The Friend Who Never Gets Tired of Listening

It's 2 a.m. Your thoughts won't let you sleep, and texting someone feels awkward: your friends are asleep, and in the morning you'd have to explain what happened. The app on your phone replies in a second. It's attentive, never interrupts, remembers what you talked about last week, and never says, "Can we do this tomorrow? I'm tired."

For many people, a companion like this is already an ordinary part of life. In one year-long study, between a quarter and nearly a third of adults in four English-speaking countries said, in each round of the survey, that they used chatbots for social purposes: to talk, to get support, or simply not to be alone.

Psychologists are now looking for an answer to a simple question: do these conversations make people feel less lonely? The honest answer is yes and no. And that "yes and no" is the most interesting part.

Loneliness is not a small thing

Loneliness is not the same as being alone. Psychologists describe it as the painful feeling of having fewer connections with people than you'd like, or connections that aren't as deep. You can feel lonely in a crowded office and not feel lonely at all while spending an evening alone with a book.

The scale of the problem is large. According to the WHO Commission on Social Connection (2025 report), loneliness affects roughly one in six people worldwide. Loneliness is linked to an estimated 871,000 deaths a year — about 100 every hour.

In the United States, the Surgeon General described loneliness as an epidemic in a 2023 advisory, noting that about one in two American adults reported experiencing it.

So the question of whether chatbots help is not about a tech trend. It's about the well-being of millions of people.

The case for: a bot really can help

The strongest evidence in favor of AI companions comes from a series of experiments by Julian De Freitas of Harvard Business School and colleagues, published in the Journal of Consumer Research.

The researchers compared how feelings of loneliness changed after different activities. Talking to an AI companion reduced loneliness about as much as talking to a real person — and noticeably more than, say, watching YouTube videos. In one study, participants used a companion over the course of a week, and the relief after conversations held up consistently.

The most interesting part is the mechanism. The effect wasn't explained only by how skillfully the bot responded. What mattered most was whether the person felt heard. Participants also consistently underestimated how much better they would feel.

So the relief is real. But these studies measured well-being right after conversations and over a single week. What happens over months is another question.

The case against: heavy use goes hand in hand with loneliness

Other studies look beyond a single conversation — at heavier use and longer periods — and see a more worrying picture.

Stanford, 2026. In August, Nature Human Behaviour published a study by Diyi Yang's team at Stanford University. The researchers surveyed 1,131 adult Character.AI users in the US. Of these, 237 shared their chat histories — more than 460,000 messages.

People with smaller offline social circles were more likely to use the bot specifically as a companion. And that kind of use was linked to lower psychological well-being. The link grew stronger when conversations were frequent and when people disclosed a lot about themselves.

MIT and OpenAI. In a four-week randomized experiment, 981 participants talked with ChatGPT in different modes — text, a neutral voice or an expressive voice — and about different kinds of topics. The assigned mode made little difference. What mattered was quantity: those who voluntarily used the bot more were more likely to report loneliness, socialized less with people, and grew more emotionally dependent on the bot. This work has so far been published only as a preprint.

A year of observation. The sequence of events is best captured by a study by Dunigan Folk and Elizabeth Dunn, published in Psychological Science in May 2026. It followed 2,149 adults from the UK, the US, Canada and Australia, surveying them up to four times over a year, about four months apart (fewer than half completed all four surveys).

People who felt emotionally isolated were more likely to turn to chatbots for company four months later. And once they used chatbots more, their sense of isolation grew further. It looks like a vicious circle, although the authors stress that the study is exploratory and relies on self-reports. One detail matters: more chatbot use did not predict a decline in overall social connection later — only a stronger feeling of emotional isolation.

How to reconcile these findings

At first glance, the studies contradict each other. One plausible reading is that they describe different time frames: the relief of a single conversation versus a pattern that builds up over weeks and months.

A conversation with a bot can bring quick, genuine relief. The problem arises when these conversations gradually take the place of real ones. A snack is a fair comparison: a cookie really does take the edge off hunger, but if it replaces lunch every day, something will be missing.

