Exploring AI's creative potential in dementia care
Memory is a fundamental part of what makes us human, shaping our identity and connecting us to our past. But for millions worldwide, dementia erodes these connections, creating profound challenges. The Synthetic Memories initiative is a research collaboration between the artist and researchers from the studio Domestic Data Streamers, Alex Mihailidis — Scientific Director of AGE-WELL and Professor at University of Toronto — and Google Arts & Culture. It explores whether art and Generative AI tools such as Google's image and video generation models can aid therapeutic engagement with senior citizens.
AI as a tool to support reminiscence therapy
The approach is based on reminiscence therapy, a technique that encourages individuals with dementia to share their personal experiences and memories. Traditionally, this involves using props like photos, music, or objects to improve mood, stimulate mental activity and enhance cognitive function. Synthetic Memories explores the potential of a new prop: using AI to generate synthetic memories. In cases where memories lack tangible records or have limited visual and auditory aids like photos and music, AI may offer new ways to enhance reminiscence therapy by generating a wider range of detailed and personalized recollections. These AI-generated visualisations are intended to complement existing reminiscence therapy techniques and can be revisited and explored repeatedly, offering a more interactive experience. It's important to note that these visualisations are not intended to be perfect recreations of the past, but rather prompts to aid in recalling memories and sparking conversation.
Edward and the Domestic Data Streamers Team working together to create prompts to create and fine-tune Synthetic memories
The team working on prompts that capture Edwards descriptions
A collage of Synthetic memories
How it works
For this initiative, we met with Edward, a research participant who has Alzheimer's disease, to capture his personal recollections. We began by interviewing him to collect detailed descriptions from his past. These descriptions were then entered into Google’s ImageFX and VideoFX tools, with additional inputs to create images and videos that have a dream-like quality and evoke the essence of those memories. We worked with Edward to select and refine these synthetic memories of his past until he was satisfied with the visualisations.
One result was a generated visual reflecting Edward's memory of hitchhiking on a roadside in Spain with his friend. The image, refined with his input, incorporated details he described, such as wearing his leather jacket and carrying a rucksack.
By exploring the intersection of memory, AI, and artistic expression, this initiative opens new possibilities for reminiscence therapy. As the research progresses, we aim to better understand how technology and creativity can work together to support memory, storytelling, and human connection.
This research is one of several initiatives where Google Arts & Culture explores the role of arts in health and wellbeing; discover more on g.co/artshealth