Twitter faced some criticism last year after some users noticed that its image cropping algorithm appeared to show bias, with a preference of showing images of white people.
In a new post on Twitter’s official blog, the company admitted that its algorithm was indeed biased, and shared its analysis of its image cropping algorithm CNN reports. Software engineering director Rumman Chowdhury revealed that Twitter came to the conclusion that the algorithm favored white people over Black people, and women over men. The company first admitted that it had some work to do on the algorithm last year.
“We tested for bias before shipping the model & didn’t find evidence of racial or gender bias in our testing,” said the company in a statement on the matter in 2020, replying to an example of the algorithm favoring Mitch McConnell over Barack Obama in two seperate images. “But it’s clear that we’ve got more analysis to do. We’ll continue to share what we learn, what actions we take, & will open source it so others can review and replicate.”
The analysis concluded that there’s around an 8 percent demographic parity favoring women, and 4 percent favoring white people. “Machine learning based cropping is fundamentally flawed because it removes user agency and restricts user’s expression of their own identity and values, instead imposing a normative gaze about which part of the image is considered the most interesting,” the research reads.
In March, Twitter made a change on its mobile apps that would show pictures in full instead of cropped. Crowdhury noted that going forward, the company is looking to depend less on machine learning for certain functions, specifically anything the company agrees “is best performed by people using our products.” The image cropping algorithm is set to be removed from Twitter soon, although an exact time frame was not provided.
