AI and the Future of Psychotherapy: Toward Therapist-Less Treatments?

Each article is available in two versions: one written by a human author and one generated entirely by AI. Click on the version you want to read and experience how far AI-generated writing has evolved.

A year ago, Veronica and I published an article asking whether we should fear artificial intelligence from different perspectives: what is the nature of what is commonly called AI, could it develop sentience and personality, and could it replace psychotherapy or psychotherapists? In the same issue of the Newsletter, Silvia also wrote, in this article, about how AI represented a paradigm-shifting event while underlining its lack of agency.

Using Kernberg’s theory of personality, we argued that AI does not have the necessary elements to develop consciousness or personality, but only a superficial simulation. Concerning psychotherapy, our final conclusion was that, while some data suggest that AI can achieve excellent adherence to cognitive-behavioral models of psychotherapy, it does not have the necessary elements to offer an efficient alternative to psychodynamic treatments like Transference-Focused Psychotherapy. TFP relies heavily on many elements of the therapist’s personality, notably the capacity to activate affective states in reaction to the patient’s affective activation, observe behavioral patterns (both in patient and therapist), and develop organized cognitive object representations of these affective, behavioral, and fantasy-symbolic phenomena into relational dyads.

Last October, during the TFP supervisor conference in Amsterdam, a fellow member, Michal Novák, took me aside during the conference dinner to express his conviction that, for a certain population of users, AI could function as a sufficient relational object – one whose responsiveness allows it to receive projections and to be experienced as relationship. It was interesting to hear a perspective different from our own. Since it was neither the place nor the time to engage in a lengthy discussion about the matter, we regretfully did not deepen the discussion. Nevertheless, his perspective left me wondering.

The next time I heard from him was when he emailed me to say that he had written an article for the website of the American Psychoanalytic Association in response to what we had written.

That was the main trigger for this Newsletter entirely dedicated to AI. We thought it would set the context for a very interesting discussion that could benefit our whole community.