AI-judged beauty contest picks disproportionate number of white winners
In September 2016, an algorithm assigned to pick the winners of a beauty contest examined selfies sent in by 600,000 entrants from India, China, the US, and all over Africa, and selected 44 finalists, almost all of whom were white. Of the six non-white finalists, all were Asian and only one had visibly dark skin. The contest was run by Beauty.ai, an initiative from the Russia and Hong Kong-based Youth Laboratories, and was supported by Microsoft and Nvidia. The reason was the lack of diversity - of both people and opinions about them - in the databases used to train the deep learning algorithm used to judge the contest. Additional factors were that 75% of the contest entrants were European and white, and the judging algorithm sometimes rejected selfies of dark-skinned people if the lighting was too dim. The problem's roots can be traced much further back into the history of photography, however, as camera film was originally designed to perform better with white skin.
Writer: Jordan Pearson