Can AI Be a Therapist? The Difference Between Chatbots and Psychotherapy

AI chatbots are increasingly used to answer mental health questions and help people express what they are feeling. Their accessibility, availability at any time, and seemingly non-judgemental tone may make it easier for some people to talk about difficult subjects. However, talking to a chatbot is not the same as working with a psychotherapist. AI may offer support in certain situations, but there is not enough evidence to say that it replaces psychotherapy.

What can AI chatbots do for mental health support?

A chatbot generates responses based on what a user writes or says. It may ask questions that help someone name their feelings, share general information, suggest ways to organise thoughts, or help a person keep track of their mood. These functions may support self-reflection and help someone clarify what they would like to discuss with a professional.

However, general-purpose AI tools are usually not designed to provide psychotherapy. Their mental health responses may be inaccurate, incomplete, or unsuitable for a person’s circumstances. The World Health Organization cautions that large language models can produce health information that sounds convincing but is incorrect, and that sensitive personal data may not always be adequately protected. 

How is a chatbot different from psychotherapy?

Psychotherapy involves more than receiving answers to questions. It is a process that develops over time between a therapist and a client, shaped by the person’s needs and life history. A therapist considers a person’s experiences, emotions, relationship patterns, and the effect of their difficulties on daily life. This understanding is not based on a single message or brief exchange.

A chatbot, by contrast, generally responds to text or voice inputs. It may not adequately assess cues such as facial expressions, body language, tone of voice, or pauses in conversation. It also does not have the same clinical assessment, ethical responsibilities, or individualised follow-up as a human therapist. The American Psychological Association advises that general-purpose generative AI chatbots should not be used as a replacement for psychotherapy or psychological treatment. Some tools may be used as an adjunct to ongoing professional care, depending on the circumstances. 

Can AI replace psychotherapy?

The answer depends partly on the type of tool being used. Some chatbots specifically developed for mental health support have shown promising results for certain groups in research studies. For example, a 2025 randomised trial reported that one purpose-built AI chatbot was associated with changes in depression and anxiety symptoms. However, this does not show that every general-purpose chatbot works like a therapist or is equivalent to professional psychotherapy.Findings need to be understood in the context of the specific tool and study conditions. 

It is more accurate to think of AI as a digital tool that may offer support in specific circumstances, rather than as a therapist. A chatbot cannot guarantee a diagnosis or a safe assessment in a crisis. If someone is experiencing severe distress, thoughts of self-harm, a risk of violence, or needs urgent help, relying on chatbot responses may be unsafe. Contacting a qualified mental health professional or an emergency support service is more appropriate. The APA also notes that chatbots’ ability to respond consistently and safely to crises may be limited and unpredictable. 

Why do privacy and personal information matter?

Messages sent to AI chatbots may contain sensitive personal information. Before sharing details about childhood experiences, relationships, health, or distressing thoughts, users should check how the platform collects, stores, and uses their data. Privacy policies and security measures vary between services.

Avoiding names, contact details, and information that could identify other people may be a safer approach. In professional psychotherapy, confidentiality and informed consent are important parts of care. With AI tools, how these protections apply depends on the platform being used.

When might AI be helpful?

AI may be useful for finding general information, organising thoughts in writing, or preparing questions before speaking with a professional. It does not replace individual assessment, diagnosis, or treatment. Information generated by a chatbot should be considered critically, and important mental health decisions should not be based solely on its responses.

Psychotherapy is not simply about providing ready-made answers. It offers a professional setting in which a person can explore their experiences within a therapeutic relationship. AI may accompany that process in limited ways, but a key difference is the clinical judgement, ethical responsibility, and individualised relationship provided by a human therapist.