From policy to practice

Artificial Intelligence (AI) is rapidly transforming public administration, but its implementation is shaped by local contexts. This thesis compares AI adoption in Belo Horizonte, Brazil, and Berlin, Germany, examining how administrative traditions influence results. Belo Horizonte, with a hybrid governance system, uses existing tech contracts for cost-effective AI implementation, prioritizing efficiency and citizen engagement, while facing secondary resource and ethical challenges. Berlin, within a strong legalistic tradition, emphasizes data sovereignty and ethical safeguards, but encounters challenges related to bureaucratic fragmentation. Using a comparative case study approach, drawing on semi-structured expert interviews, and a meso-level analytical framework, this research reveals significant differences. Belo Horizonte's approach is pragmatic and incremental, while Berlin's is more cautious and legitimacy-focused. Findings demonstrate that while administrative traditions strongly influence AI implementation, local context and strategic choices are also crucial. The study highlights the need for context-sensitive approaches to study AI adoption in public administration, moving beyond universal solutions and recognizing the importance of factors like institutional capacity, regulation, and public value strategies.