Can AI replace therapy?
21st September 2026
There’s lots of discussion now around AI and whether it’s helpful or a threat to humanity.
AI platforms, such as using Google, Chat GPT, Claude or Gemini are already being used in many areas of our lives, but I’d like to explore its current use within mental health support and therapy.
Can AI replace therapy? … let’s look at the pros and cons and explore this further.
Some ‘positives or advantages’:
- AI is a 24 hour, 365 day service - AI chatbots are available to access 24 hours a day, making them accessible to all, anytime, on demand. We could argue that having instant interaction and answers could be positive, but my concerns are whether the information fed back is accurate or even appropriate?
- Easy to ‘talk to’ - Talking to an AI Chatbot could be less anxiety provoking than speaking to a therapist or mental health practitioner, easier to organize and it may help break down the ‘stigma’ of accessing therapy.
- Can help raise self-awareness - Some tools can be useful for self-reflection or journalling.
- Can help with anxiety and stress management - AI can suggest simple techniques or strategies such as grounding or relaxation techniques to help with anxiety and stress management.
- Signposting - AI may give signposting advice, for example suggesting contacting your GP.
Some ‘negatives or disadvantages or concerns’:
- AI is not human! - AI does not understand what being human is like, how we feel, the complexity of our thinking. Therefore, it lacks empathy and understanding. It is not the same as talking to or being supported by a therapist or mental health practitioner.
- Safeguarding - AI does not have the ability to manage crisis. Therapists can react appropriately if concerned about someone’s safety, are bound by ethical codes of conduct and practice, have a duty of care and are accountable for their practice and therapy. Therapists hold ethical body and professional memberships, have indemnity insurance, supervision and need to undertake continuous professional development (CPD). All of which helps safeguard the client and therapist.
- Privacy concerns - AI uses shared data! Sharing personal information, thoughts and feelings with online platforms is not guaranteed to be secure and considering the AI model is built on collating data I think great consideration needs to be given as to where your information goes. Therapists must keep records but again are bound by ethical frameworks to be confidential and we only share what may be appropriate during safeguarding concerns.
- Lack of empathy and understanding - AI doesn’t have the depth of understanding as a human being. Therapists have years of training and experience in understanding human connection, thought patterns, body language, the ability to pick up on what’s not said, subtle cues, pauses and reading between the lines. AI reacts literally to what is being input, without empathy and understanding.
- Concerns for the young - All these concerns are magnified when AI is used for mental health support by young children and younger adults. Both of which are highly impressionable and don’t yet have all the resources or life experience to question or look at different perspectives. The danger being that what is being given or suggested by AI is truth and must be believed.
- Data led - AI responses are data led and are very much dependent upon the information it is given. Therefore, responses can be very biased and worryingly could validate harmful thinking and negative thought patterns. AI doesn’t ‘challenge’ a person’s beliefs or thought patterns as a therapist may.
- Dependency - this can be a particularly worrying factor. Some may become dependent or reliant on ‘talking’ to another intelligence which could promote isolation or withdrawal from human interaction. AI can be seen as a readily available ‘friend’ or crutch which may encourage isolation from society and promote a lack of self-belief.
- Taking away our ability to cope? - Using AI for support can take away the ability to self soothe and self-reassure. It can decrease our ability to tackle life using our own resources, essentially taking away our inner instinct, inner power and trust in ourselves. It can reduce our self-awareness and ability to develop our own coping mechanisms.
- Accountability - AI has no accountability. Therapists are accountable for the service they provide and are contactable/accessible to address any concerns.
- Un-regulated - AI, its contents, interactions, feedback and suggestions are not regulated, which again is a safeguarding concern.
- AI does not see the whole picture - AI does not consider the ‘full picture’, again just responding to what is being given in the moment. Effective therapy takes time, is built on rapport and trust, where a therapist will also consider life circumstances, experience, resources, previous traumas, patterns and support network etc.
- Not a safe space - Mental health support with AI can be accessed anywhere, anytime and does not understand whether a person is safe, in a dangerous situation, unstable or in a good mental place to explore their thoughts. Therapists provide a safe space in which to share, express and explore thoughts and feelings at an appropriate pace.
- AI must surely be right? - There is a danger of believing all that is stated by AI is true.
So, considering all the above, what feels safest to you?
I understand that AI can have some advantages, but my overall feeling is a deep concern of using AI in a therapeutic sense. I do not believe AI can or ever should replace therapy.
"One friend, one person who is truly understanding, who takes the trouble to listen to us as we consider our problems, can change our whole outlook on the world."
- Dr. Elisabeth Kübler-Ross
