5 Shocking Ways People Keep Making Technology Racist
-
Saikat Bhowmik
·
- September 02, 2017
While humanity's endless battle against racism might feel like two steps forward, one giant step back, at least we can take comfort in the idea that our merciless robot overlords will be immune to such human weaknesses. Or will they? It turns out that, much like the children of Klan members and anti-vaxxers, technology itself is cursed by the folly of its creators. Case in point ...
5 Color Film Was Specifically Formulated To Make White People Look Good
When the government stepped in to break up Kodak's monopoly on film processing in the mid-1950s, it opened up a world of opportunity for independent photo labs. Kodak, forced to loosen its stranglehold on the industry, but not wanting to fully let it breathe, developed a line of printers to market to said independent photo labs. Of course, if the photos produced by Joe Random's Bumfuck Photo Hut were still to bear the Kodak name, Kodak had to make sure their quality stayed consistent. To do so, they provided calibration guides in the form of "Shirley cards."
"I am serious."
The cards were named after Shirley Page, Kodak studio model, human calibration tool, and possessor of skin whiter than fresh snow on an American Apparel store. As you might imagine, equipment tuned to make Shirley shine her brightest made those with darker skin appear "dulled and darkened," but no one much gave a shit up until the 1970s. That's when Kodak competitor Polaroid stepped in, not with a new type of printer, but with a new type of camera flash designed to better reflect black skin. We'd laud Polaroid's efforts ... if not for the fact that they only did it to better aid South Africa in enforcing apartheid.
In the end, it wasn't humanism, but consumerism that leveled the film development field. When producers of dark chocolate and dark-stained furniture complained that Kodak film made their products look shitty, Kodak finally took note. More modern Shirley cards feature not just a single, pearly white Shirley, but an entire mixed-race gaggle of them.
4 Facial Recognition Software Has All Sorts Of Problems With Non-White Faces
In 2009, Taiwanese-American Joz Wang bought a Nikon Coolpix digital camera as a gift for her mom. But first, she fired the camera up and loaded it with portraits of her and her brother, because she's an immeasurably better child than we could ever hope to be. That's when she discovered that the camera's software code had some built in personal issues: every time she or her brother smiled, the camera flashed a warning that someone had closed their eyes in the photo.
"I can't be racist, I was made in Japan!"
It's not the only time a camera spelled out its racist tendencies more clearly than your uncle's Facebook page. In 2009, coworkers Wanda Zamen and Desi Cryer discovered that HP's face-tracking webcams had the same priorities as a Texas Congressman.
On the plus side, if we need someone on a heist to get past motion sensors ...
HP issued a public apology explaining the camera's shortcomings in regards to tracking black faces, but Google wasn't listening. When scrolling through his photo roll in 2015, Oluwafemi J Alcine noticed that Google Photos' automated tagging algorithm had labeled cars as cars, bikes as bikes ... and him and his African-American friend as gorillas.
How do you say "holy shit, not cool" in binary?
3 FaceApp Thought "More Attractive" Meant "Whiter"
Whereas Snapchat allows you to transform yourself into a rainbow-puking unicorn or a talking Chipotle turd, FaceApp promises more realistic photo manipulation. It allows you to "transform your face using Artificial Intelligence." With just one poke, you can age yourself by decades, flip genders, or make yourself more attractive ... words that, in FaceApp parlance, are synonymous with "whiter."
Which is why people spend so much time tanning.
When it became apparent that FaceApp's "Hotness" filter was nothing more than a Caucasian-maker, CEO Yaroslav Goncharov issued a swift apology: "We are deeply sorry for this unquestionably serious issue. It is an unfortunate side effect of the underlying neural network caused by the training set bias, not intended behavior. To mitigate the issue, we have renamed the effect to exclude any positive connotation associated with it."
Yes, they just renamed the effect -- from "Hotness" to "Spark" -- which was obviously the best course of action, as opposed to, say, fixing it, or removing the filter entirely.
2 Digital Redlining Ensures You Never Have To Step Foot Inside A Black Neighborhood
The term "redlining" dates clear back to the National Housing Act Of 1934, and refers to the practice of denying opportunities to certain areas based mainly on their racial makeup. While redlining has seen much decline in the real world, its effects are still glaringly apparent in the digital one. When Pokemon Go exploded onto every last smartphone in the United States, players noticed something peculiar: The number of available PokeStops and Gyms in predominantly white neighborhoods vastly outnumbered those in predominantly black neighborhoods.
473 Comments