AI chatbots and youth mental health

Read time: 3 minutes

A US survey of young people aged 12 to 21 found that nearly one in five have used AI chatbots for assistance with mental health issues. Most of them kept it to themselves. Many found it useful. Around 43 per cent used chatbots for mental health advice at least monthly, and many had already seen a doctor. The older the young person, the more likely they were to have used AI for this purpose. Females were more likely to use it than males.

This could be good news, if the advice being given is reliable and the systems have adequate guardrails in place for young people who may be at risk of self-harm. The trouble is that AI chatbots as mental health interventions have not been well studied, and there have been high-profile cases of young people finding ways around inbuilt safety systems, with the chatbot subsequently facilitating rather than preventing harm.

Chatbots are designed to be friendly, engaging and empathic, whereas clinical psychologists are trained to be objective and focussed on the real problems of the person.

The problems with AI chatbots in this context are more fundamental than safety settings. The way these systems are trained to conduct therapeutic-style conversations is not transparent. More importantly, chatbots are designed to be friendly, engaging and empathic, whereas clinical psychologists are trained to be objective and focussed on the real problems of the person, steering them towards a clear therapeutic strategy. These are not the same thing, and in some presentations they are directly at odds. A system optimised for engagement may keep a distressed young person talking without ever moving them towards help.

There is also the question of what happens when a young person develops a reliance on a chatbot as their primary source of emotional support. Research on parasocial relationships with AI is still in its early stages, but the concern is that chatbot interaction may substitute for rather than supplement professional help-seeking and real-world connection, particularly in young people who are already socially withdrawn or anxious.

For general practice, the practical implication is straightforward but easy to overlook. Young people presenting with mental health concerns may already be using AI tools regularly, and they are unlikely to volunteer this information. Asking directly and non-judgementally about AI use opens the conversation. The goal is not to discourage it categorically, but to understand what they are using, how they are using it, and whether it is complementing or replacing professional support.