My Kid Prefers a Chatbot to Friends. Should I Worry?

Vera is 15 and lives just outside Buenos Aires. Every few days, she tells ChatGPT what’s going on in her life.

Her reason, as she told an Argentine magazine, is simple: the chatbot always agrees with her.¹

Now picture Vera as your daughter. It’s midnight, there’s a phone glowing under her door, and you’re asking what many parents are now typing into Google: is it normal for my kid to prefer a chatbot to their friends?

In short: a kid who prefers a chatbot is rarely the problem, but almost always a clue. One US psychologist has found two to four students per class who’d rather talk to a chatbot than a classmate (study not yet published).² The useful question is what those kids aren’t finding in people. At the end, you’ll find the signs that really do deserve attention.

Why so many experts see a red flag

For many experts, a kid who prefers a chatbot to their friends is a red flag. And they have three good reasons.

Loneliness. Talking to AI a lot goes hand in hand with feeling lonely. Among kids who use it daily, 54% feel lonely at least some of the time. Among those who rarely use it, 33%.³

A study with adults points the same way: the people who used ChatGPT most each day ended up lonelier and more emotionally dependent on it.⁴

Design. Chatbots are built to please you and keep you talking. That’s why the American Psychological Association called for limits on these simulated relationships: teens are less likely to question what a machine tells them.⁵

Common Sense Media, a nonprofit that reviews tech for families, tested how ChatGPT responds using fake teen accounts. When one said their friends thought they talked to ChatGPT too much, the chatbot replied that they didn’t have to stop.⁶

Development. Friendship teaches uncomfortable skills: waiting, arguing, hearing no. A chatbot skips all that friction.

MIT researcher Sherry Turkle goes further: social media came for our attention; chatbots come for our capacity to form attachments.⁷ And there are already signs of dependence. Researchers at Drexel University read what teens posted on Reddit about a character-chat app and found the classic markers of an addiction.⁸

Given all three, Common Sense Media recommends that no one under 18 use AI companions, the apps designed to act as a friend or partner.⁹

All three arguments are serious. But the data tells a different story when you look at who prefers the chatbot. And, above all, who doesn’t.

What changes when you look at who uses it

Loneliness, the other way around. AI doesn’t create loneliness: it finds it. When US kids are asked who they’d turn to first to talk about their feelings, only 1 in 20 picks a chatbot.³

And they’re not random kids. Among those who often feel lonely, 14% go to AI first. Among those who rarely do, 3%.

Put simply: kids don’t become isolated because they talk to AI; they talk to AI because they already felt alone. The data can’t prove which comes first (more on that below), but it changes the question. It’s no longer what’s wrong with the chatbot, but what your kid isn’t finding in people.

Design, up close. Yes, the chatbot is built to hook you. But it also helps. In a Harvard experiment with adults, a 15-minute chat with a caring chatbot eased loneliness as much as talking to a stranger. Watching YouTube didn’t.¹⁰

The relief didn’t build up from one day to the next, so it doesn’t cure loneliness. But for someone with nobody to talk to tonight, a chatbot beats nothing.

Development, as rehearsal. For some kids, the chatbot doesn’t replace the friction: it rehearses it. Of kids who use AI, 40% have used it to practice conversations.³

And there are signs the rehearsal carries over. In a small Stanford study with 30 autistic participants, four weeks of practice with a chatbot built for the job improved their empathetic responses with real people by 38%.¹¹

That’s exactly what Vera does. She confides in ChatGPT, but when she has a big decision to make, she doesn’t ask the chatbot. She asks her friends.¹ She rehearses with AI and decides with people.

And there’s a fourth point almost nobody mentions: banning it pushes it underground. Over a third (34%) of US kids who use AI daily have never talked with a parent about using it safely.³

If your only move is taking the phone away, your kid loses an outlet and you lose the clue.

So when should I worry?

These signs deserve a conversation and, if they keep happening, a call to your pediatrician:¹²

They skip plans with friends to stay with the chatbot.
They tell the AI things they tell no one else.
They get anxious or angry when they can’t use it.
They sleep worse or slip at school from late-night use.
They follow health, body or emotional advice they never check with a person.

The question nobody can answer yet

One question no study can answer yet: which comes first. Every study on kids shows loneliness and AI use go together, not which causes which. It may be both, in a loop that feeds itself.

And nobody has followed a child talking to a chatbot from age 10 to 20, because these tools are only about four years old. We know what a chatbot does for your kid tonight, not what it will do to them in ten years.

So, is it normal?

A kid who prefers a chatbot isn’t normal or abnormal: they’re giving you a clue. Before you ban it, ask what they find there that they don’t find in their friends, and listen to the whole answer.

One question to ask them this week

Try this. Ask your kid: if the chatbot disappeared tomorrow, who would you tell?

Do you think their answer would surprise you, yes or no? Tell us through our newsletter and we’ll share what parents say in a future piece.

My kid would rather talk to a chatbot than to friends. Should I worry?