Resources / Blog

Stanford’s Margot Gerritsen on Women in Data Science

Join Infinia ML’s Ongoing Conversation about AI

Episode Highlights from Machine Meets World

This week’s guest is Margot Gerritsen, Stanford Professor and Co-Founder and Co-Director of Women in Data Science.

Highlights from our conversation include:

“Data-driven decision-making is becoming more and more important. So who is sharing in that wealth creation? Who is sitting around the table when those decisions are made, that are impacting all of us?


“As the only woman in a team of men or one of the few women in a team of men, you are seen different and you start to regard yourself as different. And as long as you’re seen different, you’re not really part of that conversation — or it’s much harder to be.”


“Algorithms themselves are constantly tweaked and changed and adapted by people, too. . . . you do tweaking to get the results that you’re expecting. And the results that you expect, of course, are biased by your own background and the background of your team.”

Photo by Christina @ wocintechchat.com on Unsplash

Audio + Transcript

Margot Gerritsen:
Data is what a lot of people say the new oil, the new gold. This is where the wealth creation sits. Data-driven decision-making is becoming more and more important. So who is sharing in that wealth creation? Who is sitting around the table when those decisions are made, that are impacting all of us?

James Kotecki:
This is Machine Meets World. Infinia ML’s ongoing conversation about artificial intelligence. I am James Kotecki and my guest today is Margot Gerritsen, Stanford professor, and Co-founder and Co-director of Women in Data Science. Margot, welcome to the show.

Margot Gerritsen:
Well, thanks very much for having me.

James Kotecki:
So Women in Data Science, when did this organization start? And I wonder when it started, can you give me a snapshot of maybe the percentage of people in data science careers who were women and then what that percentage is now?

Margot Gerritsen:
Oh, the percentage is still the same. So when I started thinking about computational sciences 40 years ago, when I was in high school, we had around 15% maybe, and that’s on average in the Western world. And that’s still the case. So we haven’t really moved it, despite all the talk, despite all the work. Why we started — in 2015, I was invited to give a talk at a conference at that time and I couldn’t make it. And when I saw the listing of the final program of the conference, I realized, again, there were only male keynote speakers. And in fact, I think there were only male speakers period. So I asked the organizer, I said, “Hey, why is that?” And he says, “Well, Margot, we asked you and you couldn’t make it.” So I said, “Well, what about all these other women that I know in this field?”

Margot Gerritsen:
And he said, “Oh, we really tried to find women speakers, but we just couldn’t find any.” And for me, that was the last straw. And half an hour later, I was having coffee with my very good friend and coworker, Karen Matthys and Esteban Arcaute, who is a former mentee and student in the Institute I was running at the time. And the three of us decided we had to do something. And so I said, “Well, why don’t we just organize a conference?” And that was just on campus. We sold out in record time. And without advertising all that much, we got 6,000 people in the live stream. So at that point we knew we really hit a nerve. And then we grew and every year we tried to do something more.

James Kotecki:
If I am just a crusty, emotionless, kind of just bottom line numbers-driven person, why is it important to have women in data science, women in AI conversations, ML conversations?

Margot Gerritsen:
From a business perspective, this is extremely good. You’re opening up a market, you’re opening up a talent pool. So, to me, it’s always been a no-brainer. I never understood. I do understand the temptation to hire people like yourself, because that seems to be risk-free or lower risk. From a social perspective also, and an equity perspective, I find it really, really important. Look, data is what a lot of people say the new oil, the new gold. This is where the wealth creation sits. Data-driven decision-making is becoming more and more important. So who is sharing in that wealth creation? Who is sitting around the table when those decisions are made, that are impacting all of us? Designing software that impact who gets insurance at what price, that impact facial recognition, that impact medicine. So to me, it just doesn’t make any sense whatsoever, James, not from a cultural perspective, not from a care for the world perspective or care for others perspective and absolutely not from a business perspective.

James Kotecki:
But that number, that 15% of people in data science are women, hasn’t really moved over the last, you said, four decades.

Margot Gerritsen:
Let’s think about why it hasn’t moved. So I mentioned one thing, that it is very hard in quite a homogenous culture that we have in data science to get people to buy into this idea that maybe hiring people that are not like you is good for you, right? Because as I said, it’s easier. That’s one. But really, I think we’re still struggling in this field with two myths. One is that to be successful in an area like data science, you must have a really strong innate ability to be successful. That’s a myth that has been debunked many, many times. But you still hear a lot of people say that — also to young girls, “Oh, you’re just not good at math.” Or, “I wasn’t any good at math and that’s okay.” Nonsense. Nobody says you’re just not good at speaking or writing, or you’re just not good at running. Everybody knows that if you just train, you can be pretty decent runner. That’s the same for this too.

