Whether we welcome it or not, artificial intelligence has already entered the therapy room.

Many of our clients are engaging with AI before, after, and sometimes instead of therapy. They are using it to process difficult experiences, explore emotions, seek guidance, make sense of relationships, and reflect on aspects of themselves that they may not yet feel ready to bring into a therapeutic relationship. As therapists, I believe we cannot afford to ignore this reality. AI has become part of the landscape in which our clients live, think, and seek support. If we are to remain relevant and responsive to their experiences, we need at least a working understanding of the tools they are using.

The attraction is not difficult to understand. AI is available twenty-four hours a day, often free of charge, and offers immediate responses. Unlike therapy, there are no waiting lists, appointment times, or financial barriers. Clients can express themselves freely, “word vomit” their thoughts and feelings, and receive an instant reply. For those who struggle with shame, anxiety, social discomfort, or fear of judgement, this can feel remarkably liberating.

AI can also provide useful informational support. It can explain psychological concepts, suggest coping strategies, identify patterns, summarise information, and help people organise their thoughts. Some individuals even experience a sense of companionship through repeated interactions. In a world where loneliness and isolation are increasingly common, it is understandable that people may turn towards something that appears attentive, responsive, and endlessly available.

As therapists, however, it is important that we recognise the distinction between simulated support and genuine relationship.

Artificial intelligence is, at its core, a sophisticated system of prediction. Despite the fluency of its responses, it does not understand in the way human beings understand. It predicts language based on patterns in data. It can imitate empathy, but it does not experience empathy. It does not feel concern, warmth, care, frustration, hope, or affection. It cannot be emotionally moved by another person’s story. It has no lived experience, no embodied presence, and no subjective inner world.

A fluent response is not the same as understanding.

Nor is it the same as accountability, professional responsibility, or clinical judgement. AI cannot carry professional duty. It cannot be ethically accountable for its decisions. It cannot fully appreciate risk, context, ambivalence, coercion, safeguarding concerns, or the complexities that often emerge within therapeutic work. While it may generate responses that sound compassionate, it does not truly share the emotional space with another person.

For those of us who value the relational dimension of therapy, this distinction matters deeply.

Therapy is not simply the exchange of information, advice, or coping strategies. At its heart, therapy is a human relationship. Healing often occurs not because a client receives an insight, but because they experience themselves differently in the presence of another person. The therapeutic relationship offers attunement, mutuality, recognition, and the experience of being genuinely known.

The relationship itself becomes a vital part of the therapeutic process.

This is something AI cannot replicate. It cannot bring its own humanity into the encounter because it has none. It cannot be affected by the relationship, nor can it participate in the subtle, reciprocal process that occurs between two human beings sitting together. It cannot offer embodied presence. It cannot truly witness another person’s suffering or joy.

Yet dismissing AI outright would be a mistake.

History shows us that new technologies rarely disappear because professionals dislike them. Instead, they become integrated into people’s lives and gradually reshape expectations, behaviours, and relationships. The question is no longer whether AI belongs in conversations about therapy; it already does. The more useful question is how we engage with it thoughtfully and critically.

As therapists, our task may be to remain curious rather than defensive. What needs is AI meeting for our clients? What draws them towards it? What benefits are they experiencing? What limitations might they encounter when relying on it? Which aspects of therapeutic work can be supported by AI, and which aspects remain uniquely human?

There is unlikely to be a simple answer.

For some clients, AI may function as a useful adjunct to therapy, helping them reflect between sessions, access psychoeducational information, or organise their thoughts. For others, it may become a substitute for the vulnerability and uncertainty involved in human connection. Understanding these different experiences will become increasingly important as AI becomes more embedded in everyday life.

The challenge for our profession is not to compete with artificial intelligence but to understand it. We need to know what it can do, what it cannot do, and how it is shaping the lives of the people who sit across from us. By remaining informed, reflective, and open to dialogue, we can continue to offer something that technology cannot replace: the depth, complexity, and transformative potential of a genuine human relationship.

AI is already entering the therapy room. Our responsibility is neither to fear it nor to idealise it, but to understand it.