Over the weekend, Twitter exploded with discussions over racial bias. According to some Twitter users, the micro-social site exhibits racial bias in photo previews. It’s not clear what’s been happening but apparently, Twitter favors white faces over blacks or darker ones. Many people have shared their “experiments” while others were also trying to provide explanations. Twitter uses AI and now they are being called racist. The issue is that Twitter displays people with different skin tones differently. It’s actually not just Twitter as the issue started with an education tech researcher discovering that Zoom shows people differently. Those with dark tones have their heads cropped out of their photos.
People tried the same on Twitter and some show apparent “bias”. There is the question now whether algorithms can be racist or sexist.
Researcher Colin Madland went to Twitter to discuss the problem. Photos of a white guy and former US President Barack Obama were put together in one image–one face on top with the other one at the bottom. People wanted to see what face will be shown on the preview. The first result showed the white guy.
A lot of people tried the same but the results differ. It’s not really clear how this came to be but we can only think of AI. Twitter showing faces of white people more frequently is racist according to many people but we’re pretty sure there is an explanation.
Trying a horrible experiment…
Which will the Twitter algorithm pick: Mitch McConnell or Barack Obama? pic.twitter.com/bR1GRyCkia
— Tony “Abolish (Pol)ICE” Arcieri 🦀 (@bascule) September 19, 2020
Twitter made a promise to investigate the issue. A large sample is needed for this. At the moment, it is premature to judge a platform based on a few examples. We won’t be surprised though if the result of experiments or investigations will be the same because after all, algorithms and any other automated system usually reflect the biases of the human makers.