Human-Agent Care Horizon Glossary
Human-agent care horizon is the Dictionary’s term for the possibility that people and AI agents, working together, can expand the practical support available to those living with cognitive difficulties and to the people who care for them.
The near end of that horizon is ordinary work: organising appointments, preparing questions for a clinician, and assembling a source-linked account of changes reported by a caregiver. These are examples of possible assistance, not evidence that a particular system is clinically effective. The further horizon concerns systems that might support memory, orientation, and continuity of care, provided their benefits and risks are demonstrated in the populations they are intended to serve.
The test is not whether an agent sounds caring. It is whether the arrangement improves daily life without taking away dignity, privacy, choice, or human contact. A mistaken reminder, an invented memory, or an agreeable response that reinforces a frightening belief can undermine the very support the system is meant to provide. Consequential information needs checking, uncertainty must remain visible, and responsibility must stay with identifiable people.
Horizon marks an ambition, not a treatment claim. The useful promise is not a machine replacing the person who cares. It is better support for that person, and more time and attention for the person receiving care.
See also
Background
- World Health Organization, Ethics and governance of artificial intelligence for health (2021). Ethical background, not evidence of efficacy for the proposed applications.