The strategic use of AI in the public sector: A public values analysis of national AI strategies

Abstract

Governments worldwide are strategically investing in artificial intelligence (AI) to improve public services and streamline internal operations. In this context, national AI strategies play a pivotal role. This study uses combined qualitative research methods analyzing 28 national AI strategies (i.e., the texts). Our aim is to delve into how governments define and position AI applications within the public sector. Specifically, the study explores how the texts convey AI's application in this context employing a public value(s) perspective. Its discursive analytical approach coupled with a comprehensive take on public value theory (Moore, 1995) engenders novel insights into national discourses on AI in the public sector. Against this background we draw on public administration and policy research in our analysis of three dominant discourses that we identify in the texts, i.e. empowerment through information, enhanced administrative practices, and improved service delivery. We find that the discourses involve different positions in relation to governments' use of AI and depend on particular actors and types of public service. Commonly, they concern government objectives to tackle critical societal issues through AI, such as in the areas of health and social care and employment. In particular, the discourse of enhanced administrative practices commonly positioned AI as a tool to optimize internal processes, resource allocation, and organizational management. On the other hand, the discourse of improved service delivery similarly placed public services front and center, while the discourse of empowerment through information framed AI as being able to enhance citizens' service experiences. Interestingly, discourses emphasizing the policymaking function, i.e., AI applied to the development of public policy,—receives limited attention. Our findings underscore strategic prioritizations. While efficiency and service delivery dominate the discourse, citizen engagement remains underemphasized. We argue that policymakers must strike a balance, ensuring AI aligns with broader societal outcomes while addressing democratic imperatives.