Comfort, Truth and AI: Moral Boundaries of Deception in Dementia Care
Lunchtime Sessions 1
17 November, 2026 1:20 – 2:05 pmElectronic Posters
Jennifer MacRitchie
Senior Research Fellow
University of Sheffield
Summary
Dementia care often involves difficult tensions between truth-telling, emotional comfort, autonomy, dignity, safety and carer burden. These tensions become more complex as conversational AI systems, virtual humans and companion technologies enter care settings. Practices such as reassurance, distraction or “therapeutic lying” are already ethically contested when performed by human carers, but less is known about whether similar actions are judged differently when delivered by AI.
This poster presents the design of a planned vignette-based survey exploring public and carer judgements of deceptive or potentially deceptive AI behaviours in dementia care. Participants will respond to short scenarios in which a person living with dementia becomes distressed, confused or at risk. Across scenarios, we will vary whether the response is provided by a family carer, professional carer or AI system, and whether the action involves a comforting falsehood, simulated companionship, concealed monitoring, or a conflict between present comfort and previously expressed wishes.
The survey will examine perceived acceptability, trust, dignity, autonomy, safety, care and willingness to use such systems, alongside open-text responses about participants’ reasoning. The aim is to identify where people draw moral boundaries around AI-mediated deception in dementia care, and what values shape these judgements. The poster will outline the study design, ethical considerations and implications for responsible development of AI companions and virtual humans that support care without undermining personhood, consent or trust.