ada könig
← all pieces
Urtext · 2026.11.11

Does AI love weapons?

Artificial intelligence will never be ethical as long as women are kept out of the rooms where it’s decided. It’s a more pragmatic claim than it first sounds, even to me.

This isn’t the first time I’ve written about the link between artificial intelligence and military applications, with everything that follows from it. I’m not alone in thinking this: yesterday I saw a post on X where Alex Karp, the ceo of Palantir, described the military as an environment in which the technology can develop in ways that are often constrained elsewhere. Today, while taking a break to look for new material for the site, I came across an interview with Darya Majidi, an Italian-Iranian entrepreneur and president of UN Women Italy (the Italian chapter of UN Women), published by the Italian newspaper Corriere della Sera, in which she argues that artificial intelligence will never be ethical as long as women are kept out of the rooms where it’s decided. Majidi came to Italy as a child after the theocratic regime took power in Tehran, and later earned a computer science degree in Pisa with a thesis on AI, in years when most people still thought of it as science fiction. Her thesis is simple, and at first glance almost reassuring: computing’s bro culture (male, nerdy, socially awkward) is a Western invention, because in the Middle East and Asia girls choose computer science far more than they do here, driven by a pragmatic calculation about jobs and independence.

The pattern she points to is real (in several Gulf countries women earn computer science degrees at higher rates than in Europe), but the research on it reads almost the opposite way. It’s called the gender-equality paradox, after Stoet and Geary’s work. One reading of it holds that it’s where freedom to choose is lower, not higher, that computing stays one of the few paths widely recognised as a secure career. Worth keeping in mind, but it isn’t where the real contest Majidi cares about gets decided, and that’s where I’d rather look.

That contest gets decided closer to home, and has far less to do with geography than the interview suggests. Majidi says as much herself, almost in passing: women in AI are mostly hired to do labelling, tagging the data that trains the models (a “blue-collar role,” she calls it, and I agree). The numbers back her instinct with a precision that stings more than the line itself: according to LinkedIn data on 2025 hiring in the sector, data labelling is the only AI role in which women account for more than half of all workers. From there it drops fast: AI engineer, 24 percent women; VP of AI, 21 percent; Head of AI, just 20 percent, with men holding 80 percent of the seats. This isn’t about who can code. It’s about who gets to decide what gets built.

And power, in AI today, is measured in capital concentrated in very few hands (the subject I keep coming back to). An analysis of the most heavily funded California AI companies, published last year, found that women hold 15 percent of board seats overall, and that more than four in ten of these companies have no women on their board at all. The part that actually surprised me: among companies funded above $100 million, the share of women on the board nearly doubles, to 19 percent, up from 9 percent at the $50-99 million tier, and all-male boards drop from 62 percent to 32 percent over that same range. More capital does bring in more women. It just stops well short of bringing in enough: even at the top of the funding scale, fewer than one in five board seats are held by women. It isn’t a one-off for this industry: it’s the same bro world Majidi describes, except here the bro isn’t the awkward nerd of her story, it’s the boardroom still mostly closed even once the money arrives.

The same week I’m writing this, Der Spiegel reported that a German startup called Helsing is launching an AI warfare platform to compete with Palantir, and that the German armed forces are already testing it. Its three founders are men. Its chairman and largest shareholder is Daniel Ek, the Spotify founder, whose own artists have spent years asking him to stop funding weapons with the company’s money. Europe building its own seat at this table, instead of renting one from Palantir, is not a small thing. But it’s worth noticing that the seat still comes with the same furniture.

There’s one thing, though, that fairness won’t let me skip. She fled a regime that stripped women of freedom in a brutal, explicit way, then studied and built a career in a country, Italy, that keeps computing away from girls in a subtler way (she cites it herself: 5 percent of graduate enrolment at Pisa, where she studied). Her own biography undercuts the clean West-versus-rest map she draws: she isn’t a case of pragmatic Eastern freedom or Western cage, she’s proof that the line she draws doesn’t fall where she thinks it does.

There’s a point in this story about tech’s bro culture in which I recognise something of myself. I have to make peace with it in the first person, because I’m a product of that world too. At university I knew guys who were brilliant with compilers and hopeless with girls: insecure, outside every social hierarchy except the one measured by exam grades. That wound doesn’t heal once real power arrives: it gets cashed in. I’ve watched more than one brilliant career built around an unpaid bill from adolescence, settled now that there’s finally leverage to settle it. It isn’t a deliberate project, a “narrative built by men for men” as Majidi puts it: it’s cruder than that, and harder to dismantle. It’s payback.

And payback, before it becomes real power, trains somewhere first. Those classmates spent their nights playing first-person shooters. Not because games necessarily make people violent, but because this was one of the few places where the match could actually be won: target, crosshair, eliminate, the score climbing. That was where a boy with no status could rehearse the vocabulary of total dominance, long before he had any real power to exercise it.

I don’t want to turn this into a settled thesis, but I wonder what would change if the people funding and designing AI’s military applications hadn’t come, in such overwhelming proportion, from that same training ground of rehearsed dominance, and from the same all-male boardroom deciding where the capital goes. Someone might call it friction, the resistance that elsewhere slows the technology down: it’s worth asking whether some of that friction comes from people who, in that room, never had to win that particular match. Not because women would be naturally more peaceful: that’s the reassuring argument Majidi herself makes, and it convinces me no more than the narrative she’s trying to take apart. But because women were never offered the same space in which to rehearse that particular wound, or the same power to turn it into leverage later in life. If that’s the mechanism, then it isn’t a female eye that’s missing from the tables of military AI. It’s something else, harder to say out loud: maybe what’s missing is someone who never had to win that match.

PS: if you’ve read this far, I’d suggest not a book but a writer. There’s a writer I love, whose work has always moved me and given me things to think about: she’s a woman, she writes science fiction, and her stories go deep into questions of gender. It’s Ursula K. Le Guin. Look her up, and pick your own book to start with.