Algorithm Bias

I never knew that an algorithm could have bias, let alone what an algorithm was until watching the various videos for this weeks discussion. The first video I watched was a Ted Talk from Joy Buolamwini where she described the “coded gaze”, which is similar to the male gaze. The coded gaze is the discriminatory ways in which facial recognition was made. Joy is an African American female whose face could not be detected by the webcam. She tested this problem by putting on a white face mask which was detected without any problem. This proves Joy’s point that faces that deviate from established norms are harder for algorithms to detect. This is problematic based on the usage of facial recognition among law enforcement. This can cause wrong facial recognition and arresting an innocent civilian. In the second video, also by Joy, she makes a spin off of Sojourner Truth’s iconic speech called “Ain’t I a Woman” but makes it an ode to algorithms not being able to recognize a female face. This video went on to show proof that algorithms have recognized iconic female faces as males or not recognize them at all. The last video from Joy described the experiment she derived to test if it was only African American females whose faces were unrecognizable. She tested this by using a variety of races and both males and females then used IBM, Face ++, and Microsoft. Her results showed that each company did better detecting males faces from females as well as lighter skinned people rather than darker. The final video I watched was eye opening to the little acknowledgement women received after doing a majority of the work to make the first computer. Later on, we learn that computer programing became women’s work yet only 18% of people who major in computer science are females. This statistic connects to the data we were given about women’s existence in the workplace from the National Center for Women Information Technology. By 2017, only 26 percent of computing workforce were women. These statistics remind me of the scene in ‘Hidden Figures” when the women understood technology more than men but got little recognition for their work. Overall, algorithms have bias just the same as the workplace.