An ethics of justice without an ethics of care: Why artificial intelligence ethics guidelines remain unimplementable in community care for older adults

Governments and international bodies have issued a great many artificial intelligence (AI) ethics guidelines over the past decade, yet evidence that they change how AI is built and deployed remains thin. This paper argues that the implementation gap in AI ethics is not only a shortfall of tools. In caring contexts, it is a consequence of moral form. Mainstream AI ethics operates, in its prevailing form, as an ethics of justice: principlist, calculative, and oriented toward what can be formalised. The claim is not that care is absent from its vocabulary. It is that the principles amenable to technical fixes, such as accountability, fairness and privacy, appear to be the ones that reach operational guidance, while dignity, relational autonomy and the prevention of suffering seem more often to be named than operationalised. Community-based care for older adults serves as a critical case. Assessed against established criteria for implementability and usability, the World Health Organization's guidance for health remains practicable in principle but not readily applicable at the point of care for this population, and the harms that matter most here, among them ageism, infantilisation, deception and social isolation, are barely registered. The paper does not call for abandoning principlism but for completing it. It sets out what a care-grounded reading changes in a single case, states what an ethics of care cannot do, and reframes implementation as ethical translation: the movement of a general commitment into the situation of a particular person. Nursing, situated where principle meets the particular patient, may hold a distinctive though not exclusive standpoint from which to contribute to that work. As a theoretical paper, this work involved no human or animal participants and required no ethical approval.