The conversation around the impact of
generative artificial intelligence
, or AI, on the labor market has shifted from
whether
it will displace jobs to whose jobs are at stake and how quickly. In this new paradigm, gender is emerging as a key fault line, and nowhere is this tension more evident than in countries where gender disparities already inform who works, as well as at what jobs and under what conditions.
A landmark May 2025
report
by the International Labour Organization, or ILO, and Poland’s NASK Research Institute shed light on these questions, while adding nuance. The report found that women around the world are disproportionately more likely than men to be employed in roles “highly exposed” to automation by generative AI, such as Large Language Model chatbots.
Pawel Gmyrek, the lead author of the ILO-NASK study, told World Politics Review that if the highly exposed jobs identified in the report were to disappear, two women would be displaced for every man. This is especially salient for countries like India, Brazil and South Africa, where state-led digitization is central to economic development strategies.
The Exposure Gradient
The ILO-NASK study, titled “Generative AI and Jobs,” draws on an analysis of nearly 30,000 occupational tasks, assigning each an exposure score based on model predictions of their susceptibility to AI automation. While it emphasizes the likelihood that AI’s impact will result in “transformation” of these jobs more than outright job loss, the report warns of steep “exposure gradients” for some of them.
About one in four global jobs representing 25 percent of total employment are “highly exposed” to AI-driven task transformation, according to the report. But exposure is not evenly distributed between men and women among them, as 41 percent of female employment in high-income countries falls into these high-exposure categories, compared to 28 percent of male employment. Notably, in wealthier nations, a significantly higher overall share of women than men-9.6 percent versus 3.5 percent, respectively-work in highly exposed categories.
“Women in the Global North are more exposed at this point,” Gmyrek noted, citing greater access to AI technologies and higher capital investments in high-income countries as reasons for this heightened risk. Differences in income levels and occupational structures also contribute: The ILO
report
found that high-exposure roles where women are more represented, such as clerical, financial and customer service jobs, are more common in high-income countries.
By contrast, the proportion of jobs that are highly exposed to AI is only 11 percent in low-income countries, whose economies tend to have a larger share of informal, agricultural and manual labor jobs that are less susceptible to automation. In South Asia, the region’s delayed digitization also tempers the immediate risks associated with AI’s impact on work.
However, this masked reprieve may not last. South Asian economies are rapidly digitizing and positioning themselves as global technology hubs. India’s
ambition
to become a developed economy by 2047, for instance, rests heavily on investments in AI, automation and digital public infrastructure, or DPI. So while women in South Asia are less exposed to AI than their Global North counterparts at this stage, explained Gmyrek, the digital gender divide is likely to widen as digitization grows and occupational structures shift.
“In lower-income countries, there’s a real risk of being left behind in terms of benefiting from the progress that this technology can offer,” said Gmyrek. “In contrast, high-income countries not only have more tools and funding to cushion negative effects, they also have better access to the technology’s upside.”
Asia’s AI Sprint
Beyond South Asia, the broader Asia-Pacific region is fast becoming the world’s second-most active region in adopting generative AI, with the Boston Consulting Group
reporting
in March that over 90 percent of businesses plan to scale up their AI adoption in the next two years. The trend is driven in part by the efforts of India and China.
For the region, which sits in the middle range of the global income bracket, Gmyrek’s research finds a moderate but revealing gender gap: Generative AI exposes 24 percent of female employment in the region to potential task transformation, compared to 21 percent for men. While this may seem like a narrower gap, deeper disparities emerge when examining the top two exposure gradients, Gradients 3 and 4.
As Gmyrek explains, this is because these two brackets account for a combined 8.4 percent of female employment, compared to just 4.9 of male employment. In other words, in the highest-risk categories, exposure is almost double for women in the region. So while men and women may face similar AI exposure at the lower end of the risk spectrum, women are overrepresented in the occupations most likely to be transformed or automated by AI.
Without deliberate, inclusive planning, the AI-driven digital shift meant to drive growth could end up deepening old divides and creating new ones, especially for women in emerging economies.
“In countries such as India, these are largely jobs in [business process outsourcing] services, customer support, call centers-roles where female employment is more concentrated,” said Gmyrek. If generative AI leads to the disappearance of these occupations, he said, the impact will be disproportionately felt by women.
However, exposure does not equal replacement, cautions Gmyrek. “We’re not saying these jobs have to disappear. It very much depends on the way policymakers approach it.”
Adapt or Perish
In several countries in the region, women’s employment
remains
concentrated in mid-skill service roles such as clerical work, back-end operations and customer support-all occupations often characterized by repetitive tasks and limited decision-making authority.
