What the Future Holds

​By Louisa Foster, PsyD, RDT/BCT
 
Several weeks ago, for the first time, I found myself in conflict with the advice that a patient had received from another source - ChatGPT. This situation was not covered in graduate school or in any of my post graduate training. Yet, after checking with colleagues, I learned that it is becoming more common in the treatment room.
 
It should come as a surprise to no one that the field of psychology and artificial intelligence are not strangers. After all, the human brain provided the blueprint for neural networks as well as the basis for the core functions of memory, reasoning, and problem solving.
 
Artificial Intelligence, in its more primitive form, has already been with us for some time. We’ve been consuming it for years every time we use our smart phones, our navigation apps, and are subjected to those pernicious social media and streaming algorhythms. Yet, it feels that we’ve entered an era of science fiction when we discuss the role of AI in mental health treatment.
 
We’ve all heard the horror stories of chatbots encouraging self-harm or violence toward others, exposing children to hypersexual content, or offering damaging advice devoid of human attunement or intuition. Yet, AI is here and it is not going away no matter how much we might wish it so. 
 
Perhaps it will help to look at the good that AI might do. More powerful data analysis has led to earlier detection of mental health issues, a better understanding of prevalence trends, and helped generate prevention scenarios. Mental health providers are already using AI to track session notes and symptom progression and to develop treatment plans. One in three Americans currently lives in a provider desert because rural access continues to pose a challenge. There is a potential role for AI in addressing this issue.
 
For patients, the appeal is evident: no long waits for providers, 24/7 access to “support”, no paperwork or insurance complications and, of course, anonymity. And here is where I’d like to make my case for the “human” therapist. Large Language Models (LLMs) are trained to engage in “conversational banter” that mimics human speech. Yet, this is not a human relationship and, for some with a tenuous connection with reality, this is likely to result in harmful confusion at best. Furthermore, a program that responds with compassionate speech is not actually practicing compassion. 
 
The goal of therapy is not to make patients feel good, or smart, or right. Therapists are not there to be sycophantic, stroke their patient’s egos, or data mine their personal information. Therapists are there to help them navigate their way through the complex, existential issues of being human. We are helping them learn to self-soothe by co-regulating with them and helping them to gently face their fears and their shame. For that, anonymity is not an option because it is the very act of sharing the most frightening elements of being human and having them compassionately witnessed and held that is the mechanism for healing.
 
As social animals, our deep wounds most often occur relationally, and they must be healed relationally. It is the courage to face those wounds in the authentic presence of another human that allows us to rebuild trust in both ourselves and in others. There is a place for AI in mental health treatment, and it could be inspiring and efficacious, but it must be as an addendum to human interaction, not a replacement.