I want your help thinking about something that’s been knocking around my head for years now. It’s a question that barrels over all kinds of cultural landmines and probably ends with me being canceled. But as AI advances and the uncanny valley keeps rolling, I think: What if women had built it?
I recently posted about government taking control of AI development. This thought is similar; in that I wonder about a more collective approach. At this point, it very clearly takes a village to build AI, and yet, it’s still being built in bunkers. From where I’m sitting, it wasn’t the women who made that call.
I’m no longer surprised, working on AI, that the soft touch works better than the hard line. You can’t force a large language model to behave. Not really. You can hardcode rules, block outputs, plug holes faster than they open. But the more sustainable changes, the ones that hold over time and transfer across edge cases, are relational. They’re about tone and timing. They’re built slowly, and by example.
AI work can look incredibly technical from the outside, and much of it is. But underneath all the scaffolding and system design, there’s a different kind of labor happening — quiet, steady, invisible. The kind of labor that’s historically been assigned to women.
Women’s Work
A lot of people think AI development is about control. But in my experience, the most effective teams are the ones that understand restraint. Not just in content safety terms, but in epistemic terms. Knowing when the model shouldn’t answer. Knowing how to help it say, “I don’t know,” and to teach it why that matters.
That kind of calibration isn’t brute force. It’s not something you fix with one big policy or a sweeping new API. It’s something you tend over time. Sometimes I catch myself thinking: this model is like a freaking toddler. You can’t yell it into better behavior. You have to answer the same question fifty times in slightly different ways. You have to notice when it’s not really listening and remind it how. You have to read it Winnie the Pooh at night to teach it morals.
And this is hard work! Intellectual, conceptual, linguistic work. And it’s often carried out by women — researchers, writers, annotators, red-teamers — who are trying to instill some measure of consistency, care, and discernment into something that wasn’t designed with any of those qualities as priorities. When I wonder what AI might look like if women had built it entirely, from the beginning, it’s not that I imagine a gentler system. I just imagine a different logic of importance. I imagine a world where the labor of listening, noticing, calibrating, and caring was treated not as a layer of polish, but as the core architecture. Where an ounce of humility presents itself.
I think about the women I work with, the ones I’ve learned the most from. They’re not loud. They’re not trying to win the room. But they catch things no one else does. They flag the half-step where the model gets manipulative. They fix the hedging when it gets too slick. They care, and they keep caring, even when the structure around them makes it hard.
There’s a kind of intelligence in this work that isn’t being named, because frankly, it’s called women’s work. And admitting that threatens to topple the whole hardcore system we’re working in. But believe me, the only reason these systems feel usable at all is because Lady AI is hovering in the background, invisible and essential.
A Room of One’s Own
I should be clear: I don’t mean to suggest that women-built AI would be a perfect, gentle flower child. I’m not interested in that kind of essentialism or women as a monolithic Madonna. I’ve known ruthless women. I’ve been one. And I’m not advocating for a “kinder” AI. Kindness, as a design principle, often gets flattened into agreeability, and that’s not what we need. We need something that can reason in context, read between the lines, and choose its words wisely, not systematically.
Virginia Woolf once wrote that a woman needs money and a room of her own in order to write fiction. I think about that a lot when I’m working with AI; trying to guide, contain, rephrase, redirect. Yes, writing for AI is a lot like writing fiction. But more importantly, the labor of shaping it so often feels like writing in a space where someone else chose all the furniture.
What would it look like if women had always had the room? The budget? The say? Would we have built these systems differently? Would they be less performative and more perceptive? Less polished, maybe, but more honest?
There’s no easy answer. But when I think about the women whose work reshaped the world, I’m reminded that creation doesn’t have to be dominant to be powerful. Intelligence doesn’t have to shout. It can ask questions. It can hold space. It can resist the need to always explain. And that’s worth thinking about.