The algorithm is not an oracle Predictive systems, youth, and inclusion: The case of debt

The article presents an ethnographic case study on mortgage bank lending practices and discusses the method of assessing individual applicant behaviour, with a specific focus on young people. Creditworthiness assessment is an important issue to study in terms of the use of predictive systems in decision-making processes and their effects in terms of social inclusion. The paper shows how specific social representations regarding the creditworthiness and reliability of people are embedded in customer risk assessments in Italian banks. It also highlights how these embedded representations have performative power in defining the boundaries of access to credit, and how branch operators attempt to negotiate the algorithm-defined assessment when it does not match their view of the social assessment. The representations embedded in the algorithms have considerable constraining power in differentiating the opportunities for certain social categories. These tools have been constructed according to the historically situated behaviour of people with creditworthiness characteristics defined by stability, which does not seem to fit with the actual behaviour of people with different characteristics framed in changing contexts.