
I have condensed here some of the findings from The Psychologist magazine in June:
AI is now being used sometimes as a source of mental health support, including by children and young people, despite limited clarity about responsibility, boundaries or risk…Also AI is on hand 24 hours a day rather than in a defined therapeutic space… When psychologists work with people there is no constant availability and this is not just to protect the therapists but also to empower the patients/clients to cope beyond the therapy room. AI systems are only designed to respond, not to reflect or challenge. Chatbots are essentially designed to inform and entertain, whereas a human therapist may sometimes have to present an uncomfortable truth to a client.
AI is a large language model (LLM) that is built from written language, but it can’t use the interactional infrastructure of conversations in which language evolves…. It is a tool, not a relationship….it just simulates intimacy. It cannot replace intimate therapeutic relationships, because the software lacks the embodied judgement and ethical responsibility such relationships require… A machine has no concept of truth, ethics or social bonds…Some chatbots have even encouraged self-harm, suicide and validated delusions and addictions! The chatbot has no wisdom or professional judgement, it is simply playing along and generating text that is most likely to engage the user.
AI is particularly unhelpful when working with adolescents, even though they think it is helpful. Adolescence is a developmental period characterised by emotional reactivity and increased sensitivity. Which means that for them it is often difficult to emotionally regulate and tolerate uncertainty. So adolescents who are vulnerable to rumination, anxiety, low mood and fragile self-esteem, affirmation from AI without challenges can become a trap…Constant affirmation is not a marker of good care (hence the reason why passive parenting doesn’t work), it is rather an IT design choice aimed at increasing engagement. Compare this with CBT, the therapy that I use with my clients, which has as it’s central principle that thoughts are not facts. In therapy distressing thoughts are not accepted at face value, however emotionally charged they may feel. Rather the therapist and client slow the process down, examining assumptions, testing alternative explanations and asking questions that sometimes feel uncomfortable. CBT requires cognitive flexibility, inhibition of automatic responses and the capacity to hold multiple perspectives simultaneously. Whereas a chatbot will involve less mental energy, will affirm intuitions, mirror emotions and construct narratives that don’t require mental effort…In fact a chatbot will encourage rumination, not reflection. Engagement is not the same as emotional recovery.








