“You can’t be it if you can’t see it”: Tackling discrimination in tech

Dr. Ayanna Howard and Dr. Timnit Gebru knew from the moment they met that they would have a lot to talk about. Ayanna Howard is the chair of the School of Interactive Computing at Georgia Tech, and Timnit Gebru is the co-founder of Black in AI. They first crossed paths in 2017 at the Tapia Celebration of Diversity in Computing, where they started a conversation about bias and discrimination in AI, issues they live every day as women of color in the robotics and computer vision fields.

When they got together again recently to continue that conversation, they invited us to listen in.

---

On how they met

AYANNA: Oh, I know! We were on a panel together.

TIMNIT: Yes, I remember exactly. It was this bias in AI panel that Sanmi Koyejo was moderating. It was so cool to find out about your work on trust in AI. After that, I was like, “Who is this person? How did I not know about her until now?”

AYANNA: I was equally impressed.

On getting started

TIMNIT: I always loved math. First of all, I always loved school. When people asked me what I wanted to be when I grew up, I said, “I want to be a scientist.” Since my dad was an electrical engineer, I knew it was an option. You have to remember, I went to an all-girls school in Ethiopia, so there was none of this “Girls are not supposed to do science” or “Africans or black people are underrepresented.” I’m just grateful for my childhood being completely different.

AYANNA: My dad was also an electrical engineer. We had role models. We could see people who were doing what we wanted to do.

I started programming in the third grade, BASIC, because of my dad. I think he thought of it as his heritage. He would bring home Radio Shack kits, and he taught me how to solder. I remember burning myself. He kept repeating, “You’ve just got to be patient.” Teaching a six-year-old to be patient with a soldering iron!

TIMNIT: I wonder if our parents weren’t electrical engineers…

AYANNA: I don’t know. I don’t know. There’s an expression: you can’t be it if you can’t see it. I think that’s extremely important.

On discrimination—and its subtle forms

TIMNIT: I knew about racism in an abstract manner when I was in Ethiopia. When I moved to the U.S. in 11th grade, the way I experienced racism was so different from the way I had imagined. People would be nice to you, and they would look you in the eye and say, “I think this class is going to be too hard for you.”

Even for someone as stubborn as me, at the end, my mom had to go talk to my guidance counselor, because I really believed that I probably wouldn’t get into any college. Then once I got to Stanford people told me I got there because of affirmative action, so it never ends.

AYANNA: I mean, some people don’t realize the power of their words, right? And you had such a good upbringing. Imagine the kids who don’t. They don’t have an advocate, so they just assume, “Oh, yes, I’m not going to college.” It’s so unfair.

TIMNIT: In Ethiopia, I went to an all-girls school, and I had an all-female kind of household, and everybody was an engineer. Also, all the women I know were working women. I really didn’t at all think about, whether or not I should have a career in this space—or whether or not it was appropriate or anything like that—until I came to the States. Then it was a battle.

The other thing I felt in undergrad—I don’t know if you felt this—is that I felt like I was so visible, and that I couldn’t mess up. That if I did badly on a test or something like that, it would be very obvious, because I’m, I don’t know, because I’m the only black person, or something like that. I still sort of feel that to a certain extent.

AYANNA: I remember the first time I realized that I was different. It was Calculus, a really hard class, a weed-out class. I went into the final with a borderline A, and I came out with a B. I went to the teacher, and he said, “You were the highest B.” I was like, “Well, you know, couldn’t you round it up?” He said, “I have to make the cutoff somewhere.” If you’re someone who hasn’t experienced something like that, you might think, “Well, but he had to make the cutoff somewhere.” But it’s how they say it: dismissive. It doesn’t feel right.

On AI and inclusion

TIMNIT: People used to ask me, “How did you get interested in bias?” I said, “That’s not what I got interested in!” I got interested in computer vision. The bias thing came out of being scared based on what I was seeing.

AYANNA: We’re getting to the point where every decision in your life is going to be based on an algorithm. Which loan you get, which school your kids get into, which hospital you’re assigned to. If there is bias against you, it’s going to affect your life. I’m not going to let this technology be used for evil.

TIMNIT: I try to collaborate with a lot of African researchers, just to try to understand what kinds of problems they’re working on. Because what happens is if you have technology in the hands of just a few groups of people, whether it is geographically located or gender or age or race, then you’re going to be prone to solving the problems that those few groups of people think are important, and you’re not going to be solving problems that other groups of people think are important.

A simple example is cars. A lot of studies have shown that in similar accidents, women sustained much worse injuries than men in cars, because the car designers had in mind men with prototypical male characteristics. Even when they were testing cars with crash test dummies, they only used dummies with prototypical male characteristics, so a lot of women and children were being disproportionately affected. This happens with medicine, too, with healthcare. You’re creating technology that is not working well for the majority of the world.

On leadership

TIMNIT: What I’d like to see is a lot more action and a lot less talking about inclusiveness and diversity. Because sometimes people talk a lot about it, and then when it comes down to it, it’s not really a priority, because they don’t want to be inconvenienced in any way. The definition of caring about something is, “What are you willing to give up? How much of your time are you spending learning about it?” That’s what I’m not seeing enough of.

AYANNA: You really have to respect individual differences, whatever they may be, when you are coming up with strategies and pushing forward. You should make everyone feel part of the team. Those individuals will then trust you. They’ll believe you. They will follow you.

One tech leader that I admire is Satya Nadella, who’s at Microsoft. Because what he’s done is he’s actually tried to make a culture shift in his leadership and in his viewpoint. The fact is: he has a child with a disability, so that changes your perspective of the world, right? Some of it is probably because he understands what it is to be different in a world that’s not made for you.

On the joy of engineering

TIMNIT: My favorite thing to do is just sit in a corner, read some papers, come up with ideas, and write code. I’m still an engineer at heart. I also used to love physics for that same reason. It’s math, it’s learning about some property of the universe, and it’s discovery.

AYANNA: One of the things I love about computer science is there is truth in it. I mean, there’s bias and all that stuff, but at the end of the day it’s math. Math doesn’t have a color. It doesn’t have a gender. Anyone who has knowledge of computer science can be creative. There’s a purity about computer science that I love. I’ll code something up, and at the end, I’m like, “Oh, that’s so beautiful.”

If you’re following your purpose, then things happen somehow, right? Things line up, and that door closes, but another one opens. That’s how you know that’s your purpose. That’s, I think, why I know I’m supposed to be doing exactly what I’m supposed to be doing. That gives me satisfaction.