Twitter is learning first hand about the challenges of eliminating racial bias in algorithms. Liz Kelley, a spokeswoman for social media, said the company needed to do "more analysis" after the cryptographic engineer Tony Arcieri conducted an experiment that suggests that Twitter's algorithm was biased in ranking photos.
When Arcieri attached photos of himself Barack Obama and Mitch McConnell in 2 different tweets, placing the photos in different positions each time, Twitter appeared to display only face of McConnell. In fact, Obama only appeared when the engineer reversed the colors of the two photos, resulting in no question of Colour of skin.
Others tried to reverse the series of photos and names to no avail. A smile with a higher contrast worked, according to what Kim Sherrell found Intertheory. Scientist Matt Blaze, meanwhile, noted that the priority seemed to be different depending on the application Twitter used. For example, the Tweetdeck it was a little more neutral.
Kelley said the social media site had checked for racial discrimination before using the current algorithm, but had not found any data at the time. He added that the platform would open the algorithms for further study and help others "review and reproduce", according to Engadget.
Of course, there is no guarantee that Twitter can fix it issue. However, the experiment shows the real dangers of algorithmic discrimination, regardless of the intention.
It could take people away from the limelight, even if it is the focus of a social media post. network or linked to a news article. You may have to wait a long time before such issues become rare.