Testing AI Tools in Public Governance: First Reflections of the AI4Gov Case Study

The integration of artificial intelligence (AI) into public governance presents opportunities to enhance decision-making processes, optimise resource allocation, and build citizen trust. This work-in-progress paper examines preliminary findings from the AI4Gov project, which explores the potential of AI and big data technologies to support evidence-based policymaking while addressing ethical challenges such as bias, transparency, and trust. The AI4Gov project features diverse use cases (UCs) tailored to address pressing governance challenges across multiple domains. In Spain, the Deputación Provincial de Badajoz pilot focuses on sustainability and water management, leveraging predictive analytics to optimise drinking and sewage water systems, enhance operational efficiency, and inform long-term investment strategies for regional water cycle management. The Greek pilot, led by the Municipality of Vari-Voula-Vouliagmeni in collaboration with the Ministry of Tourism, emphasises tourism and urban management, including innovative solutions for traffic violations and waste management. These efforts aim to improve municipal services, reduce environmental impact, and enhance citizen satisfaction. Meanwhile, the Slovenian pilot, guided by the Josef Stefan Institute, has an international scope, applying AI to monitor the Sustainable Development Goals (SDGs) and creating tools for bias detection, fairness evaluation, and policy alignment. Collectively, these use cases demonstrate the adaptability of AI4Gov technologies in addressing local and global governance challenges, from sustainability and urban management to ethical AI deployment in policymaking. In its first validation phase, AI4Gov employed tools like the User Experience Questionnaire (UEQ) and trust-focused methods to evaluate the usability, functionality, and societal acceptance of its AI solutions. Findings highlight varied user experiences and perceptions of trust across different domains and stakeholder groups. While pragmatic qualities, such as ease of use and efficiency, scored positively in many use cases, there were also some suggestions for improvements. Hedonic qualities, including excitement and innovation, consistently garnered favourable feedback, reflecting the perceived potential of these technologies. Trust emerged as a critical topic. Participants acknowledged the promise of AI in enhancing public services but expressed apprehension about data privacy, algorithmic bias, and the representativeness of datasets. For instance, trust-building exercises revealed conditional trust dependent on transparency, accuracy, and iterative testing. These insights emphasise the necessity of addressing ethical and operational challenges to ensure public confidence in AI-driven governance. As AI technologies evolve, the AI4Gov project outlines key strategies for refinement, including enhancing data quality, expanding stakeholder engagement, and improving the tools’ responsiveness and inclusivity. The second validation phase aims to integrate these learnings to foster broader acceptance and applicability of AI solutions in governance. By situating these reflections within the broader discourse on data-driven politics, this work-in-progress paper contributes to understanding the intersection of political methodology and emerging technologies. It argues that while AI offers transformative potential, its adoption in public governance requires a delicate balance of technological innovation, ethical rigor, and public trust. These initial findings not only provide a framework for further iterations of the AI4Gov project but also offer valuable insights for policymakers, AI experts, and researchers navigating the complexities of AI in governance.