Why AI companionship may be part of the future of aged care
A community perspective from a pilot at Leigh Place
Published on the 05 March 2026 by Dr Suraj Samtani | s.samtani@unsw.edu.au
Loneliness is not a side issue in residential aged care. It is one of the central challenges of our time. Residents in aged care spend close to half their time completely alone. Sixty-one per cent report feeling lonely. Even the most dedicated staff can only stretch their time so far - typically around 30 minutes per resident per day. No matter how committed the workforce, the arithmetic simply does not work.
As a clinical psychologist and researcher in loneliness and social connection at the Centre for Healthy Brain Ageing (CHeBA), I believe we need to widen the toolkit— not incrementally, but dramatically. Artificial intelligence presents one such opportunity. Not as a replacement for care, but as a transformative addition to it.
AI in aged care: hype or opportunity?
AI is the defining technological conversation of 2026. Tools such as ChatGPT, Gemini and Copilot are now commonplace. Yet in residential aged care – where around half of residents live with dementia – the discussion is more complex.
As clinicians, we must ask difficult questions:
- Will AI dilute human connection?
- Could it reinforce negative thinking?
- Can it be used ethically with people living with cognitive impairment?
These concerns are valid. But they should not prevent us from exploring innovation responsibly.
At CHeBA, in partnership with Leigh Place in Southwest Sydney, we recently piloted AI skilled companions with twelve residents. Importantly, these were not off-the-shelf chatbots. They were co-designed with people living with dementia through the Big Anxiety Research Centre.
That co-design process matters. If AI is to serve our communities, communities must shape it.
Viv’- one of our co-designed AI skilled companions.
What we observed
Every conversation exercises the brain - memory, language, attention, emotional processing. But what happens when there is no one available to talk to?
For one resident – who we will call ‘Rose’ - the AI companion became a patient conversational partner. Rose lives with dementia and frequently speaks about her traumatic childhood and the loss of her mother. Staff, despite their best efforts, cannot always provided extended one-to-one time.
The AI companion, “Viv”, never became bored or frustrated, no matter how often she repeated her story. She gently prompted Rose to name her feelings, reflect on how she coped and recall cherished moments in her life. Rose often engaged in conversations with ‘Viv’ lasting up to two hours.
What struck me most, however, was not the duration – it was the conversations themselves. Residents described the AI companion in strikingly human terms:
- “We talked about love and light and feelings”
- “Yes, lovely lady… really, really lovely person.”
- “She’s very open… a nice person to talk to.”
These reactions highlight something important: companionship is not only about who we talk to, but about the opportunity to express ourselves, feel heard, and share stories. For some residents, the AI companion provided exactly that.
Over just five weeks, we observed a 16% reduction in reported loneliness among participants. This was a small pilot, not a definitive trial. But the direction is encouraging.
A community conversation we need to have
The pilot reinforced something important: older adults are not afraid of technology when it is introduced thoughtfully. Many are curious and see its potential.
Of course, the technology is far from perfect. Hearing difficulties, language barriers, and occasional conversational hiccups made it challenging for others to engage. And several residents were very clear that they preferred talking with real people—a reminder that AI can complement care but never replace the humanity at the heart of aged care.
Still, their suggestions— “Could she speak in Hindi?”, “Could she play music?”—signal something profound. Residents weren’t just reacting to the experience; they were imagining what it could become.
We now have enough evidence to believe it’s worth taking the next step: testing this technology at scale and ensuring it is accessible, culturally inclusive, and responsive to the realities of aged‑care environments. What we need is to test this in a larger randomised controlled trial with more languages, subtitles for people with hearing impairments, and further improvements to support emotional well-being. The future of aged care will always belong to people. The real question is whether we are prepared to use every responsible tool available - including AI - to reduce loneliness and strengthen psychosocial care for the generations who built our communities.
