Facebook Apologizes After AI Software Labels Black Men As ‘Primates’
Artificial intelligence systems have repeatedly been proven to be biased against Black people.
September 04, 2021 at 3:52 pm
Facebook issued a statement on Friday to apologize to social media users after its artificial intelligence software labeled a group of Black men as “primates,” USA Today reports. The gaffe on the social media platform involved a video from The Daily Mail, which showed Black men in altercations with white civilians and police officers.
Facebook users who watched the The Daily Mail video were then asked if they would like to “keep seeing videos about primates.” Although the video was originally shared on Facebook last year, social media users are only now saying that they saw Facebook's prompt recently appearing alongside the old video.
Facebook spokesperson Dani Lever said it was “an unacceptable error.” In a statement to USA Today, the spokesperson said the platform disabled the entire topic recommendation feature as soon as they realized this was happening.
“As we have said, while we have made improvements to our AI we know it’s not perfect and we have more progress to make,” Lever said. “We apologize to anyone who may have seen these offensive recommendations.”
In recent years, artificial intelligence systems have been blamed for being biased against Black people. According to The New York Times, Google Photos faced scrutiny in 2015 after labeling pictures of Black people as “gorillas.” Amazon and other technology companies have also faced similar issues.
Google faced another problem in 2016 when image searches for “three Black teenagers” showed mugshots, while a search for “three white teenagers” showed smiling white teens.
In 2019, a congressional hearing was held on the use of facial recognition technology, HuffPost reports. Rep. Alexandria Ocasio-Cortez spoke up at the hearing, saying technology can “exacerbate” racial bias in the criminal justice system.
"When tech companies develop algorithms that automate the assumptions of people from one demo, they start to automate subconscious bias," Ocasio-Cortez later said on Twitter. "When those flawed algorithms get sold to law enforcement, it can be disastrous for anyone that doesn’t look like most Silicon Valley engineers."
When tech companies develop algorithms that automate the assumptions of people from one demo, they start to automate subconscious bias.
When those flawed algorithms get sold to law enforcement, it can be disastrous for anyone that doesn’t look like most Silicon Valley engineers. https://t.co/SKxhYUgZEj
— Alexandria Ocasio-Cortez (@AOC) May 22, 2019
According to a 2018 study by the American Civil Liberties Union, Amazon’s facial recognition tool disproportionately misidentified Black people, incorrectly matching the faces of 28 lawmakers with people in mug shots.