Closing the GenAI Adoption Gap
My thoughts keep circling back to a set of numbers that drove our GenAI drop-ins with our Gender network last week. They were the reason for the sessions and they deserve a wider airing.
The gaps we (sort-of) know
Gender. A Harvard Business School meta-analysis of 18 datasets puts women at 22% lower odds of GenAI use, while McKinsey's 2025 Global Survey reports a 21–31 pp gap in workplace uptake, depending on region [1][2].
Age. Pew's March 2025 poll shows 58% of under-30s have used ChatGPT versus 10% of over-65s [3].
Ethnicity. Among teenagers, Pew finds Black and Hispanic students 31% each using GenAI for schoolwork against 22% of White peers; adult figures are fragmentary and swing by country [4].
Accessibility. Ontario's higher-ed survey reports 47% weekly use among students with disabilities, but nationally representative numbers for workers are still absent. Other research from the University of Washington/NASCIO 2024 reports interface barriers [5].
The message?
Adoption gaps look a lot like the early digital divide, only opening far faster; GenAI usage has hit roughly 35–40% of adults in under two years, about double the speed of the early internet. GenAI app use today is a predictor of better and broader adoption of future AI tools by individuals. So, if we don't intervene now, tomorrow's productivity engine could hard-wire today's inequities.
What we did and what it taught us
We showed people how their colleagues are already benefiting from using AI in the workplace — and at home. Across informal drop-in sessions, co-hosted with our Gender Equity Network and open to all, female colleagues from diverse backgrounds demonstrated how they:
- turn meeting transcripts into actionable plans,
- translate company talking points between English and Spanish in seconds,
- draft an email from bullet points and re-word it in a friendlier tone in a jiffy,
- summarise dense research before the kettle's boiled, and
- even plan healthy mid-week dinners (a lifesaver for working parents).
Attendance spiked, questions flowed and — most importantly — confidence visibly grew and I saw that when technology is shown in context, barriers turn into curiosity.
Three takeaways
Confidence follows relevance. People don't need a white paper; they need a Wednesday-afternoon 5-minute time saver.
Visible sponsorship matters. When senior leaders fumble through a live demo, everyone else gains permission to experiment.
Trusted, ethical AI is key. These things matter to people, especially those burned by past data bias — prove that this wave is different: invite them in, show the responsible AI in practice, and back every promise with visible follow-through.
We're still in the dark on the impact of intersectionality on adoption rates. No solid figures exist for older women of colour, or disabled gig-workers. Without that lens, "inclusive" roll-outs risk missing the mark.
So is there a call to action here? Absolutely!
Organisations:
- Create show and tell, and safe-to-try sessions. Keep them small, practical — real day-to-day tasks and co-hosted by relatable colleagues.
- Show your workings. Leaders who use GenAI openly and share their journey de-risk the learning curve for everyone else.
- Bake in accessibility. Alt-text for images, plain language prompts, larger text windows — make them procurement non-negotiables.
Researchers & funders:
- Plug the data holes. We need GenAI attitude, adoption and usage data from labour-force and digital-skills research, that include diversity dimensions, particularly for people with disabilities and intersectionally — and especially outside North America.
GenAI should be a great leveller — but only if we measure who's boarding the train and who's still on the platform. So run small, peer-led sessions, make accessibility a non-negotiable, and let leaders share their own GenAI wins and wobbles, because that's how everyone learns, quickly and safely. Let's bridge the gap now, before it hardens into new structural walls.