Abstract

The article focuses on the explainability of Artificial intelligence (Ai) algorithms used in public administrations. It presents agnostic and non-agnostic Explainable ai (Xai) frameworks with the main literature about their development and application and the advantages of the possible deployment of these frameworks to the sociotechnical system employed by public administrations.As a case study, we analyse the narratives of teachers’ X users about the algorithms that assigned school-teacher positions from 2016 to 2023, an algorithmic system that has generated unexpected and potentially problematic effects on society. We argue that the Xai framework can be employed by stakeholders as a guideline for the design of transparent systems by design, to prevent or mitigate the negative effects of these technologies and provide methods and tools for inspecting the processes performed by the automated decision systems.