We Already Know What You're Going to Say

So you read our piece on AI's gender problem and you have thoughts. We get it. We've been doing this long enough to know exactly what's about to hit our inbox. So instead of letting you fire off a reply you'll regret at your next performance review, we went ahead and answered every complaint in advance.

You're welcome.

Oh, and before you get upset because this is off brand: it's not. Have you read our newsletter? We're snarky AF!

"This is a tech newsletter. Stick to tech."

It is a tech newsletter. And this is a tech story. The composition of who builds AI directly determines what the technology does, who it works for, and who it fails. When facial recognition systems can't detect dark-skinned women, that's a technical failure caused by a workforce composition problem. When 44% of AI hiring tools show gender bias, that's a product defect. When ChatGPT automatically assumes women are younger and less experienced, that's a bug.

If we covered a story about how AI image generators couldn't render hands properly, nobody would say "stick to tech." This is the same category of problem; it just makes some people uncomfortable.

"It's a meritocracy. The best person should get the job regardless of gender."

Totally agree. And here's the fun part: the data suggests meritocracy isn't what's happening.

Women founders get interrupted nearly five times more often than men in VC pitch meetings. Women receive only 18% as much praise for using AI at work, even when they're doing the same quality of work. 82% of women in tech report having to prove themselves more than male colleagues. Amazon's AI hiring tool literally penalized the word "women's" on resumes.

That's not meritocracy. That's pattern matching to previous (biased) data. Meritocracy would produce different outcomes than "83.6% of all VC money goes to all-male teams." Unless you believe men are inherently 5.6x better at building companies, which... good luck with that argument.

"Women just aren't interested in tech/STEM."

Latvia has nearly half its AI professionals as women. Finland ranks second globally in gender parity. Women's AI talent has expanded significantly in 74 of 75 economies since 2018.

If women "just aren't interested," someone forgot to tell Latvia. The interest is there. The structural barriers are also there. When only 21% of entry-level women are encouraged by managers to use AI tools versus 33% of men, that's not a preference gap. That's a management failure.

Also worth noting: computing was literally women's work for decades. Women programmed ENIAC. Grace Hopper invented the compiler. The field only became "male" when it became prestigious and lucrative. Meredith Whittaker has documented this history extensively. Women didn't lose interest. They were pushed out.

"This is just DEI/woke stuff."

Sure. Here are some numbers.

Gender-diverse teams make better decisions 73% of the time versus 58% for all-male teams. Companies with 30%+ women in leadership are 15% more profitable. Diverse AI teams show 67% fewer bias incidents and 34% higher innovation.

You know what's actually bad for business? Building facial recognition that doesn't work on half the population. Training hiring algorithms that filter out qualified candidates. Designing health AI that misdiagnoses women. Those are product recalls waiting to happen. The AI companies shipping biased products are going to learn this the expensive way.

Also, please define woke? Does it mean "not asleep"? Why is that bad? We try to see with eyes unclouded to predict the future and point at potential problems. It's difficult in a world where you can see only what you want to see. I just had an idea to research this and this is what was found. If it makes you uncomfortable, congratulations, you're human. It's uncomfortable!

"You're cherry-picking data."

We linked every single claim to its source. Click any of them. The data comes from the Stanford HAI AI Index, McKinsey, the World Economic Forum, UNESCO, the ILO, CNBC, Russell Reynolds Associates, Lean In, MIT, and peer-reviewed research in Nature.

If you'd like to send us counter-data from equivalently credible sources, our inbox is open. Grant@theneurondaily.com

"But what about [Daniela Amodei / Lisa Su / Mira Murati]? There ARE women in AI."

Yes. And we can name them all in about 30 seconds, because there are so few of them at the top that they're each individually famous partly because they're exceptions. That's... the point.

Daniela Amodei is the president of Anthropic. She's one of only four women CEOs among 39 major AI organizations examined. Lisa Su runs AMD. Mira Murati left OpenAI. The fact that you can name the women proves the argument; you can name them because they're rare.

Quick: name five male AI CEOs. Sam Altman, Elon Musk, Dario Amodei, Jensen Huang, Demis Hassabis, Satya Nadella... that took you three seconds and you ran out of fingers before you ran out of names. That asymmetry is why we wrote this.

"If women want to be in AI, nothing's stopping them."

Except the 96% of deepfake pornography that targets women. And the boys' club culture that kept qualified women from joining OpenAI's board. And the fact that women's representation in European tech roles is actively declining (from 22% to 19%). And the VCs who interrupt women founders five times more often. And the managers who don't encourage women to use AI tools. And the companies that are deleting their diversity commitments.

But sure.

"Men and women are just different. Men are naturally more drawn to technical work."

We'll let the World Economic Forum's data on 75 economies handle this one. If the gender gap were biological, it would be roughly consistent across cultures. It's not. It varies wildly by country. Latvia and Finland look completely different from the US and Japan. That's culture, not chromosomes.

"So what do you want, quotas?"

Nobody said quotas. The article discusses structural barriers (funding bias, cultural norms, recognition gaps, management failures) and the consequences of those barriers (biased AI products, less innovation, worse business outcomes).

Fixing structural barriers isn't quotas. It's things like: training managers to encourage AI adoption equally, auditing hiring algorithms for bias (which Colorado now requires by law), not scrubbing diversity pages from your website, and maybe not building chatbots that generate 4.4 million sexually explicit images of women in nine days.

The bar is not high.

"You're just virtue signaling."

We're an AI newsletter that covers the AI industry. 86% of the 6.1 million workers most vulnerable to AI displacement are women. Women are three times more likely to lose their jobs to AI. Lots of our readers are women. AI products are demonstrably worse when built without diverse teams. This affects our readers, our industry, and the products we write about every day.

Covering it isn't virtue signaling. Ignoring it would be malpractice. There's a lot of things we could cover that we don't. We're trying to get better at that now that we're growing.

"But you're just two dudes on your podcast."

You're right, which is why we interview really cool people like Carina Hong, Zuzanna Stamirowska, Eve Bodnia, Nicole Baer, Kari Briski and plenty of others. We can do better, as can everyone else. I even think we should have a third co-host to even this out, or eventually replace myself on the podcast. Who knows what the future will bring?

"I'm unsubscribing."

We'll miss you. But probably not as much as you'll miss finding out about AI developments three days before everyone else.

See you tomorrow.