Therapy and AI- pros, cons and ethics.

The current discourse around AI as a substitute for traditional in-person therapy is fascinating -equal parts exciting and disturbing. I prompted my chatbot to list the benefits, and I agree with some of its observations up to a point.

I’m not one for gatekeeping access to therapy. Everyone deserves help whether they’re experiencing a crisis or just feeling a general sense of ennui. Sometimes we just need someone to listen. Public and volunteer mental health services are often under resourced and private counselling is unaffordable for many.

The chatbot agreed with me on the benefits of immediate and low-cost access to therapy delivered by AI. Additionally it pointed out that it could provide consistent support for ‘low level concerns’ and a ‘daily emotional check-in’. Which begs the question: what is a ‘low level concern? Who gets to decide the status or urgency of an issue? Things that seem small may add up over time, having a huge impact on one’s life. As a therapist I believe that all concerns are valid and worthy of attending to.

The chatbot claimed that the perceived anonymity it provided is helpful to anyone hesitant or embarrassed to seek in-person therapy. But here’s the thing, your chatbot is afforded anonymity, not you, the user.

This brings up an interesting ethical issue regarding confidentiality and privacy. As a practicing therapist I do not use AI tools to record sessions or take notes because I do not know where the AI company is storing the data or what they are using it for. Thus I cannot claim confidentiality. Additionally, therapists who are members of professional bodies such as the BACP (United Kingdom) or HPCSA (South Africa), are bound by ethics that protect the client and therapist. Chatbots and the tech companies that develop them are untethered from any ethical concerns. This means they are unaccountable.

Speaking to the hesitancy of someone to bring an embarrassing issue to a human therapist, well, there’s the rub. I’ve previously written that ‘the relationship is the therapy’, that a therapist who embodies empathy, non-judgment and congruence creates a safe environment for a client to explore their issues. This is the foundation of person-centred therapy. In every relationship, including the therapeutic alliance, we must risk vulnerability as a means to build connection and emotional depth.

This is how we heal and grow as individuals.

My chatbot conceded that AI therapy is not suitable for assistance with issues of deep and complex trauma or severe mental crises. I would add that anyone looking for help with issues of addiction and violence should immediately seek professional help. A GP can assist here. The sense of connection that is created between two human beings is something I believe a chatbot can simulate but never fully replace. To be fully seen and heard by another person is a transcending experience.

A chatbot can be relied upon to deliver certainty in responses to prompts and questions, but beware: certainty is the enemy of tolerance. If one cannot tolerate one’s feelings how can one begin to heal.

Psychologist Carly Dober sums things up in an article for The Guardian, “Good therapy is

relational. It thrives on imperfection, nuance and slow discovery. It involves pattern recognition, accountability and the kind of discomfort that leads to lasting change. A therapist doesn’t just answer; they ask and they challenge. They hold space, offer reflection and walk with you, while also offering up an uncomfortable mirror.”

As evidenced by recent upgrades, AI companies can change the ‘personality’ of chatbots at anytime - what do you do when your robot therapist suddenly changes its demeanour ? So in conclusion, be discerning when interacting on personal issues with AI.

Your mental health and private data are at stake.

Further reading: https://www.theguardian.com/commentisfree/2025/aug/03/generative-ai-chatbot-therapy-dangers-risks