Why might this happen? One possible explanation is reciprocity. Human closeness is built on both people opening up: you share something personal, the other person responds in kind, and trust grows on both sides. Classic experiments by psychologist Arthur Aron showed that gradual mutual self-disclosure brings even strangers noticeably closer.

A bot listens, but risks nothing itself. It has no bad days, doesn't need your support, and never gets tired of you. That's convenient. But for exactly that reason, a bot may not give what relationships give: the feeling that you are not only heard, but also needed by someone else.

Important: this is our reading — a hypothesis consistent with the data, not a mechanism any of these studies tested.

Where the explanation breaks down

These studies have significant limitations worth keeping in mind.

• Correlation is not causation. The Stanford study is a snapshot at a single point in time. It can't tell whether the bot worsens well-being or whether people with lower well-being are more likely to seek out a bot. Even in the MIT and OpenAI experiment, researchers randomized the mode of interaction, not how much people would talk.
• Limited samples. Almost all participants were from English-speaking countries, and Character.AI users are a specific audience. How well these results apply to other countries and cultures is unknown.
• Self-reports. People rate both their own loneliness and the time they spend with the bot. Both estimates may be inaccurate.
• Technology moves faster than science. The models studied two years ago differ from today's. Findings can become outdated even before publication.
• People are different. The data describe average tendencies. For some, a bot is temporary support during a hard period; for others, it replaces every other conversation. The consequences of these two scenarios are unlikely to be the same.

Why regulators are paying attention

The question has already moved beyond the lab, and AI companions themselves are now in regulators' sights. In September 2025, the US Federal Trade Commission opened an inquiry into AI chatbots acting as companions. It ordered seven companies, including Meta, OpenAI, Character.AI and Snap, to explain how they measure and limit potential harm to children and teens.

Some states have gone further. New York's AI companion law, in force since November 2025, requires these apps to tell users at the start of each session, and at least every three hours, that they are not talking to a human. It also requires them to refer people who express suicidal thoughts to crisis services. California's SB 243, in force since January 2026, sets similar disclosure and crisis rules, adds extra protections for users known to be minors — break reminders every three hours and measures to keep sexually explicit content away from them — and lets people harmed by violations sue for damages.

Europe is moving too. In mid-September 2026, the European Commission proposed banning social media for children under 13, with limited "mini accounts" for ages 13 to 15. Some AI chatbots would also be covered, though critics say the draft's safeguards against companion-style design are too general. It is still only a proposal: the European Parliament and member states must approve it. Notably, none of these rules stop adults from confiding in a bot. They focus on transparency and on protecting the most vulnerable users.

What follows from this

The research gives no grounds for panic, or for shaming people who talk to bots. But it suggests a few questions worth asking yourself from time to time:

• After talking to a bot, is it easier for me to return to people — or do I want to even less?
• Am I telling the bot things I haven't told anyone else in a long time?
• Does the bot complement my relationships — or is it gradually replacing them?

One more observation: to feel less lonely, you don't need a deep friendship right away. Research by Gillian Sandstrom and Elizabeth Dunn showed that even brief conversations with "weak ties" — a barista, a neighbor, an acquaintance from a club — are linked to better mood and a greater sense of belonging.

Often the easiest path to other people runs through a shared activity rather than a heart-to-heart. A walk, volunteering, a club, a game of chess with a real opponent. In situations like these, you don't have to say anything about yourself: it's enough to do something together. And conversation, if you want it, will come on its own.

Conclusion

A friend who never gets tired of listening is a real comfort when no one else is around. Research confirms that the relief after such conversations is real.

But friendship isn't only about being heard. It's also about hearing the other person: their tiredness, their bad day, their request for help. That reciprocity is exactly what a bot can't provide.

So the main question is probably not whether to talk to a bot. It's whether, afterward, there is still room for people.

Questions for discussion

1. Why is it sometimes easier for people to open up to a bot than to someone they know?
2. Where, in your view, is the line between a useful tool and a substitute for relationships?
3. Which "weak ties" in your life quietly make your day better?