According to The Guardian, Microsoft software used to replace the human journalists running news site MSN.com confused two mixed-race members of British pop group Little Mix. In an MSN.com article headlined “Little Mix star Jade Thirlwall says she faced horrific racism at school”, the software mistakenly picked a photo of Thirwall’s fellow band member Leigh-Anne Pinnock. MSN.com has since replaced the incorrect image. The outlet’s remaining human staff have been warned that the software could automatically publish the Guardian’s article again, and told to remove the story when it does. However, they were also informed that their AI overlord could overrule their attempts to delete it. [Read: Speech recognition technology is racist, study finds] Thirwell slated the mistake on her Instagram, noting that the two band members were so frequently confused that it’s become a running joke. “@MSN If you’re going to copy and paste articles from other accurate media outlets, you might want to make sure you’re using an image of the correct mixed race member of the group,” she wrote.
Reflecting human bias
The MSN.com error was made by an early version of the AI software that Microsoft plans to use to replace around 50 news production contractors, The Guardian reports. One would hope that human editors would have spotted the error, but the truth is they have a track record of similar mistakes. In February, the BBC wrongly labeled black MP Marsha de Cordova as her colleague Dawn Butler, just a week after the broadcaster featured footage of NBA star LeBron James in a report on Kobe Bryant’s death. Then, in its coverage of the BBC’s blunder, the Evening Standard website mistakenly used a photo of another black female politician.
— I Can’t Breathe…Dawn Butler MP (@DawnButlerBrent) February 3, 2020 Those are just a few of the numerous times that a largely white media has mixed up people of color. Microsoft’s AI is likely reflecting the prejudices of its human creators, as algorithms are predominantly trained on images of white males — yet another example of how AI amplifies racial biases.
