I don't know much about therapy, but this article appears to be a good analysis of what current therapy is about, why it works, and why AI might not be able to replace it.

But some of the things mentioned, such as: your therapist gets moved by your story themselves, they are themselves imperfect, and in general you're talking to a human being you have to respect etc - these things, while they definitely do support current therapy in important ways, they are not a deliberate design choice of how therapy works. They are rather constraints that come about when you are using a human being as a therapist, made out of flesh. So, the last point the article makes - that therapy itself might mean something else in the future - is the real question, in my mind.

The current form of therapy has incredible strengths in part because it works with these human constraints. But AI based therapy might be able to fill a different gap, one that regular therapists cannot fill - on account of, for example, not being available 24/7, not having the capacity to read entire diaries or chat histories, or taking on different personas. Or, simply, being cheaper than a real therapist, which can benefit many even if the ultimate quality is worse.

I agree with the conclusion but not necessarily the why. I don't think therapy is always about projecting your other relationships on to your therapist. A lot of time its your relationship with yourself that you're trying to work through. Or sometimes you just need someone who can observe and challenge you. Or sometimes you just need the space to be honest about what's going on in your mind where there won't be the social consequences of saying that to your friends or family.

I agree that the frame of the therapist-client relationship is key. I agree that AI simply cannot replicate that. You need someone on the other end who has both the deductive ability but also the emotional intelligence to either suss out or help you suss out what exactly is causing discomfort. And AI just can't create that frame because of how we expect to use it.

What's really crazy is therapists are starting to give their clients AI slop as part of their value-add, using it to create writing for clients to take away that resonates more with clients... and the AI is prompted without all the context known to the therapist so a bunch of the specifics and assumptions made by the AI in its advice don't track to other family/friends observing the output but do relate to some hypothetical pattern which doesn't apply.

Misdiagnosis from a therapist is not new, but AI use by the therapist is.

People have had itching ears to hear what they want to hear for a long time, and recently we have all seen in the press stories of AI-only therapy sycophanticly reinforcing delusions, but it is so strange to see a professional therapist human-washing such stuff.

But it makes sense; therapists are not yet savvy to all the weaknesses of LLM analysis and framing bias and they too can be tempted themselves to ask an AI and just pass outputs upwards like various work colleagues I've had...

Still, I was quite disturbed about the unprofessionalism of the therapist when I heard about it. I think we'll see this come up more in the press. like we've seen with lawyers. But not as soon because lawyers have a validation function called a judge.

I never found a therapist able to go as deep as I wished to help me solve trauma. They always end up focusing on short term solutions to help me get more “functional”. But with AI, I could go as deep as I need, and I unlocked 30+ year old buried trauma that I could heal from in 6 months. Granted it’s not for everyone and I studied a lot of psychology to get there, but it was worth it. Fuck (most) therapists.