In India, this trend intersects with the country’s booming digital public infrastructure ecosystem, anchored by platforms like Aadhaar and the national Unified Payments Interface. These systems have digitized millions of interactions between citizens and the state, but much of their backend operation is outsourced to entry-level staff, many of them women.
According to the industry body
NASSCOM
, women constitute 51 percent of entry-level hiring in India’s Information Technology and Business Process Management, or IT-BPM, sector, which includes business processing operations like data processing, customer service and scheduling-the very roles most vulnerable to automation.
According to Ragui Assaad, a professor at the University of Minnesota’s Humphrey School of Public Affairs, businesses will have to decide whether to automate jobs and displace workers, or augment jobs and increase their productivity.
“India could be a great beneficiary of AI if it prepares its workforce properly,” he added. “It is well positioned, given its leadership in [Information and Communication Technology] and tech services.” Still, Assaad cautioned, certain functions, especially those related to business processing operations, might be automated rather than outsourced. “That risk is indirect,” he said, “but real.”
Similar dynamics are playing out in Brazil, where digital platforms for government services combine automation with human-run support centers.
A recent
study
by economist Bruno Imaizumi using the ILO’s methodology and Brazilian microdata found that here, too, women are significantly more likely than men to work in occupations with high exposure to generative AI. In the highest bracket, 7.8 percent of female workers would be affected, compared to 3.6 percent of male workers.
Zooming out across Latin America and the Caribbean, the World Bank
estimates
that 30 percent to 40 percent of jobs may be affected in some way by generative AI, with women twice as likely as men to face occupational risks.
And a 2019
report
by the McKinsey Global Institute sheds light on the global scope of the challenge. It estimated that 40 million to 160 million women, or between 7 percent and 24 percent of those employed, may need to transition to new occupations by as early as 2030 due to AI automation. “If women can make these transitions, they could be on the path to more productive, better-paid work,” the report notes. “If they cannot, they could face a growing wage gap-or be left further behind.”
The Development Bind
In India, services
account
for roughly 55 percent of the economy, with even higher shares in
Brazil
and
South Africa
. And yet, as automation begins to erode mid-skill white-collar jobs, which have long served as crucial channels for women’s upward mobility, female workers may face a shrinking pool of employment opportunities.
Ashwini Deshpande, professor of economics at Ashoka University, pointed to the digital divide as one factor driving the already large gender disparities in the Indian labor market. “Women have lower access to computers or smartphones for their own use,” she said. “As we digitize, we need to be mindful of that to not exacerbate gender gaps.”
For this reason, Deshpande and other experts argue, AI’s deployment must be consciously shaped, not left to determinism or market forces alone.
Indeed, for Shehnaz Ahmed, who leads applied law and tech research at the Vidhi Centre for Legal Policy, gender-aware policy must be “built into the very architecture of [digital public infrastructure].” And while regulation should support innovation and remain technology-neutral, she said, it must also “evolve to protect vulnerable users.”
Beyond India, the emerging evidence presents a two-sided outlook on automation across the region. An
analysis
of Chinese provincial data from 2006 to 2020 found that AI’s net effect on employment was positive, creating gains for women and workers in labor-intensive sectors, while reducing male overrepresentation in manufacturing.
Meanwhile, a World Bank
report
on the future of jobs in East Asia and the Pacific released earlier this month found that new technologies like robots, AI and digital platforms had largely boosted employment across the region, but the gains were significantly uneven.
“Like any new technology, there are jobs that are going to be displaced, but also jobs created,” said Assaad. The real challenge, he added, lies in workforce preparation. “The key tasks for governments will be in preparing human capital-training, retraining and helping workers shift to other jobs quickly.”
Without deliberate, inclusive planning, the same digital shift meant to drive growth could end up deepening old divides and creating new ones, especially for women in emerging economies.
“Investments will be needed in AI literacy, affordable connectivity and gender-sensitive design to ensure that the benefits of these technologies are broadly shared,” said Edoardo Totolo, deputy managing director of the nonprofit Accion’s Center for Financial Inclusion.
In the rush toward more digitized economies, ensuring that women aren’t automated out of progress may be one of the most defining challenges of our time.
Anisha Sircar is a journalist and researcher whose work spans economics, technology, development and society. Her work has appeared in Reuters, Bloomberg, Quartz, Rest of World, The Wire, Forbes, Scroll.in and other publications. She holds a master’s degree from Columbia University and is co-authoring a forthcoming book on personal finance and the macroeconomy.
The post
AI’s Impact on Jobs Could Widen the Global Gender Gap
appeared first on
World Politics Review
.