Moral Responsibility and AI Ethics in a Municipal Health System

Thesis Abstract

The rapid integration of artificial intelligence (AI) tools in municipal health systems raises urgent questions about moral responsibility, accountability, and ethical governance in public health practice. This study addresses the problem of how AI-driven decision support, triage algorithms, and predictive analytics impact responsibility attribution among clinicians, administrators, and policymakers within a metropolitan health department, and how ethical frameworks can guide governance to reduce harm and reinforce public trust. The aim is to examine how moral agency is distributed across human and algorithmic actors and to develop an actionable ethical governance model for municipal health systems. Specific objectives are (1) to identify normative expectations of responsibility from stakeholders (clinicians, data scientists, managers, and patients) in AI-enabled workflows; (2) to evaluate the alignment between existing policies and actual practice in AI deployment; (3) to analyze perceptions of accountability and transparency in AI-generated recommendations; (4) to assess the applicability of moral theories—particularly principlism, virtue ethics, and distributed responsibility theory—in the governance of AI in public health; and (5) to propose a context-sensitive framework for accountability that integrates technical, legal, and ethical dimensions. The study employs a mixed-methods design, combining a cross-sectional survey of 320 health professionals and stakeholders in the municipal system with in-depth semi-structured interviews of a purposeful subsample (n=40) to capture diverse perspectives. Data collection instruments include a structured questionnaire assessing perceived responsibility, transparency, and trust in AI systems, alongside interview guides exploring normative expectations and governance preferences. Validity and reliability of the instruments are established through expert panel review, pilot testing (n=30), and reliability analysis (Cronbach’s alpha ? . Eighty-five percent power is targeted for the quantitative component, with multiple regression to test predictors of perceived responsibility and institutional trust. The qualitative data will undergo thematic analysis following Braun and Clarke’s procedure, complemented by a coding framework aligned with principlism, virtue ethics, and distributed responsibility theory. A convergent parallel design will integrate quantitative and qualitative results, triangulating findings to form a cohesive interpretation. Expected findings indicate that perceived responsibility in AI-enabled decision-making is distributed among clinicians for clinical judgment, data scientists for model adequacy, and administrators for policy compliance, yet accountability tends to be constrained by ambiguous governance structures and opacity of algorithmic processes. The study anticipates significant associations between transparency of AI models, perceived legitimacy of decisions, and levels of trust (p<.05), as well as moderation effects of organizational culture on responsibility attribution. It is expected that principlism (autonomy, beneficence, non-maleficence, justice) and virtue ethics (moral character of professionals) will be invoked to justify governance measures, while distributed responsibility theory will illuminate structural allocations of accountability across equipoises of humans and machines. The research anticipates identifying gaps in current municipal policies, particularly around incident reporting, redress mechanisms, and ongoing audit requirements for AI systems. Contribution to knowledge includes (i) a theory-informed empirical map of moral responsibility in AI-enabled public health practice at the municipal level, (ii) an operational ethical governance framework that integrates risk assessment, transparency protocols, accountability trails, and stakeholder participation, and (iii) a set of policy recommendations tailored to municipal health systems, including governance roles, auditing intervals, patient consent considerations, and workforce training. The study concludes that robust governance anchored in transparent algorithmic design, explicit responsibility assignments, and continuous ethical reflection is essential for maintaining public trust and ensuring equitable health outcomes. Recommendations emphasize the establishment of an AI ethics board, routine algorithmic impact assessments, standardized incident reporting, and capacity-building initiatives that align technical practices with moral and legal norms in the municipal health context.