AI and the gender pay gap are becoming more closely linked as automation reshapes the roles women are most likely to hold and AI skills create new salary premiums that are not being shared equally. This blog explores why artificial intelligence could deepen workplace inequality unless organisations act now on reskilling, fair access, and responsible AI development.
AI is coming for the jobs women actually have, not the ones we wish they had. And the pay gap? It’s about to get a whole lot wider.
Every year on International Women’s Day, someone posts an infographic about the gender pay gap. Someone else shares it. A corporation turns its logo purple. And then, on March 9th, nothing changes
This year feels different, not because the mood is hopeful, but because the stakes have quietly become enormous. While we’ve been busy celebrating incremental progress, something much larger has been gathering momentum. Artificial intelligence is remaking the world of work faster than any policy, petition, or purple logo can respond to. And if you look at who it’s coming for first, you’ll find it has an unmistakably female face.
Let’s talk about what’s actually happening.
The Jobs at Risk Aren’t Random
When we talk about AI “disrupting” jobs, there’s a tendency to imagine it as some kind of neutral economic weather system, a storm that hits everyone equally. It isn’t. The jobs most exposed to automation aren’t distributed evenly across genders. They cluster, with striking consistency, in the areas where women are most likely to work.
Administrative roles. Customer service. Data entry. Clerical processing. These are the jobs that AI can replicate most readily, and they are overwhelmingly held by women. The data backs this up with uncomfortable clarity.
| 2xWomen are nearly twice as likely to work in roles at high risk of automation | 65MWomen’s jobs at risk globally, vs. 51 million men’s | 22%Women’s share of the global AI workforce, the people building the tools | 25%Lower rate at which women are adopting AI tools compared to men |
In high-income countries, the disparity is even sharper. In Australia and New Zealand, nearly 10% of women’s jobs are flagged as high-risk, compared to 3.5% of men’s. In the UK, an estimated 119,000 clerical roles in tech and financial services alone are projected to be automated over the next decade, roles disproportionately filled by women.
This isn’t a coincidence. It’s the product of decades of occupational segregation. The way society has quietly herded women into certain kinds of work is now colliding head-on with a technology optimised to do exactly those kinds of tasks.
Women are being automated out of the middle, while men are more likely to find themselves elevated by AI into better-paying roles. The gap isn’t just persisting. It’s being structurally encoded.
The AI Skills Gap Is the New Pay Gap
Here’s where it gets more complicated, and more interesting. Because the story isn’t just about which jobs disappear. It’s about who gets to benefit from AI when it creates new opportunities.
AI skills command a premium. Roles that require them pay significantly more. That’s not a prediction; it’s already happening. And access to those skills? Not equal.
Research from Wharton and Oxford finds a persistent gender divide in access to AI training at work. Women are less likely to be offered it, less likely to be sponsored into it, and (here’s the part that should really make you angry) less likely to be rewarded for it when they pursue it independently. IMD’s research is direct: the gender gap in AI skill access is actively fuelling a pay gap that already sees women earning around 20% less than men.
Men are more likely to list AI skills on LinkedIn. More likely to be upskilled by their employers. More likely to work in environments where AI fluency is rewarded and celebrated. Women, particularly in mid-level professional roles, are more likely to find themselves left out of that loop entirely, not through any grand conspiracy, but through the compound effect of a thousand smaller assumptions.
But Why Aren’t Women Just Using AI?
This is where the conversation usually takes a wrong turn. The response, when this data emerges, is often to suggest that women simply need to lean in harder, learn faster, adopt more readily. The problem, the implication goes, is with women’s choices.
That framing ignores some inconvenient research from Harvard Business School. Yes, women are adopting generative AI tools at a lower rate than men, roughly 25% lower on average, across studies spanning 140,000 workers in multiple countries. But the reason matters enormously.
Women aren’t avoiding AI because they’re technophobic or resistant to change. They’re avoiding it in significant part because they’re concerned about the ethics of the tools, and because they reasonably fear being judged for using them. And that fear, it turns out, is entirely rational. Studies have found that when women use AI tools to produce the same output as their male counterparts, they are perceived as less competent. The same tool. The same work. A different verdict.
So women face a double bind. Don’t use AI, and fall further behind in skills and productivity. Use AI, and face a competence penalty that men simply don’t encounter. The algorithm, as it turns out, has absorbed our biases along with everything else.
The Care Work Question Nobody Wants to Answer
Amid all the focus on office roles and white-collar work, there’s a category of employment almost entirely absent from the disruption conversation: care work. Nursing, childcare, elder care, teaching. Overwhelmingly female, chronically underpaid and genuinely hard to automate.
AI cannot hold a grieving person’s hand. It cannot read a child’s unspoken distress. It cannot build the trust that underpins good medicine. And yet society consistently prices this work as though it’s worth less than a spreadsheet. If the roles AI is coming for first are the ones women do, perhaps the lasting shift will be to finally reveal the irreplaceable value of the roles AI cannot do. Whether we’ll choose to pay for that value is another question entirely.
Could AI Actually Help?
It would be dishonest not to acknowledge the counterargument, and there is one worth taking seriously.
Some economists argue that as the labour market for routine cognitive tasks tightens, employers desperate for AI-literate workers will be forced to expand their talent pools and look beyond the usual suspects. There is tentative evidence that the gender gap in AI skills, while still significant, is narrowing. LinkedIn data shows women’s representation among those listing AI engineering skills has grown from 23.5% in 2018 to 29.4% by 2025. Progress, even if incomplete.
There is also the “time-release” argument: that AI, by automating domestic and administrative tasks, frees up time that disproportionately burdens women. If AI makes the mental load lighter, that could, in theory, create space for women to engage more fully in the workforce and in professional development. The research is cautiously optimistic on this specific point.
But optimism requires conditions. It requires deliberate policy. It requires organisations that actively invest in retraining women, not just the employees who already look like the people making decisions. It requires AI systems built with diverse teams, not ones designed in rooms where women represent just 22% of those present. It requires, in short, a level of intentionality that the market has consistently failed to deliver on its own.
What Actually Needs to Happen
Targeted reskilling programmes that reach women in the sectors most at risk, not just women who are already in tech. Transparent pay data, mandated and published, so that AI-skill premiums cannot quietly accumulate along gender lines without scrutiny. Investment in girls’ STEM education that goes beyond initiatives and becomes structural. And, perhaps most urgently, diverse AI development teams, because an AI built without women’s input will serve women last.
The window for getting this right is not infinite. Automation doesn’t pause for policy cycles. The jobs at risk are being automated now. The skills premium is accruing now. The gender gap, if unaddressed, will not merely persist. It will be encoded into systems that operate at scale, at speed, without apology.
We built the gender pay gap over centuries of accumulated assumption. AI could replicate the whole thing in a decade, at machine speed, if we let it.
International Women’s Day is a useful moment to pause. To name the problem clearly. To refuse the comfortable narrative that progress is inevitable and things are moving in the right direction on their own.
They are not moving on their own. They never have.
The algorithm doesn’t care that it’s March 8th. But we should.
Happy International Women’s Day. Now let’s get to work.
Author: Deborah Holmwood, Client Change & Transformation Partner.
Follow our LinkedIn company page to stay up to date with all our new blogs!
