This video was made by feeding hundreds of images of pharmaceutical pills into an AI. This type of early AI was used by government agencies to create discriminatory profiles of “types” of people; for example, to create facial composites from the mugshots of all people arrested for a certain type of offense in New York City within a certain timeframe. Obviously such facial composites would ignore countless factors, including the discriminatory practices of the NYPD targeting POC, to the point where 80-90% of all arrests are of people of color (mainly black and hispanic).
Averages created from non-objective data would only reflect the biases of the person that trained the particular AI model and of the NYPD, which arrests POC at 8-9 times the rate at which they arrest white people.
Facial recognition software has been proven to be mistaken on a much higher rate when it comes to people of color and women, particularly black women. (Algorithmic Learning Institute, MIT, Joy Buolamwini, “AI, Aint I a Woman”). I used this facial recognition AI but fed it pictures of pharmaceutical drugs instead of faces. The AI then generated “averages” of what a drug should look like. I do this to show the absurdity of both situations. Both “averages” can look like anything and are just reflections of the opinions of the person that trained the AI model, and of the people that collected the data that was fed to the AI. It questions the black and white perspectives of addicts being inherent criminals, and questions absolutes like minimum sentences. The piece is an abstract commentary on the implicit racism in the War on Drugs (https://vimeo.com/530638829).
Averages created from non-objective data would only reflect the biases of the person that trained the particular AI model and of the NYPD, which arrests POC at 8-9 times the rate at which they arrest white people.
Facial recognition software has been proven to be mistaken on a much higher rate when it comes to people of color and women, particularly black women. (Algorithmic Learning Institute, MIT, Joy Buolamwini, “AI, Aint I a Woman”). I used this facial recognition AI but fed it pictures of pharmaceutical drugs instead of faces. The AI then generated “averages” of what a drug should look like. I do this to show the absurdity of both situations. Both “averages” can look like anything and are just reflections of the opinions of the person that trained the AI model, and of the people that collected the data that was fed to the AI. It questions the black and white perspectives of addicts being inherent criminals, and questions absolutes like minimum sentences. The piece is an abstract commentary on the implicit racism in the War on Drugs (https://vimeo.com/530638829).