In the society we live in today, we unfortunately see racism and sexism in stores, in our classrooms and on social media. Even though it seems as if we should have left these wrong and disgusting beliefs behind us a long time ago, these acts are still happening. Because these negative beliefs are still around, many believe that they have begun to affect technology and artificial intelligence. Recent data suggests that there may be hints of bias in the training of AI technology.
Artificial Intelligence technology is all around us. AI is a type of technology that is designed to perform skills that are typically done by humans. The technology is able to problem solve, recognize faces and voices, as well as planning and learning.
AI technology is what allows us to unlock our iPhones by holding it up to our face or by using our fingerprint. Snapchat uses the same technology when it uses the filters that distort our faces or adds different objects to the picture. For the AI technology to work, the camera has to recognize the dimensions of your face to accurately unlock the phone or add the Snapchat filters.
This innovative technology has been around longer than many might imagine. It was created in 1955 by Herbert Simon and Allen Newell. Although it has developed greatly since 1955, those who understood what the purpose of this technology was were not a fan of it.
However, after the attacks on United States soil on September 11, 2001, Artificial Technology began appearing in security cameras in airports around the country. Because it was being used for the safety of the American people, AI became more widely accepted.
But with technology this advanced, how can it be designed to be racist and sexist? Several studies have been performed that show lighter-skinned men are more likely to have their phones unlocked by scanning their face with less trouble compared to a woman with a darker complexion. According to Time Magazine, one study found that light skinned men had less than 1% of issues using AI to unlock their phones. However, dark-skinned women had 35% more issues doing the same thing.
How is this happening? When Artificial Intelligence technology is being trained by companies to recognize faces, the majority of the pictures it is shown are of white males. Of course, there are pictures of those with darker skin and of females, but there is a higher percentage of men with light skin tones in these training processes.
There are numerous questions that need to be answered as to why there is not an even number of skin colors and genders being used to train this technology. In one study, according to Time Magazine, the faces of Oprah Winfrey and Michelle Obama were not able to be detected until after a few tries.
Right now, there is no clear answer to the question of if there are racist and sexist motives behind the training of AI technology. However, companies should acknowledge this issue with their products and begin working with those who train AI within the products to make sure all people are included and acknowledged.
Photo by Emma Crouch
