ALGORITHMIC DISCRIMINATION AND PROTECTION OF VULNERABLE CONSUMERS IN ACCESS TO CREDIT

Abstract

Consumer credit carries risks of discrimination against certain categories of customers. This was originally due to people's involvement in evaluating credit applications. As a consequence, since the 80s intermediaries have used automated data processing systems, which are today complemented by AI models. Despite the several advantages, AI entails the risk of discriminatory consumer classification. This paper proposes an analysis of credit-scoring AI systems under the spectrum of Italian and EU legislation to understand whether they provide an adequate level of protection against the risk of algorithmic discrimination, also from a remedial point of view.