The global economy is running short of growth engines. Populations are ageing, public debt is high and trade fragmentation is weakening traditional export routes. Yet one obvious source of growth remains underused: women.
The International Labour Organisation (ILO) estimates global female labour force participation in 2025 at 49%, compared with 73% for men. Closing this gap could expand the global workforce without requiring population growth or fiscal stimulus.
AI is often framed as a way to widen economic opportunity. Digital platforms widen market access, remote work reduces mobility constraints, and AI-assisted learning lowers the cost of acquiring new skills.
Yet AI is creating a new economic-power grid. Those with the infrastructure, capital and skills can harness its potential, while others may bear the disruption without sharing in the gains. Women are especially exposed to this divide.
Globally, the ILO finds that 29% of female-dominated occupations are exposed to generative AI, compared with 16% of those dominated by men. At the highest exposure levels, the figures are 16% and 3%. This reflects women’s concentration in clerical, administrative and business support roles. Exposure does not imply immediate job losses, but it could reduce labour intensity.
The World Economic Forum describes a “triple whammy” for women. They are more likely to work in roles that AI disrupts and those roles are more likely to be replaced than augmented, according to US data. They are also underrepresented among those building the technology, holding only 30% of global AI workforce roles in 2022, on ILO estimates.
In Sub-Saharan Africa, the risk presents differently. The ILO estimates that only about one in five jobs is exposed to generative AI, compared with roughly one in three in Europe and Central Asia. This reflects employment concentrated in agriculture, manual work and informal enterprises. But lower exposure is not necessarily protection. It may also signal exclusion from the sectors in which AI raises productivity and wages.
According to the International Telecommunication Union, only 21% of women in low-income countries used the internet in 2024, compared with 93% in high-income economies. Informal traders and domestic workers may therefore be insulated from direct AI displacement, but remain off-grid from the digital tools, credit records and scalable markets that help raise wages.
Expanding access to employment
South Africa presents an acute version of this challenge. While ageing economies are running short of workers, South Africa has too few jobs for its available labour.
In the second quarter of 2026, Stats SA data showed that female labour force participation was 54.9%, compared with 64.4% for men. Only 34.3% of working-age women were employed. The female unemployment rate was 37.5%, 7.2 percentage points above that for men. The gap between female participation and employment absorption widened from 14.9 percentage points in 2016 to 20.6 points in 2026. Encouraging participation without creating jobs is not a growth strategy.
AI could weaken an important route into formal work for South African women. Clerical occupations account for 17% of female employment, almost three times the 5.8% share for men. Private sector firms seeking wider margins may target routine entry-level and middle-tier functions first.
The government’s tight purse strings further constrain the public sector’s role as a major source of formal employment for women, particularly in education, health care and administration. With gross government debt near 79% of GDP, the Treasury has limited scope for payroll expansion. This could encourage AI adoption to contain costs and automate administrative work.
Compounding this is South Africa’s invisible unpaid care tax. Stats SA data shows women bear a disproportionate share of domestic work and caregiving. When municipal water networks fail, public transport breaks down and clinics are overstretched, households have to compensate, with the burden falling primarily on women. AI may save an hour writing a report, but it cannot fetch water, care for a relative or navigate broken municipal services. These pressures leave women with less time to reskill.
Policy should focus on who captures the productivity dividend. Existing sector education & training authority (Seta) grants and learnership tax allowances could fund accredited AI training linked to workplace experience, prioritising workers in roles most exposed to automation. Suppliers of AI systems to the government could be required to provide training and credible redeployment plans. Informal businesses need affordable connectivity, digital payment tools and better routes into formal credit markets. Childcare and reliable public services are equally important because they determine whether women have the time and mobility to participate.
In ageing economies, AI helps shrinking workforces produce more. In economies such as South Africa, where jobs are scarcer than workers, the challenge is to expand access to productive employment, particularly for women. Without deliberate intervention, AI could create a highly productive formal core surrounded by an excluded majority, with women pressed against an algorithmic ceiling they can see through but not break.
Packirisamy is group economist at Momentum Group
