Bias is not a bug of AI. We constantly capture, classify, analyze & normalize ourselves and each other online. Normalizi.ng invites you to reflect on the past and present of this normalization.
I was generously commissioned by the Science Gallery Dublin to create an online version of The Normalizing Machine – my installation exploring issues of social normalcy, machine learning and bias. The new work titled Normalizi.ng launched in October at the BIAS exhibition at the Science Gallery Dublin and have recently opened at the Art Code exhibition at Solyanka Ground in Moscow.
Normalizi.ng is an experimental online research in machine-learning that aims to analyze and understand how do we decide who looks more “normal”. When you visit www.Normalizi.ng on your phone you will be asked to take a selfie and then be presented with a series of previously recorded participants. You will then choose who looks more “normal” between pairs of noses, mouths, eyes, and faces. The machine will analyze your decisions and your face and then add you to it’s algorithmic map of normality.
In the late 19th Century the French forensics pioneer Alphonse Bertillon, the father of the mugshot, developed “Le Portrait Parle” (the speaking portrait) a system for standardizing, indexing and classifying the human face. His statistical system was never meant to criminalize the face but it was later widely adopted by both the Eugenics movement and by the Nazis to do exactly that.
The online work automates Bertillon’s speaking portraits and visualizes how today’s systematic discrimination is aggregated, amplified and conveniently hidden behind the seemingly objective black box of Artificial Intelligence.