More young people are turning to AI for emotional advice, and many find it helpful. But these increasingly private conversations raise questions about how technology is changing the way we deal with personal problems.
Almost two-thirds of young Americans who have asked an AI chatbot for mental health advice have never told anyone about it. According to a study published in JAMA Pediatrics in June 2026, nearly all of those who used chatbots for this purpose considered the advice helpful.
Think about the last time you were worried about a relationship, replaying an awkward conversation or feeling anxious late at night. You wanted to talk about it, but perhaps you didn’t want to bother a friend. Or you weren’t quite sure how to explain what was going on. Now there’s another option. You can open a chatbot, describe your problem and receive a response within seconds. You don’t need to arrange a meeting, worry about someone’s reaction or feel embarrassed about asking the same question again. It’s easy to see the attraction. What’s less clear is how these conversations affect the way people handle their problems outside the chat window.
Young people are already turning to AI for advice
Researchers behind the JAMA Pediatrics study surveyed 1,009 Americans aged 12 to 21 in November 2025. About 19% said they had used AI chatbots for mental health advice. Among those users, nearly 43% did so at least once a month, while 63% had not told anyone about using AI for this purpose. The most striking figure may be the level of satisfaction. Almost 92% described the advice as somewhat or very helpful. That tells us something about how people experience these conversations. It doesn’t tell us whether the advice was accurate or whether their mental health improved afterward.
The trend isn’t limited to teenagers. A separate Pew Research Center survey published in June 2026 found that 10% of American adults had used chatbots for emotional support or advice. Among adults under 30, the figure was 20%. The two surveys covered different age groups and asked different questions, so their results aren’t directly comparable. Still, they suggest that discussing personal problems with AI is becoming familiar to a growing number of people.
Sometimes it’s easier to talk when nobody is listening
Talking about personal problems can be uncomfortable, even with people we trust. Friends have opinions. Partners may become defensive. Family members sometimes offer solutions when all we want is to explain what happened. A chatbot doesn’t bring those complications into the conversation in quite the same way. You can take your time, change your mind halfway through a sentence or ask a question that feels too embarrassing to say aloud. You can also return to the same subject repeatedly. There is no need to wonder whether the other person is tired of hearing about it.
That freedom may be particularly appealing to someone who struggles to open up. It may also explain why private conversations with AI feel different from talking to a friend or therapist. The research doesn’t establish why people choose chatbots or why so many keep their use private. Some may simply appreciate the convenience. Others may be looking for a place to explore thoughts they aren’t ready to share.
Whatever the reason, AI offers something unusual: an opportunity to discuss deeply personal experiences without having to navigate another person’s immediate emotional response. That can be useful. It can also make certain conversations easier to postpone.
A reassuring answer isn’t necessarily good advice
Chatbots are often remarkably good at producing supportive responses. They can help someone put their concerns into words, suggest another way of looking at a situation or offer ideas for approaching a difficult conversation. But a convincing response isn’t necessarily a reliable one.
In June 2025, Stanford University researchers reported problems in tests involving five therapy-oriented chatbots. Some responses showed bias toward particular mental health conditions, while others handled sensitive situations poorly. The findings don’t mean that every conversation about emotions with AI is unsafe. They do show that a chatbot can sound understanding while offering advice that may be inappropriate for the situation.
The American Psychological Association has also warned against using general-purpose chatbots and wellness applications as replacements for qualified mental health care. Its concerns include inaccurate information, privacy and the possibility of users becoming overly dependent on digital companions.
Americans appear uncertain about the benefits themselves. In an August 2026 Pew Research Center survey, 39% of adults said chatbots do more harm than good for people experiencing loneliness, while 19% believed they do more good than harm. Those figures reflect public opinion, not measured outcomes. They do, however, show that the growing popularity of emotional conversations with AI is accompanied by considerable skepticism.
The conversation that never happens
Consider someone who has been upset because a close friend has stopped responding to messages. They might ask a chatbot what to make of the situation. Perhaps it suggests several possible explanations, helps them recognize why they’re hurt and offers ideas for approaching their friend without sounding accusatory. Afterward, the person feels better prepared to have a conversation. In this case, AI has served a practical purpose.
Now imagine that the same person returns to the chatbot every evening. They ask whether the friendship is over, whether they’ve done something wrong or whether they should send another message. Each answer provides a little reassurance, but nothing changes between them and their friend. Both people might describe their chatbot conversations as helpful. Yet the results could be quite different.
This example illustrates a possibility rather than a finding from the studies. Researchers have not established that using chatbots for emotional advice generally makes people avoid difficult situations. Still, it raises a question worth considering. If a conversation helps us understand a problem but never brings us any closer to dealing with it, how much has it actually helped?
People have always found ways to put off uncomfortable conversations. We might seek repeated reassurance from friends, replay events in our minds or spend hours trying to decide what someone else meant. AI hasn’t invented those habits. It has simply made another form of reassurance available at any hour, without requiring anyone else’s time or attention.
What AI can offer, and what it cannot
There are sensible ways to use chatbots when dealing with personal concerns. Someone might want help organizing their thoughts before seeing a therapist, finding the right words for an apology or considering a disagreement from another perspective. For people interested in mindfulness and personal development, AI may also provide an additional way to reflect on emotions and recognize recurring patterns.
But knowing what we feel isn’t always the same as knowing how to handle it. Sometimes the next step involves accepting uncertainty, having an uncomfortable conversation or listening to a perspective we would rather not hear. A chatbot can help someone prepare for those experiences. It cannot have them on that person’s behalf.
There are also limits that have little to do with everyday relationship problems. People facing serious mental health difficulties, particularly a crisis, need appropriate human and professional support. An AI-generated response, however reassuring it sounds, is not a substitute for qualified care.
What happens after we close the chat?
The recent research suggests that many people are comfortable sharing personal concerns with AI, including things they haven’t discussed with anyone else. Most young users surveyed in the JAMA study also reported finding the advice helpful. We know much less about what happens afterward.
Do these conversations help people communicate more openly? Do they encourage people to seek support when they need it? Or might some people become increasingly comfortable discussing their lives with a chatbot while avoiding difficult interactions with others? The answers will probably depend on the person, the circumstances and how the technology is used. For now, the available surveys don’t tell us enough to draw broad conclusions.
The more revealing question may be what happens beyond the screen. Someone can spend an hour discussing a relationship problem with a chatbot and come away feeling understood. Whether that conversation makes it easier to speak with the other person—or easier to keep putting it off—is something a satisfaction survey alone cannot tell us.
Sources and Further Reading
- JAMA Pediatrics (June 2026). Study on AI chatbot use for mental health advice among Americans aged 12–21. https://jamanetwork.com/journals/jamapediatrics/fullarticle/2849307
- Pew Research Center (June 17, 2026). Americans and AI: Chatbots, Smart Devices and Views on Impact. https://www.pewresearch.org/internet/2026/06/17/americans-and-ai-2026-chatbots-smart-devices-and-views-on-impact/
- Stanford University (June 2025). Research on risks and limitations of AI mental health care tools. https://news.stanford.edu/stories/2025/06/ai-mental-health-care-tools-dangers-risks
- American Psychological Association. Health advisory on AI chatbots and wellness applications. https://www.apa.org/topics/artificial-intelligence-machine-learning/health-advisory-chatbots-wellness-apps
- Pew Research Center (August 25, 2026). Do Americans Think Chatbots Help or Hurt People Using Them for Loneliness, Depression or Stress? https://www.pewresearch.org/science/2026/08/25/do-americans-think-chatbots-help-or-hurt-people-using-them-for-loneliness-depression-or-stress/
