Your AI adoption strategy is missing its most critical component

The data on women and AI is alarming but the conversation we are having about it is completely wrong.

Here are some numbers that should stop every technology leader in their tracks:  women are adopting generative AI at roughly 25% lower rates than men. Across 18 studies, more than 140,000 workers, a Federal Reserve survey found half of men used AI tools in the past year compared with just a third of women. The OECD found female workers were 20 percentage points less likely to report using ChatGPT than male colleagues in the same role.

The standard organisational response? More training. More tutorials. More access programmes.

It is not working. Because the problem was never a logistics problem.

We keep solving the wrong problem

The dominant narrative goes like this: women lack AI skills, so we must upskill them. It is a clean, linear, logical framing. It is also dangerously incomplete.

The Skillsoft Women in Tech Report found that 63% of women using AI at work reported a lack of adequate training, yet AI was simultaneously the top subject they wanted to learn. The desire and motivation are there,  so why are adoption rates still lagging?

Because access to a tool and feeling psychologically safe enough to use it are two entirely different things.

Harvard Business School researchers tested this directly. They gave women entrepreneurs explicit access to ChatGPT and removed the knowledge barrier entirely. Women were still 13% less likely to try the tool. Access alone does not close the gap. Something else is at play and that something is emotional.

The missing infrastructure

In my work on the emotional dimensions of organisational change, I see this pattern constantly. When a new technology or transformation is introduced, leaders focus on the what – the platform, the process, the timeline. They build the logical case. They run the training sessions. They tick the adoption boxes.

What they almost never build is the emotional infrastructure – the psychological safety to experiment without fear of failure, the sense of belonging in the change narrative, and the felt experience of being seen in the transition rather than managed through it.

For women navigating workplace AI adoption, this gap is especially costly. Research consistently shows that women have higher concerns around AI ethics, trust, and data privacy. A 2025 study found an 18-point confidence gap in AI skills, with young women reporting significantly lower confidence than their male peers (56% versus 74%). Deloitte UK’s research found that 28% of women were using generative AI compared to 43% of men – not because women are less capable, but because the conditions that enable confident adoption have not been created for them.

Only 38% of junior women in technical roles recognise reskilling in AI as critical to their future job success, compared to 53% of their male counterparts. BCG’s research suggests one key reason: junior women are not equally represented in the conversations and pilot programmes where AI strategy is formed. They are not in the room where the story is being written.

That is not a skills deficit. That is an exclusion from meaning-making and no training programme fixes that.

The stakes are commercial, not just ethical

Let us be direct about what happens if this does not change. Federal Reserve economists have warned this gap could amplify the existing gender pay gap, structurally, at scale. Harvard researchers flag a feedback loop: when women are underrepresented among AI users, AI systems train on male-dominant data, producing outputs less attuned to women’s needs, which further discourages adoption. The gap compounds itself.

For startups, this is not just an equity issue, it is a competitive one. The Oliver Wyman Forum surveyed 25,000 working adults globally and found that 98% believe they need to be upskilled due to AI disruption. Yet business leaders estimated that only 40% of their workforce actually needed such training. That disconnect creates anxiety. Anxious people do not experiment, they do not explore, they do not try new things. People who feel unseen, valued or heard in a change process do not engage.

When your most ethically-attuned people are sitting this technology out, you are not just losing productivity. You are losing the perspective that makes AI adoption more considered, more human, and less likely to cause damage down the line.

What actually works

Create low-stakes experimentation spaces. Sandboxed AI projects and no-judgement peer learning groups are adoption accelerators, not nice-to-haves.

See Also

Mimica raises $26.2M to make AI agents actually work in the real world

Put women in strategy rooms, not just training rooms. If women are absent from AI pilot conversations, you are signalling, perhaps unintentionally, that this is not for them.

Surface senior women as visible AI users. Senior women already lead their male peers in AI adoption by 12–16%. That confidence exists, so make it visible.

Treat ethics concerns as an asset. Women’s higher concerns around AI ethics and data privacy are not obstacles, they are the critical thinking your governance needs. Create space for them.

Change is an emotional experience

Every major organisational change I have worked through has confirmed the same truth: the logical case for change rarely determines whether people embrace it. People adopt new things when they feel safe enough to try, valued enough to matter and connected enough to a shared future that the risk feels worth taking.

The companies currently winning at AI adoption are not just those with the most sophisticated tools or the most generous training budgets. They are the ones creating conditions where every person in the organisation, including and especially women, experiences the change as something happening with them, not to them.

The data is clear. The gender gap in AI adoption is real, it is consequential and the standard organisational response is not working. What is missing is not more content. It is more humanity in how we lead change.

If your AI strategy does not have an emotional architecture, it does not have a complete strategy.