The AI psychologist cannot replace the human professional

The debate on the use of artificial intelligence in mental health care is increasingly topical and relevant. The growing interest in conversational chatbots is set in a historical context in which psychological well-being has taken on an unprecedented centrality, not least in the wake of the Covid-19 pandemic and the rapid digitisation of health and care services. Faced with a growing demand for psychological support, often inadequately intercepted by traditional services, more and more people are spontaneously turning to digital tools, perceived as accessible, fast and free of social barriers.

Advanced language models such as ChatGPT are increasingly being used as a first contact with psychological language: an initial form of exploration of one's inner world, an outlet or an automated presence that attempts to answer personal questions and suffering. Its continuous availability, ability to produce articulate answers and the illusion of an empathic relationship increase its attractiveness, especially for segments of the population that struggle to access professional forms of support.

In this framework, these tools can play a complementary role, facilitating emotional literacy or representing an entry threshold towards the demand for help. However, their progressive diffusion raises relevant questions regarding the borderline between digital support and specialist psychological intervention, the quality of the interaction they propose, and the possibility - far from remote - that they are progressively perceived as substitutes for an authentic and transformative clinical relationship. It is precisely on this ridge - between the function of facilitation and the risk of substitution - that one of the most complex current challenges in mental health is played out.

There is also a danger that healthcare systems, under economic and organisational pressure, are tempted to delegate to these technologies tasks that require human relational skills instead. In a context where mental health resources are often insufficient and waiting times for access to public psychological therapy are lengthening, artificial intelligence may appear as a simple and low-cost solution. However, this shortcut risks turning into a further form of inequality: people with less economic means would be offered automated tools, while only those who can afford it would continue to have access to comprehensive and personalised treatment paths. This technological proxy, if not carefully governed, could thus weaken the very mandate of public health, which should guarantee a human and competent care.

Psychological support and psychotherapy are complex processes based on building a relationship of trust, emotional processing and the encounter between two subjectivities. The dynamics that are activated in the therapeutic space - from affective resonances to silences - are impossible to reproduce artificially. Empathic listening, the ability to pause in doubt, to tolerate ambiguity and to remain close to the other's pain, remain prerogatives of the human being. The possibility of profoundly understanding the emotional experience of others is based on precise neurobiological foundations, as demonstrated by research on mirror neurons: brain circuits that are activated when we observe emotions, gestures and experiences in others, enabling authentic emotional resonance.