An Ethical Framework for the Implementation of Elderly-Care AI Assistants

Abstract: The rapidly expanding capabilities of large language models like GPT-4 present opportunities for improved healthcare quality and cost effectiveness. Given shifting demographics, a domain of particular interest is elderly care, which additionally is resource- and personnel-intensive. However, using artificial intelligence (AI) to provide necessary medical care, or even simply as an adjunct to this care, raises serious ethical questions. Bioethicists must be proactive in defining the criteria AI technologies should meet to be ethically implemented. Through a narrative review of best practices in geriatric medicine, AI technology and ethics, including the principles articulated in the US Office of Science and Technology Policy AI Bill of Rights, we have developed a model for an idealized elderly-care AI assistant. We consider three primary roles: (1) medical caregiver, (2) daily-living assistant and (3) emotional companion. We review the role-dependent responsibilities of the AI assistant in light of existing applications and current research directions to realistically accommodate future AI assistant capabilities. Next, we present four areas of ethical concern, each layering upon the prior: (1) safety and functional reliability (2) equity and accessibility (3) implications for social and legal organization, and (4) implications for human relationships, perceived responsibilities and the self-concept. We distill these concerns as a generalized philosophical and ethical framework for the responsible implementation of AI assistants and AI-supported medical technologies beyond elderly care.

Gyan Moorthy, MS – University of Pennsylvania Perelman School of Medicine