are highly exposed to Gen AI automation (Gmyrek et al., 2023 & 25). Women are also more likely than men to perform routine cognitive and codifiable tasks, which are at a higher risk of substitution by Gen AI across all sectors and occupations, and less likely to have analytical and abstract tasks, which are more likely to be complemented by this technology (Brussevich et al., 2019). These disparities also reflect patterns of vertical segregation, with women less likely to occupy senior or decision-making roles within the same occupational categories (ILO, 2019). As a result, women’s employment may be more affected by technology in specific occupations or tasks.

Studies find that in advanced and emerging

of vertical segregation, with women less likely to occupy senior or decision-making roles within the same occupational categories (ILO, 2019). As a result, women’s employment may be more affected by technology in specific occupations or tasks.

Studies find that in advanced and emerging economies, the share of routine and automatable tasks is higher among older women and those with lower levels of education (Brussevich et al., 2019). In contrast, in lowand middle-income countries, Gen AI exposure tends to be higher among women, more educated workers and those in urban areas, who tend to demonstrate higher digital readiness and greater access to infrastructure (Demombynes et al., 2025, Gmyrek et al., 2024).

At the same time, women are overrepresented in care-related occupations, which are less likely to be automated due to their reliance on social and interpersonal skills (ILO, 2018). However, there is increasing interest from both the public and the private sector in integrating such technologies in the care sector– for example, remote patient monitoring, automated charting and nursing care plans, and clinical prediction – given demographic changes, increasing care needs and budgetary pressures that have restrained staffing (O’Connor et al., 2023). Emerging evidence on the integration of AI in nursing reveals that if implemented without adequate consultation, training, interoperability or labour protections, AI can risk intensifying work, reinforcing existing hierarchies and inequalities, and creating new decent work deficits for a predominantly female workforce (ILO, forthcoming).

Occupational segregation is shaped by a range of intersecting and mutually reinforcing structural barriers affecting supply and demand side constraints. These include social and legal norms that affect educational and occupational pathways. Despite progress in recent years, legal restrictions continue to limit women’s occupational choices in many countries, contributing not only to occupational segregation but also labour-market

barriers affecting supply and demand side constraints. These include social and legal norms that affect educational and occupational pathways. Despite progress in recent years, legal restrictions continue to limit women’s occupational choices in many countries, contributing not only to occupational segregation but also labour-market distortions and reduced innovation and productivity (Blau and Kahn, 2017). In 2023, 21 economies limited a women’s ability to work at night, 49 prohibit women from working in hazardous jobs, and 65 bar women from working in the same industries as men (World Bank, 2023).

Persistent gender stereotypes continue to shape expectations around what constitutes appropriate work for women and men (Carranza, Das, and Kotikula 2023). These include beliefs that women are naturally more caring, more suited to repetitive or household-related tasks, have lower aptitude in science or mathematics or less leadership potential. These same norms often associate men with technical, managerial and decision-making roles, ILO Brief 9 Gen AI, occupational segregation and gender equality in the world of work while positioning women in nurturing, administrative or supportive roles (Anker, 1997).

These norms can be reinforced through biases and discrimination in workplace practices including recruitment, promotion, training opportunities and organizational culture. For instance, job postings can reflect and reproduce gender norms, discouraging women from applying or channelling them into certain occupations, while selection processes may favour male applicants even with identical qualifications. In fact, when used as a source of labour market information, Gen AI tools can reinforce such stereotypes, as they can replicate gendered perceptions of occupational segregation, appropriate remuneration, or the prestige associated with typically maleor femaledominated jobs (Gmyrek, Lutz and Newlands, 2025).

In addition, women’s entry into and retention in male-dominated occupations, as well as their

segregation, appropriate remuneration, or the prestige associated with typically maleor femaledominated jobs (Gmyrek, Lutz and Newlands, 2025).

In addition, women’s entry into and retention in male-dominated occupations, as well as their promotion prospects, may be further constrained by workplace factors when characterized by long and inflexible working hours, violence and harassment and unequal access to training, skills upgrading and leadership roles (Carranza et al., 2023).

The intensive and unequal unpaid care and domestic work done by women further constrains their time, mobility and occupational choices. Globally, women perform more than three-quarters of total unpaid care work, on average 3.2 times more than men (ILO, 2018). Recent ILO estimates reveal that care responsibilities are the primary contributor to the gender employment gap. Indeed, excessive and unequal care responsibilities keep 708 million women outside the labour force globally (ILO, 2025). This unequal distribution of care work often influences women’s decisions to seek part-time or flexible employment arrangements to be able to accommodate care responsibilities. This limits their access to a broader range of occupations, contributes to their concentration in lower-paid and less secure jobs, and hinders their ability to access and remain in decent work (ILO, 2018).

In addition, macroeconomic and sectoral policies shape the structure of labour markets and influence the distribution of employment opportunities across sectors and occupations, with gend