A recent University of Ottawa study suggests that young people using AI for emotional support may also be signalling distress. That should make us curious, not simply alarmed. The interesting question is not whether AI is “good” or “bad”, but what needs it is meeting that our families, schools and health services are not.
For many young people, talking to a chatbot may make a kind of emotional sense. It is available immediately. It does not look shocked. It will not tell your parents, roll its eyes, or put you on a waiting list. If you are embarrassed, ashamed, frightened, or just not sure whether what you are feeling is serious enough to bother anyone with, those things matter.
AI emotional support meets a real access problem
In New Zealand we should be particularly wary of pretending this is only a technology problem. We have a mental health access problem, and young people know it. They know that getting help can depend on where they live, what their family can afford, whether there is a counsellor at school, whether a GP has time, and whether the service they are referred to has space.
Seen from that angle, AI is not replacing care so much as filling a gap. Not always well, and not without risk, but often predictably. If the choice is between an instant response from a machine and waiting weeks or months to speak to a person, it is hardly surprising that some young people will try the machine first.
There is also the issue of judgement. Adolescence is a time when being seen can feel both desperately wanted and deeply exposing. A young person might want comfort, but not want the vulnerability of asking for it. They might want to test out words for what they feel before risking them with another human being. In that limited way, AI may sometimes function like a private rehearsal space.
What AI cannot reliably provide
But we should not confuse emotional responsiveness with care. A chatbot can generate plausible empathy. It can offer suggestions. It can stay available at three in the morning. What it cannot reliably do is hold responsibility for another person’s wellbeing.
That distinction matters. Good therapy, counselling, or youth mental health work is not just a collection of supportive sentences. It involves clinical judgement: noticing risk, context, patterns, family dynamics, trauma, depression, anxiety, substance use, coercion, and the many other things that may sit behind a simple statement like “I feel awful.” It also involves accountability. A clinician, counsellor, nurse, GP, teacher or youth worker is answerable to professional standards, supervision, law and ethics. A chatbot is not a relationship in that sense.
Continuity is another missing piece. Emotional distress often becomes clearer over time. What someone says today may only make sense alongside what they said last week, how they looked, what changed at home, what happened with friends, what they avoided saying. Human care is built out of this continuity. It is not perfect, and services can fail badly, but the relationship itself is part of what helps.
The New Zealand question
The temptation for adults is to respond by warning young people away from AI. Sometimes warning is appropriate, especially where safety, self-harm, abuse, or overwhelming distress is involved. In those situations there needs to be a real person, and preferably more than one, involved.
But if we stop at warning, we miss the message. Young people are telling us that privacy matters. Speed matters. Shame matters. Not being dismissed matters. So does being able to speak before a problem has become a crisis.
The challenge for New Zealand is not to build a mental health system that competes with AI by becoming more automated. It is to learn from what young people are seeking there, while being honest about what only human care can provide. Immediate access and lack of judgement should not be features unique to a chatbot. They should be basic expectations of a humane system.
Perhaps the most useful question is not “why are young people talking to AI?” but “what made that feel like the safest place to start?” If we can take that question seriously, we may learn something about technology. More importantly, we may learn something about ourselves.
