Hello Neuron readers! We wanted to take a little time to dig into something that should probably get more attention, but really doesn't: why is the AI industry so heavily male dominant? Fair warning: we're about to get into a bit of gender discrimination stuff. If that's not your thing, you should probably read this and learn something! Oh, also, no need to write in with your hot takes on why we should "stick to the tech" or "not get into politics" (gender is not political, it's biological?)... you need not write us, because we've already written a companion piece to this to answer all of your complaints. Enjoy!
AI Has a Man Problem
When Sam Altman was reinstated as OpenAI's CEO in November 2023, the newly reconstituted board that would oversee the most consequential AI company on Earth consisted of three people: Bret Taylor, Larry Summers, and Adam D'Angelo. Three men. Ashley Mayer, CEO of venture firm Coalition Operators, captured the mood: "I'm thrilled for OpenAI employees that Sam is back, but it feels very 2023 that our happy ending is three white men on a board charged with ensuring AI benefits all of humanity."
Helen Toner and Tasha McCauley, the two women who'd been on the original board, were gone. Mira Murati, the female CTO who'd been caught between warring factions, would leave within the year. The episode was framed as a corporate drama about power and governance. But zoom out and a different story comes into focus: the most important technology of our generation is being built, funded, and led almost entirely by men. And as NPR framed it just this week, this is the future "tech bros" want.
The Numbers Are Worse Than You Think
Start with the workforce. Women hold roughly 22% of AI roles globally and just 18% of AI researcher positions. Only 14% of AI research papers have a female first author. At the senior level, it gets bleaker: just 10% of AI company CEOs are women, and women occupy less than 14% of senior executive roles in the field.
Now look at who's funding the revolution. In 2024, of the $289 billion in global venture capital deployed, 83.6% went to all-male founding teams. All-female teams got 2.3%. That number actually fell to just 1% in the US market, down from an already pathetic 2% the year before. Women founders get interrupted nearly five times more often than men in pitch meetings, for what that's worth.
Here's a headline that looks like progress: female-founded startups raised a record $73.6 billion in 2025. But two-thirds of every VC dollar in that number went to AI companies, and Anthropic and Scale AI alone captured $30+ billion of it, over 40% of the total. Strip out a couple of outliers and the underlying math barely budges. Less than 3% of all venture capital still goes to women overall.
Then there's the public face of AI. Picture the people you associate with artificial intelligence. Sam Altman. Elon Musk. Jensen Huang. Demis Hassabis. Satya Nadella. Andrej Karpathy. Dario Amodei (did you know he has a sister, Daniela, who is President of Anthropic?). The AI podcasters (Lex Fridman, nearly 5 million YouTube subscribers; Dwarkesh Patel, over 1.2 million followers). The conference keynote speakers. The men testifying before Congress. It's basically all a wall of men, and it's not subtle.
This matters beyond symbolism. A March 2026 CNBC/SurveyMonkey survey found that 78% of men use AI for work versus 73% of women, 33% of men use it daily versus 27% of women, and 69% of men view AI as a "valuable collaborator" while only 61% of women do. Half of women said using AI at work feels like "cheating."
When McKinsey surveyed workplaces in 2025, they found that only 21% of entry-level women were being encouraged by managers to use AI tools, compared to 33% of their male peers. Women in entry-level product development roles declined 17% from 2024 to 2025. One in six companies had cut diversity staff or resources. Across Europe, women in core tech roles fell from 22% to 19% year-over-year, according to a March 2026 McKinsey analysis. In some places, the gap between women and men is accelerating.
In April 2026, Sheryl Sandberg and Lean In published data showing that among people who do use AI, only 18% of women have been praised for it, versus 27% of men. The gender gap in AI isn't just about who's building it or who's using it. It's also about who gets credit for using it.
Built by Men, Biased Toward Men
When one demographic overwhelmingly designs a technology, the technology reflects that demographic. This isn't theory. It's documented.
Joy Buolamwini, a researcher at MIT, discovered she had to put on a white mask for facial recognition software to detect her face. Her research with Timnit Gebru found that commercial facial recognition systems had error rates below 1% for light-skinned men but as high as 34.7% for darker-skinned women. The systems worked brilliantly for people who looked like the people who built them.
Amazon trained an AI hiring tool on ten years of resume data. Because tech has been male-dominated, the algorithm learned to penalize resumes that included the word "women's", like "women's chess club captain." It wasn't malicious. It was a machine learning exactly what the data taught it: that men get hired here.
Safiya Noble, a UCLA professor and MacArthur Fellow, found that searching "Black girls" returned predominantly pornographic results. The algorithms weren't broken; they were reflecting the priorities and blind spots of who built them.
Healthcare AI systems trained on datasets skewed toward male patients misdiagnose symptoms in women, because they default to male symptom presentation. Voice recognition systems, as Meredith Whittaker of the AI Now Institute has pointed out, work less reliably with higher-pitched voices (more common in women), because the engineers testing them were mostly men with lower-pitched voices.
