Beauty AI Algorithms Are Laden With Biases - This Computer Scientist Wants to Change That
Image Source: Courtesy of Joy Buolamwini
Too often, the best beauty stories go Untold, solely based on a person's skin color, religion, gender expression, disability, or socioeconomic status. Here, we're passing the mic to some of the most ambitious and talented voices in the industry, so they can share, in their own words, the remarkable story of how they came to be - and how they're using beauty to change the world for the better. Up next: Joy Buolamwini, computer scientist and founder of the Algorithmic Justice League (AJL).
From a very young age, I was interested in looking at ways of using computers to help people. Growing up, my dad, who's a professor of medicinal chemistry and pharmaceutical sciences, used to take me to his lab to feed cancer cells. He would also show me the way he was using computers to develop drugs to fight cancer.
I've grown up with an appreciation for the search for truth, with both arts and science, thanks to my mother being an artist. I wasn't sure at first if I would be able to blend both the artistic part of what I did with the algorithmic auditing, but I took a chance to ask myself, what would it look like to be a poet of code, somebody who blends both worlds in order to do what poets do, which is illuminate uncomfortable truth or hidden insights in the everyday interactions we have?
I've always had a bit of an entrepreneurial spirit, too. In high school, I had a little web design company, and that allowed me to make some money so I could pay for equipment for basketball, track, and cross country, and then in college, I cofounded a hair-care tech company that would analyze hair type and give people personal product recommendations.
After that, I was really fortunate to get a Fulbright fellowship to go to Zambia, and I started an organization that taught girls how to code. By the time I got to grad school, I had the experience and confidence to start the Algorithmic Justice League, an organization that combines art, academic research, and advocacy in order to fight for people who are harmed by AI, who I like to call the "ex-coded."
"I started to look into if AI systems perform differently depending on the type of face that's being analyzed, and what my research found was in fact, they do."
When I was working on an art installation as a graduate student at MIT, part of the installation was meant to track the location of my face with software, but it didn't work that well on my face until I put on a white mask. This led to my research in 2017 on facial and analysis technology that could try to guess the gender of a face. That experience of putting on a white mask to be made visible to a machine is what really made me start to ask, are these machines so neutral?
I started to look into if AI systems perform differently depending on the type of face that's being analyzed, and what my research found was in fact, they do. From looking at AI systems from a number of major tech companies, I found that the systems perform better on men's faces than women's faces when it came to guessing gender, and they performed better on lighter-skin faces than darker-skin faces. That made me think, if those results had been reversed, would these computers be out in the market in the first place? It was that experience of coding in a white mask, and then having the opportunity to show some of the largest accuracy differences in commercially sold AI products at the time, that led me to look further into issues of coded bias, which really was the seed for starting the Algorithmic Justice League.
Image Source: Olay
In beauty AI, if we're thinking about analyzing faces or having faces processed by a machine - you'll see this with the kinds of filters that are used that might lighten your skin or slim your nose as a way of claiming to enhance beauty - that's based on Eurocentric standards of beauty and marginalizes women of color. Through my research, I realized I have this important perspective to bring from my lived experience of not being seen, of being ex-coded. It's something that doesn't just impact me, but it's something that is in our larger society because we have AI systems increasingly entering all areas.
Something else to think about in the beauty space is the use of AI in employment - deciding who gets hired, who gets fired, who gets a promotion. The beauty industry employs so many people, and understandably, companies want to try to adopt the latest technologies, but we really need to be thinking about the ways in which AI serves as a gatekeeper for who even gets to participate within the industry as well.
A major piece of what we do with the Algorithmic Justice League is ask how do we go toward a world with more equitable and accountable AI? Part of it is raising awareness, because you can't fix an issue that you don't even know about, and a lot of people are not aware of coded bias. This is why I was so happy to partner with Olay on the Decode the Bias campaign to fight bias in beauty algorithms. Thinking about who's coding and how we're coding is such an essential part of making a change. The brand's initiative to send 1,000 girls to Black Girls CODE camp to help them pursue STEM careers helps make sure that the people who are creating the technologies that shape society actually reflect society. The biases continue to show what happens when we're not in the room.
"In beauty AI, if we're thinking about analyzing faces or having faces processed by a machine - you'll see this with the kinds of filters that are used that might lighten your skin or slim your nose as a way of claiming to enhance beauty - that's based on Eurocentric standards of beauty and marginalizes women of color."
When I first started my research, I experienced discouraging comments, but I didn't let that stop me from pursuing something that I felt was extremely important. Being a woman and having dark skin gave me a lived experience that led to impactful research that more mainstream peers just weren't pursuing or prioritizing at the time. And so, my experience with coding in a white mask is what actually catapulted this research.
When you're in a field where your perspective is not centered, you really need to find a community of support. I've had a very strong support system and great mentors and a really solid foundation, and so I think from the start, that's been really helpful for me. That's shown me the importance of having that kind of community, especially as a young woman of color and for girls of color. I would encourage anyone who has ideas and wants to enter this space but is afraid they're going to face pushback to push forward anyway. We still need you.