Why Are There So Few Women Leading AI Research?

Here's the answer up front: it's not that women aren't interested in AI, and it's not that they're not qualified. A new UNESCO study that tracked women through AI classrooms, research labs, startups, and investor pitch meetings across six South Asian countries found the pipeline doesn't break in one obvious place. It leaks at almost every step, and the leak gets bigger the closer you get to power. By the time you reach who actually leads AI research and who actually gets funded to build it, women have nearly disappeared from the room.

What did the new UNESCO study on women and AI actually find?

The Outlook Study on Artificial Intelligence and Gender in South Asia, published by UNESCO's Women for Ethical AI South Asia Chapter, is the first regional, data-driven look at where women stand across Bangladesh, Bhutan, India, the Maldives, Nepal, and Sri Lanka. It starts in the classroom: women make up only about a third of STEM students region-wide. In the Maldives, the gap gets almost surreal — women are 78% of university students, yet only about 2% of them enroll in science and technology programs.

The pipeline keeps leaking after school. Women are named authors on roughly 72% of AI research papers published across the region between 2015 and 2025 — they're doing the work. But they hold the lead or corresponding-author position, the credit line that actually counts for hiring, grants, and promotion, on only about 26% of those papers. Dr. Bhavani Rao R, the UNESCO Chair on Gender Equality and Women's Empowerment who co-chairs the study, has made the point plainly in her own writing on the report: the problem was never a lack of ability in women. It's that the systems around them, from classrooms to publication norms, were never designed with them in mind.

Where the pipeline leaks hardest

Follow the trail past research and into industry, and the leak turns into a flood. Fewer than 1 in 10 AI startups in India have a woman founder. Even more strikingly, fewer than 1% of India's roughly 10,000 angel investors are women — so the rare founder who makes it past the classroom and the lab often still can't get in the room where the money gets decided. It's the same story this blog keeps finding wherever we look at who actually builds your AI: the numbers thin out fastest at exactly the levels with the most say over what gets built and who it's built for. If you want the fuller global picture behind that pattern, we've also broken down the math of who's missing from AI research and looked at whether women are actually getting funded to build AI elsewhere on this blog — the South Asia numbers are a sharper, more specific version of a gap that shows up everywhere researchers have thought to measure it.

Why this matters even if you never write a line of code

You don't need to be an engineer for this to touch your life. AI systems built by rooms with almost no women in the lead don't just under-hire women — they carry that same blind spot into the products themselves: health apps that miss symptoms that present differently in women's bodies, hiring tools trained on data that already favors men, chatbots that default to a male-coded idea of "normal." The gap in the boardroom becomes the gap in the app on your phone. That's the real reason this report matters to you even if you've never touched a research paper or a pitch deck: someone else's absence from the room shapes what shows up on your screen.

One thing to try today: the next time an AI tool gets something about you wrong — flags your resume oddly, mislabels a symptom, assumes your name means something it doesn't — don't just close the tab. Leave the feedback, wherever the app lets you. That complaint is a small, real data point going into exactly the kind of gap this report is naming. It costs you thirty seconds and it's one of the few ways someone outside the room gets to push back on what's built inside it.

Quote to sit with

"We cannot all succeed when half of us are held back." — Malala Yousafzai

💌 Sarah