Can an AI companion help with loneliness? What it can and can't do

Possibly a little, for some people, and with real caveats. Studies of social robots, voice assistants and chatbots with older people have found modest falls in loneliness on average. But most were small and short, few compared users with similar people who had no companion, and some of the most quoted figures come from the companies that make the products. A companion can offer conversation, company and a little routine between calls and visits. It cannot replace the people in someone's life, and it cannot treat depression.

This article sets out what the research found, what nobody knows yet, the risks worth taking seriously, and how to tell whether a companion is helping. It is written for families and friends, and for anyone thinking about a companion for themselves.

What the studies have tested

Researchers have tried three broad kinds of companion with older people:

• Social robots. Physical machines that talk to people or respond to them, such as Paro and ElliQ.
• Voice assistants. Smart speakers such as Amazon's Echo, used as a daily conversation partner.
• Chatbots and on-screen characters. Software on a screen that people talk or type to, sometimes shown as an animated character.

Most of this research used simpler technology than today's AI chatbots. In the most recent review, only one of 17 studies used the kind of large language model behind tools such as ChatGPT.

What the reviews found

That review, published in Psychological Medicine in 2026, gathered 17 studies of conversational agents (robots, voice assistants and chatbots) with people aged 60 and over without cognitive impairment. On average, loneliness and depression scores improved after people used one, and no study reported people getting worse. The authors describe the improvements as small to moderate, and the average change in depression scores as modest.

The detail matters as much as the headline:

• The studies were small. Half of them had 18 or fewer people using the companion.
• Few had a fair comparison. Only three were randomised controlled trials. Because comparison groups were rare, the pooled results compare people with themselves before and after, not with similar people who had no companion.
• Follow-up was short. Few studies followed people up for long, so the authors say it is uncertain whether any benefit lasts.

Two other reviews point the same way. A 2025 review in Healthcare, of nine studies of AI-based tools (mostly social robots), found that six reported a significant fall in loneliness; the studies each had between 15 and 64 people. An earlier review of randomised trials of social robots, in The Gerontologist in 2019, reported that robots could reduce loneliness, but said its conclusions were limited by the lack of high-quality studies.

Read together, the reviews show an encouraging but modest signal from small, short studies. That is a reason for more research, not a promise.

Small pilots, and studies by the companies that make the products

Two individual studies show what much of the evidence looks like close up.

A voice assistant in independent living. In a 2021 pilot in the American Midwest, 16 people aged 77 to 96 who lived alone in an independent living facility were given an Amazon Echo and asked to talk to it at least five times a day. After four weeks, their loneliness scores had fallen. There was no comparison group, and the authors call it a pilot. It was funded by a University of Nebraska-Lincoln award, and the authors declared no commercial ties.

A companion robot, studied with its maker. A 2024 paper on ElliQ, a robot designed to start conversations rather than wait to be spoken to, was co-written by the chief executive of Intuition Robotics, the company that makes it, and two of its employees. The paper's most striking figure, that 80% of respondents agreed they felt less lonely when the robot was with them, comes from a customer survey that the robot itself presented after at least 30 days of use. Of 278 users approached, 173 answered. Its authors say randomised controlled trials are needed to show whether ElliQ works compared with other forms of help.

Figures from a company are not worthless, but they are not independent either. When any product claims to reduce loneliness, it is fair to ask who measured it, how, and compared with what.

What nobody knows yet

• Whether any benefit lasts. The studies in the 2026 review ran from three brief sessions to four months.
• How today's AI chatbots compare. With only one study of a large language model in the 2026 review, the evidence on older people says little about the chatbots now on sale.
• What it means for people living with dementia. The 2026 review included only people without cognitive impairment or dementia, so its findings say nothing about them.
• Who it suits. In a US survey of 825 adults aged 25 to 88, published in 2023, 68.7% did not think a talking artificial companion would make them feel less lonely, and the older people were, the less likely they were to think so.

The risks worth taking seriously

Heavier use has gone with more loneliness. In a four-week randomised study of 981 adults in the US using ChatGPT, people who chose to use it more, whatever version they were given, reported more loneliness, less socialising with real people, more emotional dependence on the chatbot and more problematic use. Their average age was about 40. The study was funded by OpenAI, which makes ChatGPT, and four of its authors worked there. It was published as a preprint (a paper shared before peer review). Because people chose how much to use it, it shows a link, not which came first.

A separate survey of 1,131 adults in the US who use Character.AI, a platform where people chat with AI characters, found that using it mainly for company was linked to lower well-being, and more so when use was intensive.

Dependence. A companion is always available and never tires of listening. That is part of the appeal, and it is also why it is worth noticing whether it is adding to someone's week or quietly taking the place of people in it.

Not knowing it is an AI. The concern is sharpest for people living with dementia. In the same 2023 survey, asked to imagine having dementia, 69.3% said they would be uncomfortable if the person supporting them let them believe an artificial companion was a real human. Since 2 August 2026, the EU's AI Act has required AI systems that talk directly with people to be designed so that people are told they are interacting with an AI, unless that is obvious.

Privacy. People tell a good listener about their health, their family and their money. What happens to those words depends entirely on the company. Our guide to the privacy questions to ask first covers what to check.

