Twitter will change its automated image cropping feature after users complained that it was biased.
The company says it’s working to “decrease our reliance” on automatic cropping so users can ultimately control how their images appear in a tweet. “We hope that giving people more choices for image cropping and previewing what they’ll look like in the Tweet composer may help reduce the risk of harm,” Twitter writes. “Going forward, we are committed to following the “what you see is what you get” principles of design, meaning quite simply: the photo you see in the Tweet composer is what it will look like in the Tweet.” It’s not clear when these changes might be implemented.
Trying a horrible experiment...— Tony “Abolish (Pol)ICE” Arcieri 🦀 (@bascule) September 19, 2020
Which will the Twitter algorithm pick: Mitch McConnell or Barack Obama? pic.twitter.com/bR1GRyCkia
Twitter first introduced the image cropping feature in 2018, saying it was meant to highlight “salient” areas of the image by using neural networks to predict where someone is “likely to look.” More recently though, a number of users called it out after a viral tweet pointed out that the cropping consistently picked out lighter-skinned people, regardless of the framing of the original image. Other users soon followed up with their own version of the “experiment,” including images of cartoon characters and even black and white dogs.
In a blog post, Twitter said that it had conducted its own tests and that “our analyses to date haven’t shown racial or gender bias,” though “we recognize that the way we automatically crop photos means there is a potential for harm.” The company also says it plans to do additional testing and is “exploring ways to open-source our analyses so that others can help keep us accountable.”