Loneliness, AI, and the Question the ABC Did Not Quite Ask

A response to the ABC News feature on AI in Australian aged care, published 17 April 2026.

On 17 April 2026 the ABC published a thoughtful feature on the emerging AI boom in Australian aged care. It captured the genuine tension in our sector: extraordinary potential on one side, and serious concerns from the eSafety Commissioner, loneliness researchers and academics on the other. It deserves to be read.

It also contained one framing error that, if left uncorrected, will hold the sector back from an intervention that 1.35 million Australians urgently need.

The framing error

Dr Pei-Chun Ko from Monash was quoted saying there is little existing research proving whether AI can reduce loneliness. That statement, presented without qualification, conflates two very different questions.

Question one. Does regular, proactive, warm contact reduce loneliness in older adults?

Question two. Can AI-delivered contact match the effect size of human-delivered contact, at a scale human programmes cannot reach?

Question one is settled science. Question two is the genuine open research question. Conflating them makes a scalable, evidence-based intervention look speculative, when in fact the underlying mechanism has been validated for decades.

What the evidence actually says

The evidence base for regular proactive contact is not thin. It is substantial, rigorous and international.

In 2021, Kahlon and colleagues at the University of Texas at Austin published a randomised controlled trial in JAMA Psychiatry. Two hundred and forty homebound older adults received a four-week programme of empathy-focused telephone calls from briefly trained lay callers. The result: statistically significant reductions in loneliness, depression and anxiety, and a 16 per cent improvement on a standardised loneliness scale. The authors concluded that this is a scalable approach to persistent mental health challenges of older adults.

In 2022, Zhang and colleagues published a systematic review and meta-analysis of 13 RCTs of remotely delivered loneliness interventions in Frontiers in Psychology. The pooled effect was a standardised mean difference of minus 0.41, with a confidence interval that excluded zero. Individual delivery and social-support strategies showed the strongest effects.

In 2024, a systematic review of reviews in Frontiers in Public Health synthesised 19 prior systematic reviews covering 101 primary intervention studies across 21 countries. The finding was consistent: individual-targeted interventions, including telephone contact, have substantive support in the literature.

The United Kingdom's National Institute for Health and Care Excellence has recommended one-to-one telephone befriending as evidence-based practice since 2015. Operational programmes at national scale, including the Canadian Red Cross Friendly Calls Program, have published supporting outcome data.

So what is actually unevidenced?

What is properly unevidenced is whether voice-based AI check-in technology can deliver that same mechanism at the effect sizes demonstrated for human-delivered programmes, and whether it can do so safely.

That is a tight, answerable, fundable research question. It is not the same as asking whether AI can reduce loneliness in principle.

This is the question we are developing with Professor Nigel Lovell, Scientia Professor of Biomedical Engineering at UNSW, for the next Cooperative Research Centres Projects round.

Why this matters for scale

Australia has 1.35 million people accessing some form of aged care. We have a workforce shortage measured in tens of thousands. Ending Loneliness Together's 2023 State of the Nation report found 39 per cent of Australians over 65 feel lonely, and 80 per cent report social isolation.

Human-delivered friendly calling programmes, as effective as they are, cannot scale to that population. Volunteer telephone outreach programmes across the sector are outstanding. They are also, by definition, bounded by the hours that human volunteers can give.

That is worth being direct about.

The safety concerns the ABC raised are real, and addressable

The article raised four specific concerns. Each is addressable, and we have addressed them.

The eSafety Commissioner called for Safety by Design, with safeguards embedded from the outset. Our supplementary protocols covering contentious topics, empathetic response standards, red flag escalation and persona continuity were designed and tested before our first call. A real-time safety layer operates as a trailing stop on every interaction.

Dr Michelle Lim from Ending Loneliness Together asked: who is behind the algorithms? Ours are documented, governed, and will be subjected to independent academic review through the CRC-P.

Dr Lim also noted that AI tends to tell people what they want to hear. Our platform is explicitly governed by clinically informed response standards, not pure agreeableness. This is a design choice, not an afterthought.

And Dr Lim asked how a platform flags when someone is in trouble. Our red flag escalation protocol routes identified risks directly to the care provider's clinical team.

These are the right questions. They deserve rigorous answers, not reassuring platitudes. The CRC-P is how we plan to provide them.

A note on robot companions

The ABC article also highlighted physical robot companions. These are interesting, and I am glad they exist. But physical robots serve communal spaces in residential settings. They do not travel home with people who have left hospital, or reach the widow in regional New South Wales who has not had a visitor in a fortnight.

A phone call does.

What the sector actually needs

Dr Michelle Lim said something in the ABC piece that every AI founder in health should take seriously. These platforms are usually built by engineers, not clinicians or psychologists. She is right.

The response is not to abandon the technology. The response is to build it differently. That means clinical governance from day one. Transparent protocols. Safety-by-design architecture. Real research partnerships with real universities. Live deployments with real care providers accountable for real people.

That is the standard the sector should hold every AI aged care platform to, including ours.

Closing

careplans ai exists because my own father needed the kind of contact that could not always be there in person. That is not a marketing line. It is the reason I left Macquarie University to do this full time.

The Australian sector has an opportunity to do AI in aged care better than anywhere else in the world. Better than the United States, where regulation is thin. Better than the countries racing to put physical robots in nursing homes without clinical evaluation. We have the Aged Care Act 2024. We have the Strengthened Quality Standards. We have research institutions capable of rigorous evaluation. We have the clinical governance culture.

What we need is to stop treating the evidence base as though it starts from zero. It does not. The mechanism is established. The engineering is maturing. The governance is deployable.

The only remaining question is whether we will build this together with the rigour and the courage the moment requires.

I think we will.

References

1. Kahlon MK, Aksan N, Aubrey R, Clark N, Cowley-Morillo M, Jacobs EA, Mundhenk R, Sebastian KR, Tomlinson S. Effect of Layperson-Delivered, Empathy-Focused Program of Telephone Calls on Loneliness, Depression, and Anxiety Among Adults During the COVID-19 Pandemic: A Randomized Clinical Trial. JAMA Psychiatry. 2021;78(6):616–622. doi:10.1001/jamapsychiatry.2021.0113
2. Zhang X, Yu H, Wang Y, et al. The effectiveness of remote delivered intervention for loneliness reduction in older adults: A systematic review and meta-analysis. Frontiers in Psychology. 2022;13:935544. doi:10.3389/fpsyg.2022.935544
3. Interventions for loneliness in older adults: a systematic review of reviews. Frontiers in Public Health. 2024;12:1427605. doi:10.3389/fpubh.2024.1427605
4. Welch V, Ghogomu ET, Barbeau VI, et al. Digital interventions to reduce social isolation and loneliness in older adults: An evidence and gap map. Campbell Systematic Reviews. 2023;19(4):e1369. doi:10.1002/cl2.1369
5. National Institute for Health and Care Excellence. Older people: independence and mental wellbeing. NICE Guideline NG32. London: NICE; 2015.
6. Australian Institute of Health and Welfare. GEN aged care data: People using aged care. Canberra: AIHW; 2024.
7. Ending Loneliness Together. State of the Nation Report: Social Connection in Australia. 2023.
8. ABC News. Australia on the verge of an aged care AI boom but experts warn of high risks. Published 17 April 2026.