Google did not intend to label black people “gorilla” in its Photo app and is apologizing for the blunder. Google was criticized on social media due to the racist connotation in the label. Its product, an image-recognition feature, automatically tags photos using an artificial intelligence program.
The mistake was brought to Google’s attention by a software developer based in New York. In his account @jackyalcine, he posted a screen shot from his smartphone on Sunday of images showing the new Photos app that sorted a picture of two black people into a category “gorillas.” He goes, “Google Photos, y’all f****ed up. My friend’s not a gorilla.”
As the Android platform owner did not intend to label black people “gorilla,” it reacted quickly before the mess reached the media. In less than two hours, chief architect of social products Yonatan Zunger, responded to the tweet, seeking access to Alcine’s account to check on the error. He tweeted, “Sheesh. High on my list of bugs you never want to see happen. Shudder.” The engineer said it is “100 percent not OK.” Google spokeswoman Katie Watson said they are “appalled and genuinely sorry.” The search titan is already working to prevent it from happening again.
According to Alcine, the error affected some photos in his collection. This is not the first time an error happened like this. In May, iTech noted the app tagging photos of dogs as horses. The application also labeled some people as seals. Google was criticized for the error, but this most recent mistake is more sensitive.
Alcine expressed his dismay for calling his friends apes, a comparison which is a racial disgrace when it comes to black people. Despite the apology, the image recognition’s error adds to the company’s present burden, considering that it is currently facing accusations for being discriminatory in its hiring practices.
Google is among the tech firms whose workforce seldom has blacks and Hispanics, just mostly Caucasians and Asians. ABC News says Google’s recent disclosures of its workforce diversity has 94 percent white or Asian and only a percent black. Due to this fact, some people in social media wondered if this can be the flaw of white and Asian engineers who are perhaps not so sensitive regarding how black people will feel about the labeling.
The error also adds up to the rising racial tension in the U.S. where blacks were killed by policemen. Another source of fuel to the fire was when nine black churchgoers were shot in Charleston, South Carolina, on June 17. The image recognition blunder also highlights the drawback when people rely on machines for tasks they are supposed to handle.
In late May, the Internet titan introduced the Photo app, which uses a program to sort thousands of pictures into various categories such as places, animals, and activities. The Mountain View, California-based company, however, warned that it is not yet perfected as they are still fine tuning it the feature. Watson said there is still a lot of work to do with the automatic image labeling.
The Mountain View firm is a leader in artificial intelligence, but it looks like its computers still have a lot to learn about judging people to avoid future sensitive blunders. In its new Photos app, the feature will suggest categories when users do their searching. These categories are developed from “machine learning,” which is the science of training machines to do human tasks. According to BD Live, Google has removed “gorilla” from the categories, so it will not appear in the suggested categories anymore.
Google did not intend to label black people ”gorilla.” It was a machine glitch. A similar image recognition error also happened to Yahoo’s Flickr, which was flooded with complaints when its automated feature tagged black people as “animals” and “apes.” Moreover, it also used “jungle gym” to label a Nazi concentration camp.
By Judith Aparri
Edited By Leigh Haugh
BBC: Google Apologizes for Photos App’s Racist Blunder
ABC News: Google’s New App Blunders by Calling Black People ‘Gorillas’
BD Live: Google’s Photos App Tagging Blunder Shows Limits of Algorithms
Photo Courtesy of Kārlis Dambrāns’ Flickr Page – Creative Commons License