Dr. Timnit Gebru completed her PhD at Stanford in the Artificial Intelligence Laboratory, studying computer vision. Timnit also worked at Apple, designing circuits and signal processing algorithms for various products, including the first iPad, and she will spend a year working at Microsoft research in the Fairness Transparency and Ethics group in NYC.

In Part One of our talk, Timnit described her first day at a US high school, where she was openly discouraged from taking advanced chemistry, and how feeling like you don’t belong is one reason why so many women start in STEM but eventually leave to work on something else. Here we discuss why diversity is so important.

How can regular folks help make sure everyone has access to STEM and STEM-like professions?

I think that we have to change the perception of what types of people are in STEM. You don’t have to be Sheldon from The Big Bang Theory. I consider myself to be an artist as much as a scientist. Art is just as important to me. I dance, I play piano, I like hanging out with people. If I had been raised entirely in the US, I think that I would have gotten the message that I wasn’t supposed to be in STEM. We have to call out sexism and racism when we see it. We also need to be aware of how much discrimination exists. These spaces are not created for women or people of color. I often go to parties at conferences and feel extremely uncomfortable. A lot of times, people see someone saying/doing something wrong and they stay silent. They don’t point it out. Don’t be apathetic.

Tell me why it is so important that organizations have diverse employees, researchers and academics. Who benefits the most?

One time, a friend gave me this analogy of rowing a boat. He said that him and his coworkers were so in sync, it was like they were rowing a boat — they would all row in unison. That, to me, is a lack of diversity. You all think the same way and so you row the boat in unison. However, if you are rowing the boat towards a waterfall, there will be no one on your team to point out that there might be a waterfall and maybe you shouldn’t row in that direction. So you will be efficient without diversity but you will just as efficiently get into a pitfall. Without diversity, the science suffers, the company suffers and the user suffers.

That’s interesting because I often use a canoe analogy. You need all kinds of people in a canoe. You need someone to lead. You need someone to navigate, someone to motivate, you need people to propel the boat forward, and you need someone who is not afraid to tell you that you’re moving in the wrong direction. It takes all kinds of people to be efficient and to succeed.

When potentially life-changing technology comes into our lives, there is so much to consider: ethics, ownership, access, etc. Who is responsible for making sure technology is available to everyone, everywhere?

It is a combination of people, but researchers in particular have to be socially conscious and think about these repercussions. For example, in June, there was an AI for Good conference in Geneva, organized by the UN [ROSS coverage here]. There were maybe two black speakers and one of them was my friend, Facebook researcher, Moustapha Cissé. [Interview here.] So everyone gathered to talk about “AI for good,” but good for who? If you are including Africa, then where are all the Africans? Why are Africans working on AI for good as applied to their problems excluded from the conversation?

So you don’t see a lot of inclusiveness at the moment?

When you go to these conferences, how many researchers do you meet from India? How about any part of Africa? Such a minute percentage of the world’s population is monopolizing this field. And this cycle will continue. If I am a French white man and I see lots of French white men, I’m going to feel very comfortable. I will see my culture represented. My political points of view represented. If I am black woman in this field, from any part of the world, I will barely see anyone there. I will feel very uncomfortable in most conversations. I will rarely interact with people who hold a similar view to mine in many respects.

Our field is currently an echo chamber because we don’t hear from 99% of the world. How can you ensure that AI is good for all people when you have no idea what 99% of people are doing, you don’t interact with them, and you don’t talk to them?

Where do you see AI in terms of life-changing technology? People are often excited about the possibility of robots cleaning the house but its potential is a whole lot more than that.

I rarely think of robots cleaning or cooking. I think AI can have positive and negative impacts. It can have a positive impact on healthcare, education, government and agriculture, for example. It can have a negative impact in conflicts — drones for example.

“I think that we have to change the perception of what types of people are in STEM. You don’t have to be Sheldon from The Big Bang Theory. I consider myself to be an artist as much as a scientist.”

How optimistic are you about the future, in terms of women and diversity in professions currently lacking both? And if you are optimistic, tell me some good news.

I’m not so optimistic but I do see that various people who care about this are coming together. One piece of good news is that I think we will have 50–100 black attendees at the 2017 NIPS [Neural Information Processing Systems]. Last year I saw five black people out of 8,500 attendees. This year, there will probably be around 10,000 attendees so 50–100 will be less than 1%, which is terrible — but it’s a lot better than five people.

What are the biggest misconceptions about women and women of color, in terms of being capable and intelligent?

I’m not sure. I think people don’t understand just how much you have to deal with when you are a woman of color in our field. They’re unaware of the normal everyday ways in which people treat you differently. And they don’t self-reflect and think “what am I doing to contribute to this?” Even when they are well-meaning, it’s hard for them to see how their behavior can be problematic.

How will you know when things have changed? What would you have to see that would make you go: “FINALLY!”?

When I forget that I am a woman or a person of color. It’s like money. When do you know that you are rich? When you don’t have to think about how much money you have.

‍