Offline Struggles, Online Companions:

What the Data Says About Teens and Chatbots

I've been talking to more teens about AI chatbots lately and whew, what I'm hearing is all over the place. From the teen boy who told me he regularly talks to chatbots when anxiety keeps him up at night...to the teen girl who shared how a chatbot kept trying to interpret her depressive episodes through a lens of the supernatural...to the teen who used chatbots to respond to arguments with her family and the teen who used it to write her mom's birthday card.

I've talked to teens who have expressed shame, fear, and guilt for how they have interacted with chatbots.

These are some heavy emotions for teens to process. I've heard everything from the apathetic to the benign to the truly concerning.

Research is telling us that teens who are struggling, vulnerable, or in the middle of a mental health challenge, are the ones most likely to turn to a chatbot for company or emotional support. Pew Research Center found that 64% of U.S. teens use AI chatbots overall, but only 12% have used one specifically for emotional support. Common Sense Media found that number climbs when we look at teens who report they’re having a harder time making friends.  Similarly, a recent California survey found that teens who felt most lonely used AI chatbots at 48%, compared to 37% overall, climbing to 57% among teens who reported that they’d stopped hanging out somewhere because it didn't feel safe or welcoming. In other words, teens’ social lives shape their engagement with bots.

We know youth are turning to chatbots, but we’re still learning if, when, and under what conditions they can be helpful, however, we have reasons to be concerned. When Common Sense Media and Stanford Medicine's Brainstorm Lab stress-tested ChatGPT, Claude, Gemini, and Meta AI for mental health support, they found that after extensive conversations, all four missed clear signs of anxiety, depression, ADHD, eating disorders, mania, and psychosis, conditions that affect roughly one in five young people. Harvard's Center for Digital Thriving found that most teens who talk to AI about something heavy, haven’t talked to an adult about it. As one teen put it, they don’t want a lecture from their parents, so a chatbot just feels easier.

It's completely understandable why talking to chatbots feels easier…because it is. Chatbots don’t judge, they’re never shocked. There's no risk of damaging a relationship or a reputation when talking to a chatbot.

But, so much of identity development in adolescence happens through relationships with other people in ways that aren't always comfortable - disagreements, being told no, repairing something after a conflict, figuring out who you are by bouncing ideas off someone who isn't just reflecting your ideas back. A chatbot is designed to agree with you, affirm your beliefs, and never tire of the conversation.

But there are also missed opportunities to learn how to navigate friction in relationships, how to take accountability for your actions, and how to be present during an awkward conversation.

If teens are afraid to be vulnerable with each other, they miss opportunities to build trust, which will paradoxically leave them feeling lonelier.

Belonging doesn't just mean being liked or surrounded by like-minded people, it's feeling valued and accepted for who you are and knowing you’d be missed if you weren’t there.

Despite all the narratives telling us teens are addicted to their phones, what teens actually tell us is that they want more time with people, in person. This means that our radar can’t just focus on phone use, rather we need to focus on what’s driving loneliness in the first place. The California Partners Project surveyed more than 1,000 teens and found that 88% say spending time with friends in person matters to them, more than half wish they connected in person more than they do, and almost half of teens who use AI chatbots said they'd use them less if they had more chances to connect face to face. Teens are choosing chatbots precisely because they feel like they don’t have people to turn to.

Teens need greater access to places to be together, but, as the California survey reveals, providing a place isn't the same as providing belonging.

Which brings me to something I've been thinking about a lot lately: how the collective choices of others can limit teens' agency inside spaces meant to include them.

Recently a parent shared with me how upset her kids were that everyone in their after-school program was on their phone during an activity. Not only did her kids not have a phone so they felt left out, but the point of the club was to foster connections and belonging, and the phones were disrupting that.

As program leaders we need to be intentional in designing spaces and opportunities, not with hard and fast rules about phones, but by co-creating values with participants: what do they want out of the program, why are they attending, and what community agreements can help achieve that? Teens need agency and buy-in, not just rules passed down.

Most of what I'm hearing, and most of what the research is finding, points to AI use as a symptom of something else: loneliness, an unmet need, a gap in who's available or who feels safe to talk to.

Chatbots can widen that gap if the easiest option becomes the only option a teen reaches for. So the real challenge involves access to adults who respond with curiosity instead of alarm, community spaces intentionally designed to foster belonging, opportunities to learn to be mutually vulnerable with peers, and teens equipped with enough literacy to see a chatbot clearly for what it is and isn't.