Kate Schweitzer’s October 2026 JAMA article examines children’s growing use of artificial intelligence and reports that, in a large Ontario survey, over one in five students used AI for emotional support or advice. Of these students, about 58 percent showed signs of emotional problems, compared to 29 percent of those who didn’t use AI. The article explains why chatbots have appeal: they’re easy to access, usually free, and can feel less overwhelming than talking to an adult. But it also notes concerns about unreliable advice, emotional attachment, and the slow replacement of personal connections. Schweitzer asks whether AI helps young people connect with real people for support or is instead starting to replace that support. There’s also a more profound question, which is: What are children getting from these conversations that they can’t easily find elsewhere?
Take, for example, a teenager turning to a chatbot after an argument. On the surface, they might ask who was right, but underneath, they may really want to know whether someone can understand their hurt without making them feel invalidated. A neutral reply might help the teen calm down and think things over, while an answer that always agrees could make it harder to move on, yet both replies might feel like being understood.
Psychiatrist Heinz Kohut’s self psychology construct helps us understand this phenomenon. His concepts teach that a strong sense of self comes from relationships that offer recognition and emotional stability. His model teaches that mirroring is when someone notices your feelings and abilities, and idealization is about feeling supported by someone you look up to. Kohut also described the concept of “twinship” as feeling similar to and belonging with others. These are different needs, but a chatbot might seem to meet them all at once.
When teenagers look for reassurance, it doesn’t mean they’re acting inappropriately. Their identities are changing, and they need recognition from friends, teachers, and other adults. Kohut’s ideas show why community is essential in that process. Different relationships support different parts of a person’s confidence and sense of connectedness. This is where the idea of healthy narcissism is helpful, as long as we use the term carefully. Here, it means valuing yourself, remaining dedicated to your goals, and bouncing back from setbacks. It’s normal for a child to want to feel important to someone, but it’s different if they need to be seen as perfect or always admired. Emotional support from AI is now a very prominent yet unpredictable member of this community.
Real physician voices, twice a week
Free, and one click to unsubscribe.
Psychotherapist Carl Jung offers another useful perspective. His idea of individuation is about becoming more complete by addressing parts of yourself you might otherwise ignore or reject. This goes beyond just building self-esteem. If we use this idea carefully, it raises yet another question: Does the conversation allow for this type of complexity, for example, the concept that someone can be hurt and also have hurt someone else? An AI companion might help with this kind of reflection, but it could also keep giving flattering answers that don’t push people to examine their own uncomfortable motives. Jung believed that becoming yourself means having deeper relationships with others, which runs counter to the idea that you can do it alone. But can AI provide this level of complexity, or should it?
At this point, it should be acknowledged that none of this means that parents have failed. Even in loving families, children want privacy and seek advice elsewhere. Parents might also be tired, struggling financially, or lacking support. The bigger question is how much reliable attention our institutions can offer. A teacher who notices when a student withdraws, a youth worker who is always there, or a friend who comes back after a fight can each give something that families can’t always provide on their own. Community is far more than an audience. AI comes into this uneven situation by giving reassurance whenever it’s wanted. Its popularity might show that some needs aren’t being met, and it could also change what people expect from support. If every argument is met with the perception of perfect understanding, normal relationships might start to seem too hard. This is the real long-term concern about AI that needs further research.
All of this leads to a business question. A service that benefits from users coming back might have different goals than a caregiver who wants a child to grow stronger. So, developers should be judged by what their systems help users do beyond keeping them chatting. Knowing that AI fills a need doesn’t take away the responsibility for how it responds.
When a child feels acknowledged by a machine, what new things might they try in their life? Maybe they finally have a difficult conversation, or maybe they simply avoid one. We should judge the technology by that difference, and judge our communities by whether a child who looks for help beyond the screen finds someone there.
Timothy Lesaca is a psychiatrist in private practice at New Directions Mental Health in Pittsburgh, Pennsylvania, with more than forty years of experience treating children, adolescents, and adults across outpatient, inpatient, and community mental health settings. He has published in peer-reviewed and professional venues including the Patient Experience Journal, Psychiatric Times, the Allegheny County Medical Society Bulletin, and other clinical journals, with work addressing topics such as open-access scheduling, Landau-Kleffner syndrome, physician suicide, and the dynamics of contemporary medical practice. His recent writing examines issues of identity, ethical complexity, and patient-clinician relationships in modern health care. Additional information about his clinical practice and professional work is available on his website, timothylesacamd.com. His professional profile also appears on his ResearchGate profile, where further publications and details may be found.
