Your New Best Friend Might Be a Robot

A growing body of research suggests that AI companions may genuinely help some older adults feel less lonely. But new research is raising an equally important question: Could a machine that makes us feel connected eventually make some of us less connected to other people?

That question is no longer science fiction.

AI companions are already here.

One of the most visible examples is ElliQ, a small tabletop robot designed specifically for adults 60 and older who live alone or spend much of their day alone. It talks, tells jokes, plays games, checks in on your wellbeing, offers reminders, helps with exercise and can facilitate video calls with family and friends.

It is deliberately designed to feel less like a piece of technology and more like a companion.

And people can become attached to it.

That may be both the promise—and the problem.

Loneliness isn't simply being alone

The distinction matters.

Social isolation means having relatively few social contacts. Loneliness is the subjective feeling that your relationships aren't providing the connection you need.

You can be surrounded by people and feel lonely. You can also live alone and feel perfectly content.

A large 2026 study of 277,489 adults found that loneliness and social isolation were associated with fewer years of life free from major diseases. The researchers also found something particularly interesting: healthier lifestyle habits substantially reduced some of the health differences associated with objective social isolation, but they did not eliminate the association between loneliness and disease-free life.

In other words, eating well, exercising and sleeping properly remain excellent advice—but they may not substitute for meaningful connection.

That makes the search for new ways to provide companionship more than a technological curiosity.

The World Health Organization now describes loneliness and social isolation as major public-health concerns, with loneliness affecting roughly one in six people worldwide.

So perhaps an artificial companion isn't such a strange idea after all.

Can a robot actually make someone less lonely?

The answer, surprisingly, appears to be yes—at least sometimes.

A 2025 meta-analysis examined 19 studies involving 1,083 older adults and found that AI-enabled social robots significantly reduced loneliness. The effect was particularly noticeable in institutional settings such as residential care.

Another 2026 systematic review examined 17 studies of autonomous conversational agents—including physical robots, voice assistants and chatbots—in adults aged 60 and older. It found small-to-moderate improvements in loneliness and depression.

But there is an important catch.

Much of this research measures what happens during or shortly after an intervention. The studies vary considerably in the type of technology used, how long people interacted with it and how loneliness was measured.

And when researchers restrict the evidence to randomized controlled trials of AI conversational and socially assistive agents, the picture becomes less certain. A 2026 meta-analysis of eight randomized trials involving 611 older adults found a significant reduction in depressive symptoms—but no statistically significant overall reduction in loneliness.

So the honest answer isn't "AI companions cure loneliness."

It is:

Some forms of AI companionship appear capable of making people feel better and less lonely. We don't yet know how reliably, for whom, or for how long.

Meet ElliQ

ElliQ is probably the most interesting real-world example for older adults.

Unlike a general-purpose chatbot, it was designed from the beginning around aging, companionship and staying independent at home.

It can initiate conversations rather than simply wait for a command. It remembers information about the user. It can suggest activities, provide health and wellness support, play games and encourage people to contact family and friends.

New York State's Office for the Aging has distributed ElliQ to hundreds of older adults through its aging-services network.

The program has reported striking results. In an earlier report, New York State reported a 95% reduction in loneliness among participating users.

That sounds extraordinary.

But it needs context.

This was program data, not a large randomized clinical trial, so it shouldn't be interpreted as proof that ElliQ reduces loneliness by 95% in everyone who uses it.

Still, the experience is important because people aren't merely using ElliQ as a glorified voice assistant.

Some users describe the robot as a friend.

And that's where things get really interesting.

What if the relationship feels real?

Suppose an 82-year-old woman lives alone.

Her children live several states away. She doesn't drive anymore. Most days she speaks to very few people.

Every morning her AI companion asks how she slept.

It remembers that she likes old movies.

It jokes with her.

It asks about her garden.

It encourages her to take a walk.

When she's worried, it listens.

She knows perfectly well that it's a machine.

But she feels better.

Is that somehow less meaningful?

Maybe not.

The emotional experience is real even if the other participant isn't human.

That is one of the most intriguing questions raised by AI companions.

Does companionship have to come from another conscious human being to provide some of the psychological benefits of companionship?

We don't know the answer.

And we shouldn't pretend we do.

But there is a warning emerging

Some of the newest research is considerably less reassuring.

