Both zoom and Twitter came under fire this weekend for their respective problems with algorithmic bias. Zoom is an issue with the virtual backgrounds of the video conferencing service, and Twitter is: This is an issue with the website’s photo crop tool.
It started when Ph.D. Student Colin Madland tweeted about a black faculty member’s problems with Zoom. According to Madland, Zoom removed the head whenever the faculty member used a virtual background.
“We reached out to the user directly to investigate this issue,” a Zoom spokesperson told TechCrunch. “We strive to provide a platform that includes everyone.”
However, when discussing this issue on Twitter, the algorithmic bias issues heightened when Twitter’s mobile app only previewed the image of Madland the white by default.
“Our team tested for bias prior to shipping the model and our testing found no evidence of racial or gender bias,” a Twitter spokesman said in a statement to TechCrunch. “However, these examples show that we need to do more analysis. We’ll keep sharing what we’re learning, what action we’re taking, and making our analysis open source for others to review and replicate. “
Twitter pointed to a tweet from his chief design officer, Dantley Davis, who was doing some of his own experiments. Davis found Madland’s facial hair affected the outcome, so he removed his facial hair and the black faculty member appeared in the cropped preview. In a later tweet, Davis said he was “just as irritated about it as everyone else. However, I am able to fix it and I will. “
Twitter also pointed to an independent analysis by Vinay Prabhu, chief scientist at Carnegie Mellon. In his experiment, he wanted to find out whether “the tendency to crop is real”.
In response to the experiment, Parag Agrawal, CTO of Twitter, said it was “a very important question” to address the question of whether cropping bias is real. In short, sometimes Twitter cuts out blacks and sometimes it doesn’t. But the fact that Twitter does it at all is enough to be problematic.
It also addresses the larger problem of the spread of bad algorithms. The same types of algorithms lead to biased arrests and imprisonment of blacks. They are also the same algorithms as google Photos of black people used to be called gorillas, and Microsoft’s Tay bot used to become a white supremacist.