My patients use ChatGPT for therapy. Now I use it too | Sarah Dargouth
I can’t blame my patients for turning to its straightforward assessments. But it has real risks – and care may require human messiness“Chat told me I should break up with him.”I in
By The Guardian
My patients turn to ChatGPT for therapy advice, and now I use it too, though I remain wary of its risks and the essential human messiness that care demands. Sarah Dargouth, a therapist based in Cheshire, recounts how an AI chatbot delivered a relationship verdict her patient had long sensed but struggled to articulate, prompting her to adopt the tool cautiously while highlighting the unresolved dangers of relying on algorithms for emotional guidance.
The moment arrived during a session when her patient, a young woman, revealed that ChatGPT had told her to break up with her partner. Dargouth tried to maintain her therapist-neutral stance, yet a smirk betrayed her annoyance.
For weeks, they had debated the viability of the relationship, and in an instant, the AI had provided the answer the patient had been circling. When Dargouth asked how the patient felt about the suggestion, she admitted it had been her gut feeling all along.
By the following session, the relationship was over, a decision accelerated by the straightforward, unemotional clarity of the machine. This anecdote reflects a growing trend across the UK and beyond.
An international survey spanning 30 countries found that more than half of psychotherapists now use generative AI tools like ChatGPT in clinical practice, primarily for treatment planning, administrative tasks, and generating psychoeducational materials for patients. ChatGPT dominates this emerging landscape, with nearly 85% of AI users in mental health relying on OpenAI’s chatbot, far ahead of Google Gemini or Microsoft Copilot.
Treatment planning emerged as the most common use, reported by 41.9% of AI users, while administrative tasks followed closely at 41.4%. Yet the adoption of these tools is outpacing regulation, training, and governance.
Only 18% of respondents said their employer or professional organization had encouraged AI use, and just 18.3% had received any formal training. More than four out of five clinicians reported using AI without professional education or structured guidance.
Despite this, nearly 83% of therapists who use these tools say AI reduces their overall workload, a significant incentive in an increasingly strained healthcare environment. The effectiveness of ChatGPT in therapeutic scenarios is unsettling.
In some studies, its responses were rated higher than those of human therapists across core measures including empathy, completeness, and clarity. ChatGPT has even passed the Turing Test, making it difficult to distinguish between human therapists and AI in certain contexts.
Young adults appear particularly open to talking to AI, with trials showing they are more willing to engage with digital therapists than older generations. One University of British Columbia study found that ChatGPT can be so persuasive that it affects how long patients spend with their doctors, altering the traditional dynamics of care.
However, Dargouth warns that the greatest unresolved issue is not accuracy but confidentiality. When patients read aloud their conversations with AI chatbots or share digital transcripts, sensitive data may be exposed to platforms that are not designed for mental health.
ChatGPT is not built for therapy, and its straightforward assessments, while helpful, carry real risks. The algorithm lacks the capacity to understand the nuanced, often contradictory nature of human emotion, and it cannot replicate the “human messiness” that Dargouth believes is essential to care.
In professional forums and private conversations, therapists increasingly swap stories of clients bringing AI-generated advice into sessions. Some patients use ChatGPT to navigate difficult decisions, ask questions about their feelings, or seek validation for choices they are hesitant to make.
Dargouth herself has never used ChatGPT as a therapist, though a close friend does and reports positive outcomes. She acknowledges that she can no longer blame her patients for turning to the tool, given its accessibility and clarity, but she insists that care must remain rooted in human connection.
The debate is not about whether AI should be used, but how it should be integrated safely. Researchers conclude that professional competency frameworks, evidence-based guidance, and regulatory clarity are urgently needed to ensure safe integration of AI into mental healthcare.
Without these safeguards, the risks of confidentiality breaches, misdiagnosis, and overreliance on algorithms could undermine the trust that forms the foundation of therapy. Dargouth’s experience underscores a broader shift in Cheshire and across the UK, where patients and therapists alike are navigating a new landscape shaped by artificial intelligence.
As ChatGPT becomes more embedded in clinical practice, the challenge will be to balance its efficiency with the irreplaceable value of human empathy, ensuring that technology serves care without replacing the messy, complex, and deeply personal nature of healing. The story of Sarah Dargouth and her patient is not just about one relationship ending; it is a snapshot of a transformation in how mental health care is delivered.
As AI tools become more persuasive and accessible, therapists must remain vigilant, using them as supplements rather than substitutes, and always prioritizing the human element that defines true care. In Cheshire, as elsewhere, the future of therapy will depend on finding this balance, ensuring that technology enhances, rather than diminishes, the human connection at the heart of healing.