Transparency in public administrations in the age of artificial intelligence: concepts and key actions

Abstract

The emergence of artificial intelligence (AI) in all areas of human activity necessitates a redefinition of the concept of transparency in public administrations. The opacity of algorithms can introduce and amplify cognitive, ethical, and cultural biases, influencing information, sources, and results, as well as automated decisions based on AI tools made by public administrations. This affects a wide variety of instruments used by institutions, from classification algorithms for allocating public aid to tools for medical diagnosis. To guarantee the transparency of their operation, the State must undertake the task of analyzing, cataloging, classifying, and controlling the algorithms it uses, and making this information available to citizens. This is further complicated by the fact that, in many cases, understanding these systems requires technical knowledge that the general public lacks. This paper presents this problem, along with some key action points and success stories.