Should AI be in the therapy room? What about using it to create materials for clients, or for note-taking and documentation? And how should we respond to clients who are already using AI to supplement their therapy?
There are so many questions coming up about AI right now for mental health professionals.
As location-independent therapists, the answers will often depend on where you’re licensed and what regulations apply to you. We don’t have all the answers, and we’re not here to tell you if or how you should use AI. But ignoring the topic completely isn’t helpful either, which is why we recently made it a month-long focus inside the LIT Community, facilitating conversations, sharing perspectives, and exploring it from many different angles.
This post is part 2 of our AI series. In Part 1, we looked at using AI for the backend of your business: admin, organisation, and marketing. Here, we turn to the more complex clinical side.
How AI is already showing up in therapy
AI tools are rapidly becoming part of everyday life, including how people seek mental health support. Recent studies suggest that one of the most common uses of LLMs (large language models, aka ChatGPT and the like) is for therapeutic support: talking through their thoughts, understanding their emotions, or researching what they’re experiencing.
Whether they’ve mentioned it or not, some of your clients may already be doing this. And at the same time, AI is increasingly being marketed to therapists for clinical workflows to speed up note-taking, session summaries, documentation, and practice management.
However you feel about any of this, it’s already part of the wider mental health landscape. The question isn’t really whether AI will affect your practice, but how you want to handle it when it does.
When clients use AI for support
Clients might use AI to help them understand what they’re experiencing, whether that’s anxiety, relationship conflicts, burnout, or grief. Some use it to organise their thoughts before a session. Others bring AI-generated insights into the room and want to explore them with you.
How does it feel when a client mentions using AI? It can be tempting to dismiss it or feel a little defensive. But as with most things clients bring, one of the most useful starting points is curiosity. Some questions that might open up an honest conversation:
- What made them turn to AI in that moment?
- What felt helpful about it, and what didn’t?
- Was there something they needed that felt hard to get elsewhere?
Sometimes AI use points to an unmet need: privacy, immediacy, a way to process something before speaking it out loud. A client using AI isn’t necessarily replacing therapy. In many cases, they’re trying to support themselves with the tools they have.
At the same time, there are real concerns about AI in the mental health space, such as the sobering reports of AI-induced psychosis. This recent segment on AI chatbots from Last Week Tonight with John Oliver is also well worth watching.
The harm caused by AI can also be more subtle. AI systems carry biases shaped by their training data, but clients may not recognise this and consider AI responses to be neutral or authoritative, thus internalising these negative biases. And while AI can sound convincingly like a professional, it lacks clinical judgement or any understanding of the specific person it’s talking to. That’s why as therapists, we’re well placed to help clients think critically about what AI has told them, and to notice when it may be doing more harm than good.
Using AI as a tool in your therapeutic work
Some colleagues find AI helpful for creating between-session materials. It can be effective at creating journaling prompts, psychoeducation handouts, reflection questions, and communication scripts for assertiveness practice. But others find it more frustrating than useful, or have concerns about accuracy and bias that make them reluctant to rely on it.
If you do use AI to create materials, your clinical judgement remains essential. While it’s important to review it to catch errors and hallucinations, your oversight is also essential because you understand the client’s history, risks, culture and context. While AI can generate a starting point, it can’t review its own output to determine whether it is appropriate for the client. That’s a uniquely human task.
AI and documentation: a bigger question than it might seem
Documentation is one of the fastest-growing areas of AI use among therapists, and also one of the most debated. Some colleagues use AI tools to draft session notes or summaries and find it meaningfully reduces their admin load. We’ve heard from several neurodivergent colleagues who shared that it’s been an important accommodation, freeing up their executive functioning capacity for the work itself.
Others have strong reservations. And for good reason.
What often gets overlooked in conversations about AI documentation tools is that this doesn’t just affect your workflow. If client data is being processed through a third-party tool, clients may need to be informed and may need to consent. Depending on the tool, this could mean disclosing that sessions are being recorded, transcribed, or analysed by an AI system. That changes the dynamic in the room. For some people, it will be a non-issue. But for other clients, it may significantly affect their sense of safety in the therapy room.
There are a few things worth reflecting on if you’re considering this:
- What exactly is being entered into the tool, and what could be inferred from it even if it’s depersonalised?
- Where is the data stored and how is it used?
- What are your obligations to inform clients under your jurisdiction and ethical framework?
- And does the AI output actually reflect the nuance of what happened in the session, or does it just sound polished?
There isn’t one right answer here, and it can be even more tricky to figure out as the regulatory landscape is still catching up with the latest AI technology. At the same time, it’s clear that questions of client privacy and consent are not optional considerations but something that every therapist should be thinking about when it comes to using AI in their practice.
Weighing up AI’s place in the therapy room
Like it or not, AI is likely to keep showing up in your clients’ lives and potentially in your own practice. Ignoring it doesn’t make it disappear, but embracing it without reflection isn’t the answer either.
What seems most useful is something in between: staying curious about how clients are using it and what needs it might be meeting, keeping up with developments as they unfold, and being clear with yourself about where your own limits are, ethically, professionally, and practically.
We’d love to know where you stand on all of this. Is it a dilemma you’re navigating right now, or something you’re keeping an eye on from a distance to see how things develop?
If you’d like a space to explore questions like this with colleagues who are working through the same things, the LIT Community is exactly that. You don’t have to figure it out alone. Start your membership here and join the conversation.
