The AI Abstainers: Why Women May Be Right to Opt Out
Laetitia@Work #96
Hi everyone,
By using AI less than men, are women opting out of the future or are they in fact seeing it more clearly? Are they more risk-averse, as is often said, or rather more risk-aware?
The gender gap in AI use seems to be real. The numbers are striking enough to demand an explanation. Women are somewhere between 20 and 25 percent less likely than men to use generative AI tools, according to a Harvard Business School meta-analysis. Another recent study found that only 14.7% of women reported using generative AI at least weekly, compared to 20% of men. Meanwhile, the jobs most vulnerable to AI-driven automation are disproportionately held by women. The Brookings Institute found that 86% of the workers in roles most likely to be disrupted are women.
So women are slower to adopt the very technology that threatens them most. Is it a paradox? Not really. The original Luddites — the 19th-century British textile workers who smashed the machines replacing them — understood exactly that the new machines would concentrate power in the hands of owners and replace their craft with cheaper output. They didn’t smash looms out of confusion and stupidity. Yet when we call someone a Luddite today, we use it as an insult. We probably shouldn’t.
Women who hesitate to embrace generative AI may be operating with similar clarity. If your job — administrative assistant, receptionist, legal clerk, office manager — sits squarely in the crosshairs of automation, enthusiastic AI adoption is not obviously in your interest. Training the system that replaces you, accelerating its capabilities, smoothing its integration into your workplace: these are acts that primarily benefit your employer, not you. The question “why aren’t women using AI more?” should be replaced with “why would they?”
Women’s hesitancy may be attributable to discernment. A case should be made for “fierce ambivalence” (as was made by Mara Bolis in the Stanford Social Innovation Review). Technology is not neutral, and who benefits from its adoption is a political question, not a technical one. The reasons behind the gap are more interesting — and more flattering to women — than the usual framing suggests. Here is my analysis of the AI-use gender gap👇.
The Penalty for Trying
One underappreciated reason women hesitate is that they have more to lose from being seen using AI. A study published in the Harvard Business Review found that when male and female engineers submitted identical AI-assisted code for review, women received competence ratings 13% lower, compared to just 6% lower for men. The same output produced with the same tool leads to different consequences.
So women’s hesitancy may be a calculated read of a biased environment. In ambiguous situations — and “did AI write this?” is inherently ambiguous — evaluators fall back on stereotypes. Technology is still culturally coded as male, so when a woman uses AI, observers tend to assume the tool did all the thinking. It’s like a digital Matilda effect, where the tool gets the credit and the woman gets the doubt. By contrast, when a man uses the same tool, he is credited with the strategic intelligence to deploy it well. Women who have spent years navigating exactly this kind of double standard anticipate it here.
👉 Also read Why women over 50 are the future of work in the age of AI. Laetitia@Work #95
A System Trained Without Them
There is a second, less-discussed reason that doesn’t get enough attention: the outputs of generative AI are frequently less useful to women because the systems were built largely without them. Women are underrepresented in the engineering teams that designed these models, in the policy discussions that shaped their deployment, and crucially in the training data that gives them their sense of the world.
The result is a technology that can feel subtly foreign or hostile. For example, a German study found that AI chatbots advised women to ask for significantly lower salaries than identically profiled men — with gaps of up to $120,000 a year for the same role. Outputs on professional scenarios, historical figures, and even basic personal finance questions tend to reflect a perspective that is implicitly male. Women notice the hallucinated citations, the lazy generalizations and the gender-blind advice.
The Burden of AI Slop
Women may use AI less partly also because they are the ones who most clearly see where it fails. They are already the ones cleaning up after it.
In many workplaces, the unglamorous work of reviewing, correcting, and polishing AI-generated content falls disproportionately on administrative and support staff, roles overwhelmingly held by women. The people who rewrite the slop don’t need a study to tell her where the productivity gains are, or aren’t. Greater AI literacy may in this case lower adoption. When you actually understand what the tools do and do not do, enthusiasm tends to become more measured.
Moral Weight and Environmental Anxiety
The aformentioned research also found that among users who expressed concern about AI’s environmental impact, the gender gap in usage jumped to 9.3 percentage points. Among those concerned about AI’s mental health implications, it widened to 16.8 points. And among older women worried about climate, the gap reached nearly 18 points.
Researchers describe this as alignment with “greater social compassion and moral sensitivity among women” — though a more precise reading is that women have been socialized to weigh collective consequences more heavily than individual advantage. No, I don’t believe that women are genetically wired to be superior moral beings! But whatever the origin, the behavioral pattern is consistent: women are statistically more likely to act on their moral concerns. According to IEA estimates, a ChatGPT query uses nearly ten times as much electricity as a Google search, and global AI-related water demand is projected to skyrocket. That a growing number of women have made a deliberate choice to abstain, connecting the carbon footprint of generative AI to real-world environmental consequences, is a rational response.
The Contrarian Case for Abstaining
What if the women opting out are not falling behind, but preserving something the enthusiastic adopters are gradually losing?
There is an emerging body of concern about what habitual AI delegation does to cognitive skills like writing, reasoning, and problem-solving. The same tools that offer efficiency gains may, with prolonged use, erode the capacities that made the user valuable in the first place. If women are, on average, engaging more selectively and critically with these tools, they may be better positioned to maintain the deep expertise that AI currently cannot fully replicate. In a world saturated with AI-assisted mediocrity, genuine human judgment and craft may become more and more valuable.
Thus women may be like the canary in the coal mine. Ask what the canary knows that you don’t.
What Needs to Change
None of this means the gap is fine. The career risks are real. Major employers are beginning to tie AI adoption to performance reviews and promotions, and women who remain on the outside of that shift will face compounding disadvantages. The answer is to use these tools deliberately and critically, while holding their developers to the highest possible standards on transparency, bias, and environmental accountability.
Women’s lower adoption of generative AI is not due to irrational fear but to informed “risk awareness” about bias, opacity, and potential harm, yet opting out risks amplifying gender inequality in careers and influence. The article argues for “fierce ambivalence”—engaging with AI while demanding accountability—so women can shape its development and ensure more equitable outcomes.
Women’s skepticism is data. The industry should be reading it, not waiting for women to get over it.
💡For Nouveau Départ, I wrote several new articles (in French):
→ Subscribe to receive my future podcasts and articles directly in your inbox!