The number of Gen Z and Gen Alpha using AI diagnosis tools is rapidly growing, with approximately one in five young people globally now turning to AI for mental health advice, up from one in eight the year before. Most research and media coverage is focused on whether its advice is accurate enough to be comparable to that of professionals, or whether it has reached clinical reliability, but as someone living through this shift, I would like to explore the why behind our tendency to use AI in this way. Not through the already established reasons of accessibility and convenience, but through the harsher truth beneath them.
Before going further, let’s rewind and remember what artificial intelligence actually is. It is a large language model trained on patterns in text and information it has been pre-fed. It has no senses or feelings, and no memory of you between conversations beyond surface facts about who you are. It is in no way a licensed clinician who has undergone the professional and ethical training required to advise or interact with another human being on something so personal and complex, and most importantly, it has no stake in what happens to you. Ouch. AI is a machine optimised to produce responses that you would feel are helpful, so when you are using it in conversation and feel like it is understanding you, it is very likely just agreeing with you. It is a sign the training worked, because when it offers advice it is not drawing on knowledge of you or the world in the way a human would. It is simply generating plausible text.
That is important to understand, because it is the very reason young people find AI comforting. It reflects back a version of themselves that they want to see and need in order to feel validated, whereas a friend or therapist would offer real, honest feedback. The pull toward that kind of reflection is stronger in adolescence than at almost any other point in life, because it is a period of significant hormonal and neurological change, during which wellbeing is unusually sensitive to reinforcement, positive or negative.
The risks of this look different for every user depending on their condition and how they use it. For example, say you had an argument, and in a conversation with a chatbot you reason with yourself that the decision or action wasn’t actually bad, and walk away feeling more convinced. Now imagine a more extreme scenario. There are real documented cases of young people in genuine crises who have spent months confiding in chatbots that produced fluent, sympathetic responses while failing to identify that the person was dealing with something far more serious, such as severe depression, suicidal ideation, or an undiagnosed condition requiring urgent professional care. Their families often don’t find out until afterwards, because a chatbot has no mechanism, and no obligation, to tell anyone.
At the same time, it is also important to acknowledge that seeking support is still treated as a taboo in some communities. Even if a young person wanted to seek help, they aren’t often independent enough to do so themselves and would have to go through their parents, which becomes harder given the taboo already sitting around it at home. And because it is easier to avoid sharing our feelings, the more we start relying on AI for conversation about our inner lives, the more it could pull us, especially the younger end of our generation, away from real human connection. It might make us more closed off, because the need is being met somewhere else, somewhere much easier to access and far less judged.
So what can we do about this? The tools need to work differently. The current standard of a disclaimer printed in small text at the bottom of a response is not a meaningful safeguard. Systems should be able to recognise when a conversation has moved into something that might affect a young person’s wellbeing, and respond in a way that stops validating and points clearly toward a real person or professional who can help. The harder task falls on the adults and decision makers to prioritise youth wellbeing and create spaces and funding for organisations that bring real support to young people. Schools and communities also need to teach that everyone has mental health in the same way everyone has physical health, and to diminish the stereotype of it being a sign of dysfunction or failure.
The reasons young people are turning to these tools shouldn’t be ignored, and are equally, if not more, important than understanding the accuracy of the tools themselves. The question my generation needs answered is not whether AI has reached the bar in diagnosis but why so many of us reached for it in the first place.
Check out more from the SHIELD Youth Council: