AI Facial Profiling, Levels of Paranoia is an interactive installation and a “physiognomic machine”. Both fascinated and critical towards recent psychometric research this artwork uses pattern recognition to predicts the potential danger of an individual based on the biometric analysis the face. The artist refers to research papers or companies such as Faception which claim to be able to recognize a person’s tendency to be or become a terrorist, criminal, gay or academic based on their face.
For this installation the artist used machine learning models presented in the papers and a dataset she scraped and collected from YouTube videos where a person handles a gun. The faces were extracted from those videos and used to train the category of persons portraying a high risk of being a threat. The category of low risk was trained on a set of selfies.
The visitors who get classified by the machine are scanned with a gun like devices representing the violence such surveillance causing feelings of terror rather than evoking a secure space. The classification result is stamped on a document with the taken image and the person is declared one of the binary options either HIGH or LOW risk. Then the document continues to move forward on a blue belt, familiar from airport security, and is sorted in to one of two transparent boxes. The persons face and classification is visible for the visitors until the next documents classified into that category cover the previous images taken by the installation.