Imagine you’re sitting across from your therapist and they mention something about using an AI tool in your work together. Your first reaction might be curiosity. Or it might be something closer to alarm. Is a chatbot going to analyze me? Is my therapist being replaced by a machine? Am I going to be talking to a screen instead of a person?
These are fair questions. And the honest answer is that most of what people imagine when they hear “AI in therapy” is not what’s actually happening — or what should be happening — in a thoughtful clinical practice.
AI in Therapy Is Not What You Think
When AI comes up in the context of mental health, the conversation tends to jump straight to chatbots and apps — tools that try to simulate a therapeutic interaction without a therapist present. Those products exist, and they have their own set of limitations and concerns. But that’s not what we’re talking about here.
The kind of AI integration that interests us at Therapy Aligned is something different. It’s about a clinician using AI as a tool within the therapeutic relationship — not as a substitute for it. Think of it less like replacing your therapist with software and more like your therapist picking up a new instrument. The musician is still the one making the music.
The Third Chair: A Framework for AI in the Therapeutic Room
Andrew has proposed what he calls the 5-Level AI-Integrated Psychotherapy Framework — a public Open Science Framework project that describes how AI might sit alongside the therapist and client in session, not as a participant, but as a resource.
The idea is straightforward. In a therapy room, there are traditionally two chairs: the client’s and the therapist’s. The 5-Level AI-Integrated Psychotherapy Framework introduces a conceptual third chair — a place for AI tools that the therapist can draw on when clinically appropriate, always with the client’s knowledge and consent.
Critically, the therapist remains what researchers call the “human-in-the-loop.” That means no AI tool makes clinical decisions. The therapist decides when and whether to bring AI into the conversation, interprets whatever the tool offers, and filters it through their clinical judgment and their knowledge of you as a person. The technology serves the relationship. It doesn’t steer it.
This framework is still being developed and tested — it’s a public Open Science Framework project, not a finished product. But it represents a careful, clinically grounded way of thinking about a question that the field needs to take seriously: how do we use these powerful new tools without losing what makes therapy work?
What This Might Actually Look Like in a Session
Your therapist might use an AI tool to help you co-create a prompt — a structured question or reflection exercise — that you work through between sessions. Rather than a generic worksheet, this could be something shaped by the specific themes you’ve been exploring together, tailored in the moment to where you are in your process.
In session, an AI tool might help surface patterns across what you’ve shared over time — recurring themes, emotional shifts, connections you haven’t noticed yet. Your therapist would review these with you, not as conclusions, but as starting points for deeper conversation. The insight still belongs to the therapeutic relationship; the tool just helped bring something to the surface a little faster.
None of these examples involve AI acting autonomously. In every case, the therapist is guiding the process and the client is informed and involved.
The Safeguards: Privacy, Consent, and Clinical Judgment
Any responsible use of AI in therapy has to start with three non-negotiable commitments: privacy, consent, and the primacy of clinical judgment.
Privacy means that your therapeutic material is treated with the same — or greater — care as it always has been. Introducing a tool doesn’t mean your session notes are being fed into some open system. Consent means you know what’s being used, why, and you have a genuine choice to opt out. This isn’t a terms-of-service checkbox. It’s an ongoing conversation between you and your therapist.
And clinical judgment means that the therapist’s assessment always takes precedence over any output an AI tool generates. If an AI suggests something that doesn’t fit your experience or your clinical picture, your therapist sets it aside. The tool is subordinate to the relationship — always.
Why This Matters for You
The goal is more tools, not less humanity. Therapy works because of the relationship between two people. Nothing about thoughtful AI integration changes that. What it can do is give your therapist additional ways to support your growth — more precise reflections, more personalized between-session resources, more ways to track and honor the progress you’re making.
At Therapy Aligned, we believe technology should serve the therapeutic relationship, not replace it. Andrew’s ongoing research through the 5-Level AI-Integrated Psychotherapy Framework explores how to bring these tools into clinical work responsibly, transparently, and always in service of the people we work with.