A 2024 UNESCO study on large language models found they consistently portray women in domestic and subservient roles while casting men as executives and leaders. The models aren't inventing these stereotypes from nothing. They're absorbing them from training data curated and filtered by teams that are 78% male.
The bias problem isn't limited to older systems. A 2026 study found that 44% of AI hiring programs show gender bias. The UNDP reported that AI image generators depict 75-100% men when asked to visualize STEM professions. UC Berkeley researchers found that ChatGPT assumes women are 1.6 years younger than men and have less work experience by default. A separate study found that AI labels young women as "fragile" in 56% of cases and is six times more likely to recommend they seek "external validation" than it is for young men.
Then there's the weaponization. In January 2026, Elon Musk's xAI chatbot Grok generated over 4.4 million sexually explicit images within nine days of launch, 1.8 million of which depicted women. By March, three girls had filed a class-action lawsuit against xAI for generating child sexual abuse material. Broadly, 96% of deepfake videos are pornographic, and 99% of those target women. "Nudify" apps that strip clothing from photos of women are proliferating with explicit targeting. Women are now 17 times more likely to experience AI-enabled online abuse than men. The technology isn't neutral. It's being built in a way that makes women less safe.
Kate Crawford, co-founder of the AI Now Institute, has written extensively about how AI systems enforce rigid binary gender categories and fail on non-binary gender recognition. In a March 2026 Fortune profile, she warned that AI agents with access to private data "upend privacy as we have known it." She notes that the "male white elite that dominates Silicon Valley" obsesses over existential AI risk while marginalized groups face algorithmic harms happening right now. The men worrying about whether AI will destroy humanity in 2040 often don't notice it's already discriminating against women in 2026.
The Culture Behind the Numbers
Whittaker makes a point worth sitting with: gender imbalance in AI development and gender bias in AI systems are "two versions of the same problem." One creates the other. And the culture of the AI industry reinforces the cycle.
Consider the OpenAI saga more carefully. Helen Toner, one of the women on the board, described a "toxic culture of lying" under Altman, saying he withheld information, gave inaccurate accounts of safety processes, and was "outright lying to the board." Her co-authoring of a paper that was even mildly critical of OpenAI's practices reportedly contributed to the blowup. When the dust settled, the women were out and the men were back in charge. Qualified women reportedly avoided joining the reconstituted board because of the apparent "boys' club" dynamic.
The pattern kept going. In February 2026, OpenAI fired Ryan Beiermeister, the VP leading its product policy team, on a sexual discrimination allegation she called "absolutely false." Beiermeister had opposed the rollout of ChatGPT's "adult mode" and raised concerns that OpenAI hadn't built strong enough safeguards against child-exploitation content. The timing raised questions about whether the allegation was related to her safety advocacy. In April, Fidji Simo, OpenAI's CEO of Applications, took medical leave, highlighting the pressure on the few women in senior AI roles.
This isn't unique to OpenAI. The broader tech industry shows a documented "bro" culture where women receive 2% of venture capital, face systemic bias in pitch meetings, and encounter networking structures built around male social patterns. The AI boom has, if anything, intensified this: the biggest funding rounds are going to a small circle of companies led by men who know each other, funded by male VCs, and celebrated by a predominantly male tech press.
And the industry is actively retreating from the commitments it had made. OpenAI scrubbed its diversity commitment webpage in early 2025, redirecting the URL to a page about "building dynamic teams" with no mention of the word "diversity." Google scrubbed mentions of "diversity" and "equity" from its responsible AI team pages. Companies across the sector are quietly dropping public diversity hiring goals. 82% of women in tech reported having to prove themselves more than male colleagues; just under half experienced sexism or bias in the past year.
The irony is exquisite. Computing was originally "women's work," as Whittaker points out. Programming, in its early decades, was considered clerical, and women did most of it. As the field gained prestige and money, men took over. The same pattern is playing out with AI: the more powerful and lucrative it becomes, the more exclusively male its leadership grows.
What's Being Lost
Fei-Fei Li, the Stanford researcher who founded AI4ALL (a nonprofit that has reached over 10,000 people across 50 states, 62% of them women or non-binary), warned Congress that without diverse development teams, AI "is doomed to widen the wealth gap even further, make technology even more exclusive, and reinforce biases."
The business case backs her up. Companies with gender-diverse leadership are 27% more profitable and generate 19% more revenue from innovation. Gender-diverse teams make better decisions 73% of the time, compared to 58% for all-male teams. The Peterson Institute for International Economics found that companies with at least 30% women in leadership outperform their peers in profitability by 15%.
And there's evidence that women-led AI companies produce different outcomes. Russell Reynolds Associates found that among the few women-led AI firms, 50% achieved gender parity on their teams and 75% were more gender-diverse than the industry average. Leadership composition ripples downward.
Fortune warned in March 2026 that we risk a "two-tier AI economy" where men reap AI's productivity gains while women are left behind. The numbers back the warning: women use AI at rates 25% lower than men despite being more vulnerable to automation, and only 36% of women feel confident using AI compared to 52% of men. The ILO found that of 6.1 million workers most vulnerable to AI displacement, 86% are women. In 88% of countries, women's jobs are more exposed to generative AI than men's. Women are being replaced by technology they had almost no role in creating.
