"This article explores the change in calculation methods induced by deep learning techniques. While more traditional statical methods are based on well instituted categories to measure the social world, these categories are today denounced as a set of hardened and abstract conventions that are incapable of conveying the complexification of social life and the singularities of individuals. Today AI models try to overcome some criticism raised by rigid social categories by combining a "spatial and temporal expansion" of the data space, producing a global transformation of the calculation methods."

Support: Digital