Abstract
With growing utilisation of digital health technologies for the prevention and management of dementia, this presentation will delve into the ethics of using voice assistants powered by large language models (LLMs), within dementia care. The first part of the presentation will introduce value sensitive designs, complemented by the Ethics by Design for AI framework, to discuss the key principles and ethical considerations for development of such technologies. Secondly, I will introduce our prototype, GRACE, an LLM-based voice assistant that leverages elements of cognitive stimulation therapy to provide cognitive interventions. The aim of GRACE is to help individuals with dementia maintain independence and enable aging-in-place, and thus also improve their quality of life. To conclude, I will present results from two pilot studies conducted with GRACE, among (a) healthy younger adults, and (b) healthy older adults. These pilot studies provide an insight to perceived ethical values and considerations by study participants, and a use-case of how ethical frameworks can be applied to new technologies.

Joining link: medsci.zoom.us/j/91784291038