The whiplash around AI these days is real: One minute, we’re hearing from the most powerful AI leaders of our time that the technology could arguably kill us all. A minute later, we’re seeing an ad for the latest AI therapy-bot or app appear in our algorithm, offering a cheaper, more convenient alternative to human therapists. All this is happening within the larger geopolitical and environmental context of the debate surrounding AI data centers and the enormous ecological cost of running these systems. This begs the question: Can you, or should you, consider an AI therapist? Human Versus AI Throwdown: Therapy Edition If we momentarily put aside carbon footprint concerns, recent research directly comparing the effectiveness of AI versus human mental health professionals is telling. When researchers showed therapists transcripts of conversations between patients and 1) an AI therapy bot and 2) a human therapist, the professionals couldn’t reliably tell the difference between the two. Their performance hovered around chance (53 percent accuracy). What’s more, the transcripts with the AI therapists scored higher on quality. …

Source: NoSystem Images / Getty Images Signature It’s a not-so-well-kept secret. Therapists are nosy. Maybe not all therapists, but c’mon. Who would want to go into a profession where you listen all day to other people’s problems, secrets, quandaries, and so forth? Nosy people. Sure, we want to help you, absolutely. But lots of people want to help others and don’t become therapists. Those people (after a socially acceptable number of minutes of other people going on about personal problems) say, “Dude, you should really go talk to somebody about that.” Therapists can’t say that. They are the somebody. Over the therapist years, I learned so much more than my nosiness and desire to help others ever suggested I might. And I owe that to all my clients. Thank you, each and every one. Thank you for making me keep learning, all the time. Whenever you had a problem, it wasn’t that it became my problem, but if I didn’t know about it, I wanted to learn. From reading studies that taught me—for instance—that couples …

When I speak with colleagues and students about AI therapy, the concern that I most often hear is that AIs are “mindless”—that they are “robots” that “parrot” whatever melange of cognitive behavioral therapy and other modalities that they may have picked up from textbooks and Wikipedia. What a client wants from her therapist is a real person, with a real mind, who will respond to her genuinely; their concern is that an AI therapist—by which I mean, roughly, a large language model used or adapted for a therapy-style conversation—can simply never do what a human therapist can. I think it is useful to approach the question from the opposite direction. Let us take seriously the hypothesis that an AI does have a kind of mind – that it has representations, utilities, and perhaps even moods and emotions – and ask whether the kind of mind that it can be supposed to have is suitable for the work of doing therapy. My view is that it may not be, and that this points to a different …