1APR

Women face double the AI job risk of men

2 min read

12:41UTC

A 135-country study found women hold nearly twice the share of highest-risk AI-exposed jobs globally. In wealthy nations, the gap is almost threefold.

← Back to Trump declares victory and withdrawal Jump to analysis ↓

ConflictAssessed

Key takeaway

Women hold nearly double the share of jobs most exposed to generative AI displacement.

A joint study by the International Labour Organisation and the World Bank, published in March and covering 135 countries and two-thirds of global employment, found that one in four workers holds a job with some generative AI exposure.⁠¹ In the highest-risk category, 4.7% of women are exposed globally, compared with 2.4% of men. In high-income countries the gap widens: 9.6% of women versus 3.5% of men.

Clerical and administrative work is the exposure vector, roles historically concentrated among women. Advanced economies show 34% overall workforce exposure; low-income countries show 11%.⁠² Wealth correlates with AI penetration into clerical roles where women are concentrated. Wealth also correlates with the gender disparity.

This dimension has been absent from the displacement debate. CFO surveys, tech layoff trackers, and congressional proposals: none address gendered impact. The 502,000 roles the Atlanta Fed projects will be cut are not distributed evenly. Clerical and administrative roles most exposed to generative AI are held disproportionately by women, and no current policy framework accounts for it.

Deep Analysis

In plain English

A major global study found that women are about twice as likely as men to work in jobs that generative AI is capable of automating. In richer countries the gap is even larger: nearly three times as likely. This is because AI is best at handling text, data, and communication tasks, which are the core of administrative and clerical work. Those roles have historically been held more by women. No current government policy specifically addresses this disparity.

Deep Analysis

Root Causes

The gender exposure gap traces directly to occupational segregation. Administrative, secretarial, and clerical roles, which have been disproportionately held by women in OECD economies since the 1960s, are the roles with the highest generative AI task coverage. This segregation was not caused by AI; it reflects decades of labour market structure that AI is now exploiting.

The high-income country amplification effect (9.6% vs 4.7%) reflects the greater penetration of digital work in wealthy economies. In low-income countries, clerical roles are less computerised and therefore less exposed to generative AI substitution. The paradox is that development increases women's AI exposure risk.

What could happen next?

Consequence
AI-driven displacement will widen the gender pay gap if affected women are absorbed into lower-wage service roles, repeating the 1980s-90s pattern from PC-era clerical displacement.
Medium term · Medium
Risk
No current legislative proposal in the US, EU, or UK specifically addresses gendered AI exposure, meaning the disparity will compound without deliberate policy intervention.
Medium term · High
Meaning
The 135-country ILO dataset is the first evidence base capable of grounding gendered AI policy; its publication removes the 'we don't have data' objection to targeted intervention.
Long term · High
Opportunity
Countries that invest now in reskilling women in AI-adjacent roles have a window before displacement concentrates to capture the productivity benefits on both sides of the transition.
Medium term · Medium

Source Landscape

This story draws on neutral-leaning sources

Primary parallel: The introduction of word processing and spreadsheet software from the mid-1980s through the 1990s eliminated approximately 40% of typing and data-entry roles in the US, a workforce that was roughly 80% female. However, the overall female labour force participation rate rose during this period as new service and administrative roles absorbed displaced workers, often at lower wages.

Counter-parallel: The 2000s offshoring wave disproportionately displaced women in call centres and back-office processing in the US and UK. Unlike the PC transition, this displacement was not offset by comparable local job creation; many affected women exited the formal labour market. Generative AI exposure is more structurally similar to offshoring than to PC adoption.

Immediate: If the Atlanta Fed's 502,000 projected US cuts follow the ILO gender distribution, women could account for 60-65% of AI-attributed displacement despite representing 47% of the workforce. At a median clerical wage of approximately $42,000, this represents $13-15 billion in lost household income concentrated among lower-income female earners.

Structural: ManpowerGroup found AI roles command 67% higher salaries than traditional software positions. The 3.2-to-1 demand-to-supply ratio for AI skills is primarily a male-dominated pipeline: women represent approximately 26% of computing and AI-related graduates. Displacement risk is front-loaded for women; benefit capture is back-loaded and structurally less accessible.

Consensus view: The ILO's Director-General Gilbert Houngbo called the study the most comprehensive cross-country evidence yet that AI automation risk is structurally gendered, and called for its findings to be integrated into the G20 AI governance agenda.

