Artificial intelligence to support public sector decision-making: the emergence of entangled accountability

Abstract

Public organizations adopt Artificial Intelligence (AI) to streamline decision-making processes to improve rationalization and efficiency of service provision. However, several cases of AI deployment have generated doubts and questions among public audience due to distortions emerged in the services delivery. Hence, increasing calls for holding AI accountable have been raised. Building on this stream of research, we posit that the deployment of AI changes the decision-making processes it informs. We show that public organizations adopt AI, it entangles with the legal and administrative rules that underpin organizations to structure the decision-making process. These techno-legal entanglements alter the decision-making process and hence the accountability of the public organizations. To shed light on these transformations, we rely on evidence from two selected cases of AI adoptions: UKVI in the UK and COMPAS in the US. We theorize the emergence of an entangled accountability in which responsibilities are shared between the machine and the human contribution in the decision-making process of public organizations.