Analysis
The data on AI and gender points in two directions at once, and both are bad. Women make up only 29% of AI-skilled workers globally, per Randstad data cited by Bloomberg, even as AI-adjacent roles have become some of the fastest-growing, highest-paying jobs available. At the same time, the World Economic Forum estimates women will make up 57% of Americans likely to see their jobs eliminated or significantly changed by AI-driven automation. Women are underrepresented in the jobs AI is creating and overrepresented among the jobs AI is displacing -- that is not a coincidence of two unrelated statistics, it is the same structural pattern showing up on both sides of the ledger.
I don't think this is primarily a pipeline problem, and I think treating it as one is why it hasn't improved. The access gap is the more damning number: only 35% of women report being offered access to AI tools in their current roles, versus 41% of men -- that's not a skills-training failure, that's a distribution failure inside companies that are choosing who gets AI tooling and who doesn't. Women also report 14.3% lower confidence that their AI training adequately prepared them, which reads less like a competence gap and more like a training-investment gap: if you're not given access, you can't build confidence using the tools.
“41%) is happening today, inside companies that could close it with a policy change, not a decade-long pipeline fix -- and that's the lever worth pulling first.”
The part that should worry every VC reading this is the C-suite data point buried in the same reporting: the underrepresentation spans from leadership to rank-and-file roles inside AI companies themselves, per LinkedIn data. That means the imbalance isn't just showing up in who gets hired for AI jobs at other companies -- it's baked into the founding and leadership layer of the industry building the technology in the first place, which shapes product decisions, hiring practices and who gets access to capital to start the next generation of AI companies.
Room for disagreement: it's possible some of this gap reflects pre-existing occupational segregation rather than new AI-era discrimination -- women have historically been underrepresented in computer science and engineering roles that feed directly into AI-skilled positions, and closing a 29% figure requires closing a decades-old pipeline gap that predates this specific technology cycle by years. If that's the primary driver, the fix is upstream in STEM education and hiring pipelines, not in this quarter's AI tool rollout decisions, and treating it as a same-year fixable access problem may be too optimistic. Either way, the access-gap number (35% vs. 41%) is happening today, inside companies that could close it with a policy change, not a decade-long pipeline fix -- and that's the lever worth pulling first.