Counter-view: Economist David Autor at MIT argues that clerical job displacement, while real, has been overstated in previous automation waves; women who lost clerical roles in the 1980s PC revolution largely transitioned into service roles, and the same adaptive capacity may apply to generative AI exposure.

Key tension: Whether the 2x global and 3x high-income-country exposure gaps translate into equivalent displacement ratios, or whether women in clerical roles demonstrate the same adaptive resilience as in prior automation cycles.

Sources:International Labour Organization / World Bank·International Labour Organization / World Bank

Mentions:World Bank →International Labour Organisation →

First Reported In

Update #3 · The AI jobs data contradicts itself

International Labour Organization / World Bank· 28 Mar 2026

Read original →

Causes and effects

This Event

Women face double the AI job risk of men

The first global dataset on gendered AI displacement reveals a disparity that no current policy framework addresses.

Led to

Goldman counts 25,000 jobs lost monthly

Goldman's finding that 86% of the most vulnerable displaced cohort are women reinforces the ILO-World Bank gender displacement risk finding.

Occurred 6 Apr 2026

Read story →

Different Perspectives

Wang Yi

China's foreign minister and Djibouti's issued a joint communique on 21 October saying the war should not reach the Horn of Africa and stressing free navigation through Bab al-Mandeb. The text names neither Iran nor the Houthis, and Djibouti hosts China's only overseas military base, opposite the Yemeni coast.

Ministry of External Affairs

India's foreign ministry called the attacks on Saudi civilian and energy sites "unacceptable" on 22 October and said they must stop at once. An Indian died at Dammam, Indians were injured there and at Shaybah, and about 4,000 Umrah pilgrims remain stranded by one MP's count.

Iraq

Iraq closed the al-Sheeb crossing with Iran to travellers and trucks from 21 October and said other crossings would close gradually. Unlike September's closure, which ended on a staged timetable, this one has none, and Iran's interior ministry said it received no notice.

Abdel Fattah el-Sisi

Egypt's president said on 21 October that Cairo's support would "defend Saudi territory, and does not include an attack on Yemen". Araghchi warned that Egyptian involvement would further complicate the situation, and Cairo's limit matches the one Ankara set on 8 October against attacking another state's soil.

UK Ministry of Defence

The MoD said on 20 October that RAF Typhoons over Saudi Arabia will not strike the Houthis, but extended tanker support to Saudi operations including against Houthi targets in Yemen. The tanker line sets no limit on strikes, so British aircraft may now refuel Saudi jets on bombing runs.

Turkey

Ankara sent the Mecca Joint Defence Agreement to parliament around 20 October, and its published text treats an attack on one party as an attack on all. Article 16 brings it into force only on the last ratification, so the 5 October invocation in Riyadh could not yet bind Turkey.

Key Figures

Countries covered135

Female employment at highest AI risk (global)4.7%

Male employment at highest AI risk (global)2.4%

Female employment at highest AI risk (high-income countries)9.6%

Male employment at highest AI risk (high-income countries)3.5%

Advanced economy workforce exposure34%

Low-income country workforce exposure11%

In everyday terms

If you or someone you know works in an administrative, clerical, customer service, or secretarial role in a high-income country, the ILO data suggests that role sits in the highest AI exposure category. The risk is not necessarily immediate job loss, but a narrowing of available positions as employers gradually automate task components without announcing formal cuts.

What’s Uncertain

Exposure vs displacementHigh AI task exposure does not automatically translate into job displacement. The ILO study measures exposure risk, not actual displacement rates. The relationship between the two depends on employer adoption speed, task complementarity, and labour market conditions.

Low-income country data qualityThe 135-country coverage includes many low-income countries where employment data is collected less rigorously. The 11% low-income exposure figure may be understated if informal sector roles are not captured.

Sources

2 centre

International Labour Organization / World Bank · International Labour Organization / World Bank

Entities Involved

World Bank →International Labour Organisation →

Evidence & sources

3 sources

SourceILO and World Bank joint study, March 2026, covering 135 countries and two-thirds of global employment
SourceAnthropic observed exposure research, Massenkoff and McCrory, 2026: workers in high-exposure roles are 'older, female, more educated and higher-paid'
SourceDallas Fed, January 2026: employment decline concentrated in workers under 25 via collapsed job-finding rates

Understand the Context