Abstract
<title>Abstract</title> <p>This paper examines the adoption of artificial intelligence (AI) in the Italian public administration through a mixed-methods analysis of 176 initiatives. Multiple correspondence analysis and cluster analysis are combined with qualitative content analysis of selected case descriptions. The results identify three recurrent patterns of AI adoption: centralized internal learning, citizen-facing communication, and internal analysis and monitoring. These models differ in their degree of organizational embeddedness, which is associated with variations in decision-making autonomy, organizational impact, and governance mechanisms. The findings suggest that AI adoption varies according to the extent to which applications are integrated into core administrative functions, particularly in applications involving core administrative and decision-support functions. Policy implications are discussed.</p>