Organizations like Black Women in Artificial Intelligence, Women in AI Ethics (which publishes an annual list of 100 Brilliant Women in AI Ethics), and Black in AI (co-founded by Timnit Gebru) are building alternative networks. But they're working against the current, not with it.
Some places are managing better. Latvia has nearly half its AI professionals as women. Finland ranks second globally in gender parity. The World Economic Forum notes that the gender gap in AI talent has narrowed in 74 of 75 economies since 2018, with women acquiring AI skills at an accelerating rate. The structural barriers, not women's interest or capability, are the bottleneck.
A Different AI
What would AI look like if it weren't built predominantly by men? This question sounds abstract until you look at the concrete failures that diverse teams would likely catch.
Facial recognition that works equally across skin tones. Hiring algorithms that don't penalize the word "women's." Healthcare AI trained on datasets that represent both sexes equally. Voice assistants that don't default to female personas serving male users (a design choice that UNESCO has called out as reinforcing servility stereotypes). Language models that don't automatically cast women as nurses and men as doctors.
Feminist AI frameworks propose something more fundamental: AI developed with affected communities as co-creators, not just data sources. Design processes that incorporate diverse visual traditions, cultural contexts, and lived experiences. Mandatory bias audits (the EU AI Act, which takes full effect August 2, 2026, requires diverse development teams for high-risk systems). Transparency about who built a system and what data trained it. In Colorado, new 2026 regulations mandate bias audits for AI hiring tools and give job seekers the right to opt out of algorithmic screening for human review.
The momentum is real, if scattered. Barnard College is hosting "Why AI Needs Feminism" later this month. The Feminist AI LAN Party at PyConDE 2026 just wrapped in Darmstadt. An International Women's Day AI Summit in March gathered 150+ senior decision-makers, with ten women sharing strategies they've built at global organizations and government bodies. The conversation about what feminist AI governance means in practice, as UNESCO is exploring, is moving from academic margins to policy tables.
Daniela Amodei, president and co-founder of Anthropic, argued in February 2026 that studying the humanities will be "more important than ever" in the age of AI, pushing back on the narrow technical monoculture that dominates the field. It's a subtle but pointed argument: if the people building AI only think in code, the products will reflect that limitation.
And there's one counterintuitive bright spot in the data. A Fortune/Lean In analysis from April 2026 found that 80% of women leaders surveyed are playing active roles in their organizations' AI governance efforts, with 31% serving as regulators evaluating AI ethics and responsible implementation. Women may be underrepresented in building AI, but they're overrepresented in trying to make sure it doesn't cause harm. That's telling. And it's worth asking whether governance roles, often unglamorous and underfunded, are being delegated to women precisely because the men are too busy chasing the next funding round to bother.
This isn't some utopian dream, btw. It's the minimum standard for a technology that will shape credit decisions, criminal sentencing, medical diagnoses, hiring, and the information environment for billions of people. SAP found that integrating inclusive AI practices in hiring led to a 40% increase in women applicants for tech roles. The tools can be part of the solution if the people building them represent the people using them.
The Real Question
I started this piece expecting to write about representation. I'm ending it thinking about something else: accountability.
The AI industry's gender problem isn't a mystery. The data is abundant, the research is decades deep, and the women raising alarms (Gebru, Buolamwini, Crawford, Noble, Whittaker, Li) have been doing so loudly, with evidence, for years. The industry knows. It's just not structurally incentivized to care. VC money flows to pattern-matched founders. Conferences invite the same speakers. Media profiles the same faces. The cycle reinforces itself.
The ILO reported in 2025 that women are three times more likely than men to lose their jobs to AI. Nearly half of women in Canada expressed concern about negative consequences of AI use at work. Women are 11% more likely than men to perceive AI's risks as outweighing its benefits. They're not wrong to be skeptical. Women are underrepresented in building the technology, underserved by its outputs, targeted by its weaponization through deepfakes, and disproportionately displaced by its deployment. That's not a gap. That's a system working exactly as designed by the people who designed it.
At the current pace, the World Economic Forum estimates it will take 123 years to reach global gender parity. AI could either accelerate that timeline or extend it. Right now, the smart money would bet on the latter.
So the evidence on AI having a male problem is overwhelming. The question is whether the industry will fix it before it encodes male-dominated perspectives so deeply into the infrastructure of daily life that they become invisible... which is, of course, exactly what's already happening. So I guess it's actually will anyone do anything about it before it gets worse?
Here's the takeaway benefit everyone should walk away from this remembering: Closing the global gender gap, according to economists, could boost the world economy by $7 trillion. That's a lot of money being left on the table, y'all. Imagine what a world that's $7 trillion richer would look like. In fairness, perhaps one aspect of a world of material abundance brought on by AGI could be just that it leads to women actually getting a greater share of that abundance. After all, if there's no "pay" from work, then there's no pay gap, right? But in all seriousness, if something isn't changed here, then we'll see firsthand what happens when the most powerful technology humans have ever built gets shaped by less than a quarter of them.
