Can Technology Be Racist?

 She didn't really pose it as a question, more like a statement with a hint of uncertainty. 

My BFF and I were in a tony taco joint newly opened in our hipster, yet diverse, south side neighborhood in Chicago. 

I was explaining to her a video I found that showed the automatic soap dispenser on a bathroom sink in a posh Atlanta hotel wouldn't dispense soap to a black man's hand but did so freely to his white friend. 

The video's narrator and creator T.J. Fitzpatrick, was attending a conference at the hotel and noticed when he put his hand under underneath the soap dispenser nothing came out. But when his friend "Larry," a white guy did it the soap readily spilled onto his palm. He dubbed the sink "Whites Only," and laughed it off. 

I told my BFF that the LED sensors work by refraction of light so dark skin sometimes won't "trigger," the desired response. As explained in this cool article by Max Plenke in Tech.Mic., black absorbs light so black skin may not trigger the LED sensors. Me, being a UX researcher and designer, I at first, only thought about the design implications. My mind quickly went to questions like, "Did the company do usability tests on this product with a diverse testing group?" "Who did their user recruiting for the testing?" "How easily this could have been changed with a little diversity in the user groups!" 

But my BFF's question caught me off guard. And it made me wonder...

Can technology be racist? 

The Downside of Man-Computer Symbiosis

Yes, I know full well that we haven't reached iRobot status yet, where AI is so advanced that it outsmarts, out-thinks it's own protocols and begins to make up new ones. (Although we're enticingly close.)

So the idea of something as inocuous as 1's and 0's taking on wholly unthinkable human traits like racism seems so far fetched. (Although if you believe Elon Musk we're all in a computer simulation anyway so it's really not that weird to speculate upon.)

To me racism is a loaded entity. It implies a deliberate intent to exclude, harm or even destroy people simply for no other reason than the color of their skin. 

Not getting soap isn't a matter of life and death. But after a bit of thinking I pondered her question...indeed could some of the products we're creating be following in the unsavory mode of some of our most repugnant human qualities? Afterall, with AI in its infancy, technology products are still being developed by humans. And no matter how good our intentions, it seems we may be allowing the most undesirable of human qualities to slip into our technological lives. 

Me being a UX researcher, I flicked out my fastest Google typing fingers and began to scour the Internet in search of an answer. But really all I had to do is look over my own life and my use of technology. Once I did that the question "Can technology be racist?" quickly turned into "Is technology racist?" Well, let's see. 

The Lasting Legacy of the Shirley Card

I really wanted to see how the LED sensors could be so sensitive to white skin but not black skin. Like, was that natural (not racist) or did someone program that in (maybe not racist but definitely biased). And then I remembered something I learned way back in print design called the "Shirley Card." 

I used to be a newspaper designer back in college. I was the Sports Editor and often our Wednesday night basketball games were the last thing that went into the paper before we sent it off to the printer for a Thursday morning publishing run. When that happened I worked with our Editor to get the paper ready for publication and that meant color balancing. That's when I learned about the Shirley Cards. 

Back in the 1940s when color in film began to be popular in photos, movies etc., Kodak, the dominate photo company at the time, set color balancing standards for skin tones in photographs using "Shirley Cards." These cards were photographs of women in various poses but they all had one thing in common - the women were white. 

So from the very early days of color photography until the late 1990s the color film standard was set for white skin. Which meant lighter-tone skin became the starting point for any technology that used color in image capturing. Back then if you took a picture of your cute black granny at the family reunion, it was calibrated using Shirley not someone who looks like her. Which means you saw the whites of her teeth and eyes but not much else. 

Well, you say, color photography has advanced significantly. And it has. As this video by Vox shows Shirley Cards began to be diverse after the 1980s. But as the same video points out, the legacy of the "Shirley Cards," still lives on today in even the most advanced technology.

Experts say the Shirley Cards injected a "racial bias," into image capturing technology and for some making our technology "racist." Though many of the photography color bias has been corrected, you can still see Shirley's imprint in all of our image capturing technology today from basic digital photography, Instagram filters, to more advanced technology such as facial recognition. Here are just some examples: 

• In 2010, Microsoft caught heat for its Xbox Kinect motion-sensitive game controllers. Until a few of GameStop employees stated the controller didn't recognize black faces. Consumer Reports refuted that assertion because they couldn't replicate it and blamed the inability to work on poor lighting. But the question was raised.
• This July, Google got a bit of jolt when web developer Jacky Alcine tweete this "Google Photos, y'all f*cked up. My friend's not a gorilla." Apparently, it's AI photo app had tagged a photo of his black friends with the word "gorilla." Apologies ensued, yada yada. Read more if you want but you get the idea. 

• A recent article in The Atlantic, showcased how racial bias may be seeping into the most advance technology to date: facial recognition. Read it here for all the gory details but basically facial recognition algorithms are taking on their creators' own biases. Algorithms created in China and Japan recognized Eastern Asian faces more readily than white ones. While algorithms created in France, Germany and the U.S. were much better at recognizing white faces than those of people of color. The differences in the algorithms recognition factors suggest our technology is inheriting its creator's racial biases. 
• Finally, I came across this intriguing headline "Racist camera: No I did not blink, I'm just Asian." Written by Joz Wang, an American blogger, who had recently purchased a Nikon S630 for her mother, the post pointed out a glaring glitch in the camera's facial recognition software. In an effort to help users take better photos, the camera is equipped with the facial recognition feature that detects if someone is blinking. It asks, "Did someone blink?" But the camera kept asking it even though the Asian woman pictured was clearly not blinking. You can see the photo here.

So Really Can Technology Be Racist?

I realize these few examples out of the whole of humanity isn't a lot. But I think they point out some very glaring biases when it comes to image capturing technology and facial recognition software. And further research shows that this color bias in imagery is affecting a lot of the technology today. From infrared technology, to facial recognition, it's definitely something UX designers and product developers should keep an eye on. So how can we ensure the products we create are inclusive? Well that's pretty easy:

• Hire diverse product teams. From UX researchers, designers, UI folks and developers have people of color, women, seniors, people living with disabilities, liberals, conservatives, gaming nerds and bookworms or whatever. The more diverse your product teams, the better your products, the better your products the more money you make. Diversity=money. Simple. (Don't believe me? See the research.)
• Make sure diversity is a key tenet of your UX research and testing. A diverse usability test can help point out color bias in products early on. But this also includes users with disabilities, reading impairments or other accessibility issues. I've found that when I have accessibility as a key tenet of my design, my design is much better for all users. Build diversity into your user recruitment and testing as part of the natural order of things. This can only be good for your product design but it can also help you avoid embarrassing launches. A diverse usability test would have caught the soap dispensers color bias easily and fixed the glitch with no embarrassing viral videos in tow. 
• Understand the strong connection between diversity and true innovation. Are you tired of the one-page, scrolling, big imagery, little text websites that are ubiquitous on hipster websites? Yeah me too. Why are they like that? Lack of diversity. We all know that the technology industry is not very diverse. It's made up mostly of white males who set the standard. Research shows diversity spurs innovation and it's the diversity of the innovators that leads to such great product development. Emi Kolawole, former Innovation Editor at the Washington Post now at Stanford explains the link between diversity and innovation so much better than I. 

Answering the Question

Can technology be racist? I suppose one can argue that since a human can and humans create it then the answer is yes. But I prefer a more nuanced view. I think technology can be exclusionary, biased and lacking in innovation if the technologist are of one ilk. Our entire world is made up of a diverse ecosystem that functions in relative harmony creating amazing experiences every second of the day. Design should be the same.