What a companion can reasonably do

These are things a companion can do, not outcomes anyone has proven:

• Be there between calls and visits. On a quiet afternoon, or an evening when nobody else is around to talk to.
• Hold an ordinary conversation. About the day, the garden, a memory or the people they love, at their own pace, with nothing to learn.
• Keep a gentle routine. A reminder about medication or an appointment, or a morning chat that becomes part of the day.
• Remember what matters to them. Names, stories and what was said last time, so a conversation picks up where it left off.

Some people will enjoy this and some will not, and both are reasonable.

What it cannot do

It cannot replace people. Some of the clearest evidence on easing loneliness comes from ordinary human contact. In a randomised trial published in 2021, 16 callers aged 17 to 23, with brief training in empathetic conversation, phoned 240 adults who received support from Meals on Wheels Central Texas. After four weeks, the people who were called reported less loneliness, depression and anxiety than those who were not. Our guide to loneliness in later life sets out the rest of what helps.

It cannot treat depression. A companion that brightens someone's day is not treating an illness. Depression has effective treatments, including talking therapies and, for some people, antidepressants, and the NHS says it is important to seek help from a GP if you think you may be depressed. If someone seems low for weeks, has lost interest in things they used to enjoy, or talks about being a burden, encourage them to see their GP and offer to help them do it. If you are worried about their immediate safety, call your local emergency number.

Is it helping? Questions to ask after a few weeks

If you or someone you love tries a companion, these questions are worth asking together after a few weeks, and again after a few months.

Is the companion helping?

1. Is it adding to the week, or taking the place of something in it?
2. Are calls, visits and outings with people happening as often as before?
3. Is it enjoyed, or used out of a sense of duty?
4. Is it clear, every time, that it is an AI and not a person?
5. Does everyone know who can see what, and how to stop sharing?
6. Has low mood lasted for weeks? That is one for the GP, not the companion.

Where Vee fits

We make an AI companion, so it is fair to say plainly where it sits in all this.

Vee is a friendly AI companion that someone can talk to every day on a tablet or phone, just by speaking, about their day, their memories and the people they love. It can remind them about medication and appointments, and it remembers what matters to them. The first time Vee meets someone, it introduces itself as an AI companion, a computer and not a person. Whenever anyone asks whether it is a person, it says plainly that it is an AI.

Vee is designed to keep someone company between calls and visits, never to replace people. The family receives a short note on how the person is doing, never their conversations. The person can tell Vee to stop sharing at any time; after that, the family hears from us only if the person needs help. Vee notices change; it diagnoses nothing, and it is not a medical device.

We do not claim that Vee reduces loneliness, because we have not measured that yet, and until we have, we will not claim it. During the pilot, which is free for families, we are testing something narrower. Is Vee good company between calls and visits? And do the short notes help a family notice when someone has gone quiet, so that the people who love them can reach out sooner?

Sources

1. Satake Y, Costello H, Naran N, Ishimaru D, Ikeda M, Howard R. Autonomous conversational agents for loneliness, social isolation, depression, and anxiety in older people without cognitive impairment: systematic review and meta-analysis. Psychological Medicine, 2026.
2. Yang Y, Wang C, Xiang X, An R. AI applications to reduce loneliness among older adults: a systematic review of effectiveness and technologies. Healthcare, 2025.
3. Pu L, Moyle W, Jones C, Todorovic M. The effectiveness of social robots for older adults: a systematic review and meta-analysis of randomized controlled studies. The Gerontologist, 2019.
4. Jones VK, Hanus M, Yan C, Shade MY, Blaskewicz Boron J, Maschieri Bicudo R. Reducing loneliness among aging adults: the roles of personal voice assistants and anthropomorphic interactions. Frontiers in Public Health, 2021. Funded by a University of Nebraska-Lincoln Layman Award.
5. Broadbent E, Loveys K, Ilan G, Chen G, Chilukuri MM, Boardman SG, Doraiswamy PM, Skuler D. ElliQ, an AI-driven social robot to alleviate loneliness: progress and lessons learned. The Journal of Aging Research and Lifestyle, 2024. Three authors, including its chief executive, work for Intuition Robotics, which makes ElliQ.
6. Fang CM, Liu AR, Danry V, et al. How AI and human behaviors shape psychosocial effects of extended chatbot use: a longitudinal randomized controlled study. arXiv preprint, 2025. Funded by OpenAI.
7. Zhang Y, Zhao D, Hancock JT, Kraut R, Yang D. The rise of AI companions: interaction with AI companions and psychological well-being. arXiv preprint, 2025, revised 2026.
8. Berridge C, Zhou Y, Robillard JM, Kaye J. Companion robots to mitigate loneliness among older adults: perceptions of benefit and possible deception. Frontiers in Psychology, 2023.
9. EU Artificial Intelligence Act, Article 50: transparency obligations.
10. Kahlon MK, Aksan N, Aubrey R, et al. Effect of layperson-delivered, empathy-focused program of telephone calls on loneliness, depression, and anxiety among adults during the COVID-19 pandemic. JAMA Psychiatry, 2021.
11. NHS, Depression in adults.
12. Virtubond, Privacy policy.

Sources checked 2 November 2026.