Reconstructing human agency in the age of artificial intelligence: a religious ethics framework for algorithmic decision-making

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Artificial intelligence (AI) increasingly shapes consequential decision-making while leaving unresolved how human moral agency should operate when computational systems acquire growing functional and practical influence. Contemporary AI ethics provides substantial accounts of fairness, accountability, oversight, flourishing and responsibility, yet a more specific problem remains: functional agency, moral agency and normative authority may become separated when algorithmic outputs acquire de facto action-guiding weight. This study develops the Religious Ethics Framework for Algorithmic Decision-Making (REF-ADM) through comparative religious-ethical conceptual analysis, philosophical hermeneutics and normative framework construction across Christianity, Islam, Judaism, Buddhism and Hinduism. The analysis identifies normative convergence without conceptual equivalence and derives eight differentiated capacities: dignity, conscience, intentionality, responsibility, justice, compassion, stewardship and relational accountability. REF-ADM organises these capacities within a recursive five-stage architecture of algorithmic input, ethical interpretation, intentional moral deliberation, responsible implementation and relational accountability with ethical learning. Its originality lies not in proposing novel ethical values, but in systematically organising them around moral displacement when functional influence and de facto normative authority increase without a corresponding transfer of moral agency. REF-ADM is advanced as a middle-range normative framework for high-stakes AI-assisted decision-making; its propositions specify requirements of accountable judgement and remain open to subsequent empirical examination.