Composite portraits generated by AI, 2020
In collaboration with Laurent Weingart.
In the 1880s, the British anthropologist, statistician and founder of the theory of eugenics Francis Galton began experimenting with what he called “composite portraits”. Galton's composite portrait practice refers to the application of statistics as a sociological tool and reveals a way of seeing based on statistical inference and reduction that can be understood as an antecedent of the representation mechanism of a facial recognition method known as “eigenface”.
Producing an image of a phantom-like blur of multiple overlapped faces, "Disembodied Identities" inquires the eigenface as portrait of machine recognition in relation to a historical backdrop of practices in identification portraiture, and exposes the machinic perception of the "suspect" face through these algorithmic processes often invisible to those under their gaze.
Through its aesthetic approach, the project evokes the problem of "disembodied identities", or the existence of visual representations of individuals circulating independently of their physical bodies. A paradox between clarity and obscurity, it problematizes the ways of seeing of these biometric technologies under development designed to target individual identities, classifying our bodies according to standardized systems conforming to a hierarchy of values such as suspicion and risk.
The "Portraits of Threatening Type" were computed from 600 facial images of profiles of individuals who present a high risk of dangerousness and the "Portraits of Non-Threatening Type" were computed from 506 facial images of profiles that present a low risk of dangerousness.
These profiles come from the classification of thousands of individuals by a "profiling machine" driven by machine learning that detects an individual's ability to handle firearms and predicts their potential danger based on a biometric analysis of their face.