The gender gap in AI adoption isn't a bug—it's a feature worth studying before we automate ourselves into a corner.
The Summary
- Girls Who Code CEO Tarika Barrett says student skepticism about AI—especially among women—reflects legitimate concerns about energy use, errors, and tech concentration
- Women are disproportionately wary of AI tools, creating a measurable usage gap along gender lines
- Barrett argues this hesitation should guide how we build AI, not be dismissed as technophobia
The Signal
Girls Who Code has spent a decade trying to close the gender gap in computer science. Now AI is redrawing the map while they're still walking it. The irony is sharp: just as more women enter tech, the industry is automating away entry-level coding jobs and graduating classes are booing speakers who mention LLMs.
Barrett's insight cuts against the usual "move fast and break things" gospel. She's not trying to sell women on AI or shame them into adoption. Instead, she's asking what it means that one demographic is systematically more skeptical about technology that everyone agrees will reshape work.
"We have a deeply held belief that the quality of our technology, the future of AI in particular, depends on who's going to build it."
The usage gap is real and measurable:
- Women report higher anxiety about AI errors and hallucinations
- Environmental concerns about energy consumption skew female
- Wariness about concentrating power in tech billionaires' hands tracks gender lines
This isn't about capability. It's about what you notice when you're not already sold on the pitch. Women in tech are reading the same headlines as everyone else about job displacement, energy consumption, and who profits. They're just less likely to handwave those concerns away because the technology is cool.
The timing matters. Computer science enrollment is finally recovering from its post-dotcom crash. Then AI shows up and companies quietly reduce headcount for junior developers—the exact roles that were supposed to be the pipeline. Tech executives openly discuss how many fewer programmers they'll need. Students notice. Women, already underrepresented and under-supported in tech, notice harder.
Barrett is framing this skepticism as a design input, not a barrier to overcome. If the people most concerned about AI's downsides opt out of building it, we get AI designed exclusively by people who think the concerns are overblown. That's not a diversity problem—it's a quality control problem.
The shift from "learning to code" to "learning to work with AI" changes Girls Who Code's entire value proposition. Writing Python was a skill with clear career paths. Prompt engineering and AI oversight are murkier. The jobs are less defined. The moats are narrower. And the question of whether you're building your replacement is no longer theoretical.
The Implication
If skepticism tracks gender, and gender diversity actually matters for AI safety and utility, then the industry has a measurement problem. Companies say they want diverse teams building AI. But if the current trajectory selects for people who are comfortable moving fast despite concerns about errors, energy, and concentration of power, diversity efforts will keep hitting the same wall.
For people trying to navigate this: Barrett's right that concerns are signal, not noise. If you're worried about AI hallucinations in healthcare or education, that worry should shape what you build, not disqualify you from building it. The Web4 infrastructure layer needs people who ask hard questions about energy costs and error rates, not just people optimized for shipping fast.