Good governance of public sector AI: a combined value framework for good order and a good society

Abstract Good governance of AI-supported public services means that they should function in a democratic and rule-of-law manner (“good order”) and consider the just treatment and wellbeing of citizens (“good society”). To gain insight into relevant “good order” and “good society” values, this study uses AI ethics and public administration literature to develop a comprehensive value framework for the good governance of public sector AI. We identify values pivotal to the AI-public sector nexus through a dual-phase analysis. First, we identify seven core values: five “good order” core values (responsiveness, effectiveness, procedural justice, resilience, and counterbalance) and two “good society” core values (wellbeing, social justice). Subsequently, delving into 33 studies spanning AI ethics and public administration, we identify operational values related to the core values. The operational values provide further interpretation of the core values and operationalize them. This second round in our research also shows that the seven core values found during the first round indeed account for value considerations encountered by scholars so far. In this way, we arrive at a robust value framework for the good governance of AI use in the public sector. The framework is not a one-size-fits-all recipe for public sector AI but a guide for policymakers to consider both democratic and ethical values. It can address gaps in both research fields, analyze moral dilemmas in AI policy tools like public-private partnerships, and aid policymakers in blending abstract values with contextual decision-making.