Facebook has apologized for an AI-powered feature that asked users if they wanted to “keep seeing videos about Primates” after they finished a Daily Mail video featuring Black men.The New York Times reported that the video featured “clips of Black men in altercations with white civilians and police officers” and “had no connection to monkeys or primates.” The video was published in June 2020; the report didn’t say for how long the offending prompt was shown.Facebook told the NYT the label appearing on that video was “an unacceptable error” and said it would investigate the feature to “prevent this from happening again.” The recommendation engine that applied the label to that video has been disabled while the company looks into it.This isn’t the first time something like this has happened. Google Photos sorted an image of two Black people into a category labeled “gorillas” in 2015, for example, and Twitter recently confirmed that its image-cropping algorithm was biased in favor of white individuals.Bias has also been found in algorithms related to facial recognition, healthcare, and other areas. Yet tech companies have made AI an increasingly vital aspect of their services, apps, and hardware despite ongoing concerns regarding the fallibility of…
Read More