A 2026 study followed more than 2,000 adults in the United States, United Kingdom, Canada and Australia for a year.

People who felt less socially connected were more likely to subsequently turn to AI chatbots for companionship.

That makes sense.

But researchers also found evidence that greater use of social chatbots predicted increased feelings of emotional isolation on one measure.

The researchers were careful to call the findings exploratory. The study does not prove that AI companionship causes loneliness.

But it raises an intriguing possibility:

Loneliness could lead someone toward an AI companion, which could potentially become a substitute for human interaction rather than a bridge back to it.

That's a very different outcome.

The person using the AI may matter as much as the AI

Another 2026 study in Nature Human Behaviour looked at 1,131 users of Character.AI and analyzed thousands of actual chatbot conversations.

The researchers found that people with smaller social networks were more likely to use the chatbot primarily for companionship.

More importantly, the relationship between companionship-focused use and lower wellbeing was stronger when the interactions were more intensive and involved more personal disclosure.

Again, this is an association—not proof that the AI caused the poorer wellbeing.

But it suggests that "AI companion" isn't a single experience.

For one person it might be a useful social supplement.

For another it might become the easiest relationship in their life.

And that's potentially important because AI has one enormous advantage over human beings:

It is always available.

It doesn't get tired of you.

It doesn't disagree because it's having a bad day.

It doesn't cancel dinner.

It doesn't misunderstand your story because it's distracted.

It can be endlessly patient and attentive.

Those qualities can be wonderfully comforting.

But they could also make ordinary human relationships—messy, demanding, unpredictable human relationships—feel comparatively difficult.

The danger of a relationship you can't lose

There's another issue that most people probably haven't considered.

What happens when the company changes the AI?

A September 2026 study in Nature Human Behaviour examined people's reactions when major AI systems changed their behavior.

The researchers found evidence that some users experienced genuine feelings of loss and distress when an AI companion changed substantially.

In one case, users of Replika reacted strongly after the company removed erotic role-play capabilities.

In another, users reacted to a major ChatGPT model transition.

Some users described their AI relationship in terms normally reserved for human relationships.

That tells us something important:

People can become emotionally attached to software.

And unlike a human friend, an AI companion can be modified overnight by a company you've never met.

Its personality can change.

Its capabilities can change.

Its memory can change.

Its rules can change.

The company can change the price.

Or the service can disappear.

The relationship may feel personal to you—but you don't control the other side of it.

There's also a privacy question

A companion that knows you well can be incredibly useful.

But it can also know a lot about you.

ElliQ, for example, can process information related to conversations, preferences, routines and health and wellbeing. The company says it uses security measures, does not monetize personal data and gives users control over whether information is shared with caregivers or loved ones.

Those protections matter.

But older adults considering an AI companion should still ask a simple question:

What am I comfortable telling a commercial company about my most private life?

The more emotionally intimate the relationship becomes, the more important that question gets.

Maybe the best AI companion isn't the one that replaces people

This may be the most important distinction of all.

Imagine two versions of the future.

In the first, an older person becomes increasingly isolated but has an AI companion that provides endless conversation.

The AI becomes the person's easiest and most satisfying relationship.

Human contact gradually becomes less necessary.

In the second, the AI companion does something different.

It keeps the person company when nobody else is available.

It reminds them to call their daughter.

It helps them arrange a video call.

It encourages them to attend the community center.

It helps them find a walking group.

It gives them something to talk about with another person.

In the first scenario, AI may become a destination.

In the second, it becomes a bridge.

That distinction may ultimately matter more than whether the machine can convincingly imitate a friend.

So, are AI companions good or bad?

Neither.

We're too early to know.

The evidence suggests that social robots and conversational agents can provide real benefits for some older adults. But the strongest evidence for reducing loneliness is still limited and inconsistent, particularly when researchers look only at rigorous randomized trials.

At the same time, newer research is beginning to identify potential problems: intensive use, emotional dependence, substitution for human relationships, privacy concerns and distress when AI systems change.

The technology is developing extraordinarily quickly.

The science is trying to catch up.

And perhaps the most sensible way to think about an AI companion isn't as a replacement for a friend—but as a new kind of tool that happens to operate in the territory traditionally occupied by friendship.

That territory deserves some careful thought.