Margot Gerritsen:
And the other myth is a very, very sad, a myth really, but it’s perpetuating and it’s perpetuating everywhere: that this natural ability is not found as much in girls or women as it is in boys or men. That has also been debunked many times. I don’t know how many times we have to say this, but as a culture, as a society, we have to move away from these silly notions. As a girl growing up at school and in culture, that is what you hear. And it really puts girls off at a very young age. What I find incredibly painful and difficult is if women really want this, when they really want this, that they are discouraged for the wrong reasons.

James Kotecki:
It’s so important, the people that are going into these roles and doing these tasks. And the kind of people who are funneled into this, the determination is being made by cultural factors, subtle factors, subconscious factors that, to some degree, seem completely removed from kind of this hard-edged, high tech stuff. It’s like this really nuanced everyday decision-making that we have to kind of figure out inside our own minds and the minds of our culture.

Margot Gerritsen:
Even if we get more girls or women into these fields at university or into apprenticeships or starting at companies, they don’t stay. Obviously, they don’t really feel welcome. They don’t really feel included. That’s often the case, right? And a lot of people leave because they feel still like an imposter, that they don’t have what it takes. They don’t feel part of the network. And that is because they are the minority. As the only woman in a team of men or one of the few women in a team of men, you are seen different and you start to regard yourself as different. And as long as you’re seen different, you’re not really part of that conversation — or it’s much harder to be. Now I’ve been that my whole life and I’ve always thought I did pretty well in this, but it hasn’t been easy for me either.

Margot Gerritsen:
I’ve had a lot of misogyny. I’ve had harassment at every level. And that’s really tough. So it’s no wonder that a lot of women say, “As long as I’m so different . . .” And you don’t even see the role models so much, you don’t really see people like yourself being promoted and thriving — it’s really, really tough to stay. But it will remain difficult if we’re below that threshold, after which you’re no longer seen as different. And that threshold is probably 25%, 30%, that’s been my experience. In the Institute that I ran at Stanford, we changed the percentage of female students, from 5–10% to over 25%, a third, sometimes even 40%. And the culture in the whole Institute changed because of this. And when you normalize, it helps everybody. It helps the men because they’re totally comfortable. They’re also not questioning the women anymore, because there’s so many around and they’re doing really good things. But to normalize this, you need a higher percentage.

James Kotecki:
Are you worried about a potentially negative feedback loop here where there aren’t that many women in data science and so the algorithms that are increasingly influencing our lives are not as influenced by women and therefore are more likely to be discriminatory against women — and you could obviously insert other kind of less-represented groups in data science into this as well — and are you worried that at some point it just becomes locked in where these algorithms are kind of self-perpetuating a feedback loop of discrimination and bias?

Margot Gerritsen:
Yes, of course I am. And I think that point where we get locked in, to some extent, we’ve already passed. There’s a lot of codes that are out there that are being used over and over again. It often takes the other person to point this out, too. There’s something really frustrating about this when you’re a woman. I’ve been in situations where I advised companies for their software and looked at the data on which they were training and said, “Hey, did you realize that in this data you have hardly any women or you have hardly any people of color?” And very often would say, “Oh gosh, we hadn’t seen it.”

Margot Gerritsen:
And you often don’t see it if you’re surrounded by the same people every day. I can’t really blame them directly, but of course I blame that culture. Algorithms themselves are constantly tweaked and changed and adapted by people, too. You know this is manual labor often, right, to really get an AI algorithm or machine learning algorithm working. And you do a lot of tweaking and you do tweaking to get the results that you’re expecting. And the results that you expect, of course, are biased by your own background and the background of your team.

James Kotecki:
Well, Margot Gerritsen, Co-founder, Co-director, Women in Data Science and a Stanford professor. Thanks so much for joining us today on Machine Meets World.

Margot Gerritsen:
You’re very welcome, James. It was really fun to talk to you.

James Kotecki:
And thank you so much for watching and/or listening. You can always email the show. It’s mmw@infiniaml.com. Please like this, share this, give those algorithms what they want. I am James Kotecki and that is what happens when Machine Meets World.