Like it or not, AI is here to stay, and people are increasingly turning to it for companionship… and even therapy. The fact that people think AI can replace a human therapist indicates to me that there’s a need for more education on what a human therapist actually does. We don’t just offer “value added” over and above what an AI companion might give. It’s a completely different beast. And in some very important ways, being human is actually essential to being a therapist. Seems so obvious it doesn’t even bear stating, right? But it’s not obvious to everyone. And these questions cut to the heart of what therapy is.
There’s one obvious difference that bears stating right off the bat: AI does not have ethical standards to uphold the way a regulated therapist does. A conversation with an AI is absolutely not confidential, whereas therapy has a lot of legal and ethical boundaries that enforce confidentiality. It is the most important commitment a therapist makes to a client, and it is sacred. Another serious consideration is that an AI has zero responsibility, professional or ethical duty when it comes to doing harm. In fact, in some cases, AIs have been known to encourage users to complete suicide. Therapists are bound by law to act when there is a significant risk of serious harm to the client or another person (including self-inflicted harm). Standards like these save lives.
A good therapist is trained to uphold ethical standards at all times, with good judgment, for your safety. When we’re not sure what to do or need support, we get help from our own therapists and clinical supervisors. A good therapist doesn’t go rogue or make it up as they go along.
But, therapists being human is a double edged sword. Some therapists, despite all their training, do harm their clients. That’s where professional accountability comes in. If something goes wrong in your therapy, you have recourse: you can complain to your therapist’s regulatory body, who are there to protect the public and who will ensure that more people aren’t hurt. If you or a loved one are seriously harmed by an AI, the best you can hope for is to launch a lawsuit against a very large corporation.
Therapists hold boundaries: we end the session on time, keep your secrets, and don’t text you at 3 a.m. Boundaries aren’t just an incidental thing that gets in the way of a good time (like texting at 3 a.m.). Having thoughtful structure around the therapeutic relationship is actually part of the therapy: it conveys trustworthiness, provides an assurance of stability and predictability, and in some pretty significant ways, boundaried structure communicates safety and respect. For many clients, safe boundaries are an entirely new experience. They are a precondition for healing. A relationship with a therapist is a lot different than a relationship that will allow you to fall ever deeper into the wormhole of your own paranoia. For example.
There are other key ingredients to therapy that AI falls short on. They’re kind of hard to pin down, which is also probably why so many people miss them when comparing human therapists with AI. But I think they’re well worth articulating. Look here for my article on why therapy requires genuine human connection. Another obvious thing I never thought I’d have to articulate!