A graduate student from Switzerland’s EPFL technical university has just won a hefty prize of $3,500 for determining that not only does Twitter favor slim and youthful looking faces, but it also favors lighter skinned people with warmer toned features. His name is Bogdan Kulynych, who was awarded the prize after presenting his findings last week.
It’s not often that a company decides to open up their algorithms for scrutiny, but Twitter decided it was time. They sponsored the “saliency” algorithm contest in order to find problems with how they crop photos on your timeline and paid out bug bounties when outsiders found bugs. This is such an uncommon practice now because companies usually put security first before anything else; however, this bounty is another example of why tech giants have been so successful lately, as people are always looking for ways around these heavy-handed systems.
Artificial intelligence is revolutionizing computing by effectively tackling messy and tedious tasks like captioning videos, spotting phishing emails, recognizing your face to unlock your phone. But AI algorithms trained on real-world data can reflect the world we live in: a place where bias manifests itself across all aspects of our lives. Twitter’s bounty program is designed for volunteers who spot these biases to eventually be corrected.
Kulynych found that the target model is biased towards deeming more salient the depictions of people that appear slim, young, of light or warm skin color and smooth skin texture, and with stereotypically feminine facial traits. This bias could result in exclusion of minoritized populations and perpetuation of stereotypical beauty standards in thousands of images.
Kulynych’s system compared the saliency of an original photo to a series of AI-generated variations. He found that faces which appeared younger and thinner were often scored higher.
Twitter Algorithmic bias challenge: https://hackerone.com/twitter-algorithmic-bias?type=team