Exploring human–AI collaboration for management control in public healthcare

Purpose This study examines how controllers in public healthcare envision human–AI collaboration in management control when dealing with uncertainty, complexity and equivocality, focusing on anticipated advantages and implementation challenges in an early-adoption setting. Design/methodology/approach We conducted an exploratory qualitative study with 14 controllers from Italian public healthcare organisations. Four closed-ended questions provided a preliminary picture of current AI adoption, while three scenario-based open-ended questions elicited prospective assessments. The framework method was used to analyse the data. Findings AI adoption remains limited and uneven, and is rarely embedded in everyday management control. Under uncertainty, controllers link AI to better forecasting and faster access to information, but worry about data quality, privacy and overreliance. In complex settings, they expect AI to help integrate diverse information and enable more systemic control, provided processes are standardised and human oversight remains. In equivocal situations, AI may support interpretation and communication, but its impact is limited by limited hands-on experience and the difficulty of codifying contextual nuances. Originality/value Based on the Jarrahi (2018) decision-making scenarios, the study develops a tentative conceptualisation of how public healthcare controllers envisage AI collaboration. The concerns of the participants inform three proposed arrangements: validation, recursive integration and contestable sensemaking. Their formulation and the accountability-augmentation principle are theoretical elaborations informed by prospective evidence and accountability research.