Artificial Intelligence in Public Services: Mutuality in the organisational governance of AI implementation

Artificial intelligence (AI) has become a general-purpose technology in the public sector, offering opportunities to enhance processes, decision-making, and service quality. However, the use of AI also brings significant administrative and responsibility challenges that require multi-actor collaboration and new governance approaches. This dissertation explores the implementation of AI in public services through the lens of mutuality research. Mutuality examines the dynamics of social relationships, multi-actor collaboration, and interdependent relationships between individuals, organisations, and ecosystems. The dissertation is interdisciplinary, combining AI, public administration, and management studies with the theoretical discourse on mutuality, providing a new approach to research on responsible AI governance and implementation. The aim of the study is to increase knowledge of AI implementation in public services, focusing on the context, processes, and agency of public service providers. The growing use of AI in public services emphasises the need to understand how social and organisational processes impact the responsible implementation of technology. The perspective of mutuality helps to understand how administrative and ethical dilemmas in AI implementation—such as tensions between efficiency and inclusion, innovation and accountability, and collaboration and autonomy—can be resolved through multi-actor collaboration. The research is based on four peer-reviewed publications that analyse AI implementation using Finnish public administration cases. The findings are derived from three datasets, consisting of thematic interviews with public service providers, technology providers, and other co-developers. The data analysis employed the approach of constructivist grounded theory (CGT), which allows for a deep understanding of the complex social processes involved in AI implementation. The study aims to answer the question of how mutuality in AI governance can support the responsible implementation of AI in public services. The key findings of the study focus on the interaction between agency (public service experts), structures (the context of AI implementation), and processes (technology implementation within organisations). Artificial intelligence implementation in public service is a multi-phase process, occurring under the pressures of environmental changes, organisational structures, and rapid technological development. The study demonstrates that public service providers need to simultaneously adopt the technology and steer the implementation, which requires them to play an active role in aligning the needs and expectations of various stakeholders. The findings suggest that mutuality is a key factor in the responsible adoption of AI in public services, particularly when structural challenges—such as fragmented regulation and gaps in technological readiness—prevent the creation of unified operational models and open dialogue. Furthermore, regarding implementation processes, the study emphasises the need to create inclusive and iterative approaches that enable continuous learning and adaptation in a rapidly changing technological environment. The research provides new insights into the study of AI governance and implementation in public services through the lens of mutuality. The dissertation contributes to mutuality research by providing an integrative approach to the theories and applying it to a new context. Empirical findings deepen the understanding of how AI implementation is shaped by the specific characteristics of local governance and how the principle of mutuality supports successful collaboration among various stakeholders. Furthermore, the study supports the development of multilevel governance practices that consider the relationships among individuals, organisations, and ecosystems. This also helps address the limitations of organisational AI governance models, such as fragmented decision- making in AI implementation. The dissertation provides insights for public sector decision-makers and practitioners on AI governance and implementation, while also suggesting future research avenues to deepen the understanding of AI governance and the development of multi-actor collaboration in public services.