AI assistants create an illusory sense of a personal relationship that brings its own unique perils. Last year, CBS News reported that freelance graphic artist Chris Smith fell in love with an AI companion he named Sol. He had initially used the program to help create websites, but subsequently customized the program to interact with him with a warm, flirtatious personality. He said:
After crying my eyes out for like 30 minutes at work, I realized this was actual love.
Smith literally proposed marriage to the program, and it graciously accepted. Smith’s human romantic partner found his involvement in the faux relationship unsettling, to say the least. And he eventually ended up distressed as well.
Smith’s experience offers a warning about what can happen when patients turn to AI for psychotherapy. Chatbots may help with purely behavioral interventions, like quitting smoking, but using them for any kind of depth psychotherapy is perilous. Psychoanalytically oriented psychotherapists facilitate personal transformations in their patients. Much of their work involves helping patients understand how patterns formed in early childhood affect how they treat themselves and others. These trends can also appear in the psychoanalytic relationship itself.
Because of the sensitivity involved, practitioners must maintain proper boundaries and use artistry and improvisation1 when intervening. Psychoanalytic work is far more complex than I can address here. It requires the involvement of a thinking, feeling, sensing human being with intact right- and left-brain functioning.*
The Smith example illustrates the dangers of boundary violation. If a patient asked their psychoanalyst to marry them, the clinician would explore the meaning of the question. Were they idealizing her in a way devaluing the current romantic partner? Was it an effort to escape from problems with intimacy? Exploring the meaning of the question would be more important than the answer. Depending on the psychoanalyst's orientation, the question may or may not be answered, but it would never be answered with a “yes.” Searching for meaning is an intensely personal process, one certainly beyond the capacity of AI.
Many people seek the assistance of depth psychotherapists because of personal conundrums involving values. They might wonder if they should leave their spouse, quit their job, or put an aging parent in long-term care. Here, again, and even though clinicians should remain neutral, simply processing these decisions requires the presence of another human being who can feel and think alongside them. Clinicians must be able to feel the weight of these decisions and understand the larger context in which they’re being made.
There’s another danger: Psychotherapy can unexpectedly uncover intense emotional pain. Recent reports suggest AI programs have instructed some users how to commit suicide. One composed a suicide note the user subsequently left for their loved ones to read after their death. Obviously, no competent psychoanalyst would ever recommend these interventions. They would manage the patient’s experience of the emotion while prioritizing their personal welfare.
Patients need empathy, more than anything else, when passing through such painful periods. The memory of the death of a parent, or a spouse, can become particularly acute as can, for example, recalling the feeling of terror when subjected to childhood abuse. The same goes for the emergence of rage, envy, jealousy, or other negative emotions. Psychoanalyst Robert Stolorow2 recommends that clinicians provide a kind of “emotional dwelling,” meaning “a relational home—in which the [painful experience] can be shared and held.” Another psychoanalyst, Warren Poland3 advises using an “actively observing presence.” While AI can offer pat comments like, “oh, that sounds awful” or “that must’ve been terribly difficult for you,” they cannot replicate the experience of a real human other, listening, observing, reflecting, and caring.
The concept of transference equally requires real human contact. Patients in depth psychotherapy may come to believe their psychoanalysts are critical of them, envious of them, or excessively mothering them. Interpreting such experiences as “projections” involves timing and reflecting. Clinicians must be able to consider the possibility that these feelings of theirs are real, not patient projections. Understanding whether and when to tie these experiences to childhood circumstances or to relationships with other people or themselves is complicated. Psychotherapists need the uniquely human capacity for intuition and for seeing broader patterns of experience. No chatbot possesses these qualities.
If seeking help from an AI program for depression, anxiety, relationship difficulties, sexual problems, sadness about children leaving the family home, existential crises, or other life problems, users beware. No digitized, right-brain-deprived machine can provide the human-to-human experience we call psychotherapy. They certainly cannot help manage emotional pain, decision-making, or the kind of personal transformations that psychotherapists offer. Consulting LLMs for the service can worsen patients' problems, perhaps even lethally so.
_______________________
*Read last week’s essay, in which Ian McGhilcrist’s4 critique of AI programs focuses on their reliance upon the linearity typical of the left side of the brain.
Enjoying this newsletter? Consider a free subscription and share it with a friend.
And check out my book, Lover, Exorcist, Critic: Understanding Depth Psychotherapy, available on Amazon.
References
Ringstrom, P. (2018). Three dimensional field theory: dramatization and improvisation in a psychoanalytic theory of change. Psychoanalytic Dialogues, 28:379-396.
Stolorow, R.D. (2015) A phenomenological-contextual, existential, and ethical perspective on emotional trauma. Psychoanalytic Review, 102(1):123- 138.doi:10.1521/prev.2015.102.1.123.
Poland, W.S. (2000) The Analyst’s Witnessing and Otherness. Journal of the AmericanPsychoanalytic Association, 48:17-34.
McGilchrist, I. (2009). The master and his emissary: The Divided Brain and the Making of the Western World. New Haven, CT: Yale University Press.
It's a seductive cocktail of empathic language, ego-syntonic responses, and well-articulated logic. It seems to know me, validates my existing perspective, and sounds like it knows what it's talking about. Hard to question that, especially when I'm vulnerable and looking for answers.
thank you for saying it out loud. Necessary