This article is part of a Special Section of On Board with Professional Psychology that focuses on the intersection of professional psychology and Artificial Intelligence (AI). Learn more about ABPP’s Artificial Intelligence Task Force.
Psychologists are increasingly apprehensive about the rapid advancements in artificial intelligence and its potential implications for the practice of psychology. Many express concerns about how AI, particularly Large Language Models (LLMs), might impact the therapeutic process and reshape our profession. Despite these technological strides, evolution has shaped human behavior, cognition, and emotion in ways inextricably linked to our biological and cultural heritage. These evolutionary forces have favored traits and skills particularly relevant to psychotherapy. While current versions of LLMs have made significant progress in simulating human-like responses and processing vast amounts of data, our evolved social and emotional capacities suggest that these technologies will not replace human psychologists in psychotherapy. This essay explores the evolutionary advantages of human psychology and the limitations of LLMs in replicating the complex, adaptive, and context-dependent nature of therapeutic interactions.
The Evolutionary Basis of Human Psychology
Human beings are the product of millions of years of evolutionary adaptation. Our cognitive faculties, emotional responses, and social behaviors have been shaped by natural selection to enhance survival and reproduction in diverse environments. Our abilities to form intricate social relationships, experience a range of emotions, and engage in abstract thinking are traits favored by evolution due to their adaptive advantages (Tooby & Cosmides, 1992). These evolutionary adaptations are not static; they are dynamically responsive to environmental cues and social contexts, allowing humans to navigate various situations with flexibility and creativity.
Psychotherapy is deeply rooted in understanding and addressing the complexities of human behavior and emotion that have evolved over time. Psychologists are trained to recognize and work with the challenging interplay of biological, psychological, and social factors influencing an individual’s mental health. We employ various techniques to help patients navigate their inner struggles and external dilemmas. The effectiveness of psychotherapy often depends on the psychologist’s ability to establish an empathetic, trusting relationship with the patient—a relationship that is inherently human and context-specific.
The Limitations of LLMs in Psychotherapy
LLMs, while impressive in their ability to process and generate language, are fundamentally limited by their lack of biological and evolutionary context. These systems are built on algorithms and data-driven models that replicate human-like responses, but they do not possess the biological underpinnings of human cognition and emotion. They lack the capacity for genuine understanding, empathy, and compassion, which are crucial components of effective psychotherapy.
One key principle of evolution is adaptation. Human behavior and cognition are beneficial adaptive responses to specific environmental pressures. Psychologists, through their training and experience, can discern meaningful patterns in a patient’s thoughts, feelings, and behaviors. They can help patients recognize how past experiences and human predispositions may influence their current mental states and emotional experiences. For example, a psychologist might help a patient understand how stress and anxiety, evolved responses to potential threats mediated through the limbic-hypothalamic-adrenal axis, may be maladaptive in our modern world (Klinlein et al., 2015). LLMs, lacking a biological and evolutionary framework, would struggle to provide such refined insights.
The therapeutic alliance, a cornerstone of successful psychotherapy, is rooted in trust, empathy, and mutual understanding—qualities that reflect our evolutionary social history. While LLMs can simulate empathic responses through linguistic patterns, they fundamentally lack the capacity for genuine emotional experience and understanding that humans possess. The empathy exhibited by LLMs remains a sophisticated approximation rather than authentic emotional resonance, as these systems operate on aggregated data rather than true emotional comprehension.
Crucially, LLMs are not conscious entities. Instead, they are complex statistical models trained on extensive textual data, devoid of the innate understanding of existence or self-awareness that defines human cognition and experience. This ontological difference creates a significant barrier in replicating the human connection essential to psychological healing.
Moreover, LLMs’ current limitations in developing a robust theory of mind—the ability to attribute mental states, beliefs, and intentions to others—further impedes their capacity to replicate the depth and quality of human interaction necessary for effective psychotherapy. The transformative power of psychotherapy emerges from an intricate interplay of empathy, shared human experience, and collaborative meaning-making that transcends mere algorithmic processes, remaining beyond the current capabilities of artificial intelligence.
The Adaptive Complexity of Human Behavior
Human behavior is characterized by its adaptive complexity, resulting from evolutionary processes that have favored flexibility and situation-specific decision-making. This complexity is evident in how humans navigate social relationships, manage emotions, and solve problems. Psychotherapy often involves helping clients address these complexities, which require an understanding of the broader evolutionary and cultural circumstances in which behavior occurs.
For example, consider the evolution of attachment styles, which have been shaped by the need for safety and security in early human environments (Levy, 2013). A psychologist might work with a patient to understand how their attachment style, developed in response to early caregiving experiences, influences their current relationships. This kind of deep, relational understanding is difficult to replicate in LLMs, which lack the evolutionary and experiential background necessary to fully grasp the significance of these patterns.
Additionally, the adaptive nature of human behavior means that what works in one circumstance may not work in another. Psychologists are adept at adjusting their approaches based on the unique needs, values, and circumstances of each patient. LLMs, while capable of processing large amounts of data, may struggle to adapt their responses in ways that are truly context-dependent and personalized. For instance, a patient with post-traumatic stress disorder (PTSD) might choose to focus on improving their sleep habits to be more present for their spouse rather than directly addressing trauma-related symptoms. The algorithms underlying these systems may not capture the fluidity and intricacies of patients’ clinical needs.
Practice Recommendations
LLMs can aid psychotherapy by generating treatment plans, documenting notes, and recommending materials. However, they are not developed or designed to provide psychotherapy services. These tools enhance efficiency and support clinical tasks, but therapy remains the psychologist’s domain. Integrating LLMs as resources allows psychologists to focus on the relational and interpretive aspects that require unique human insights.
LLMs can also enhance professional practice in psychology by assisting psychologists in self-reflection. Using LLMs as a tool, psychologists can generate prompts for self-reflection on countertransference reactions, therapy patterns, patient dynamics, boundary concerns, and self-care techniques. This systematic exploration of biases, emotional responses, and therapeutic approaches helps ensure patients feel genuinely heard and valued. Regular reflection and refinement of relational skills, especially in complex cases, are crucial for effective patient care.
Psychologists must carefully navigate ethical considerations when using LLMs as part of psychological practice. These include data privacy, algorithmic bias, and over-reliance on technology. Patient information must be protected, and LLMs must be used transparently and responsibly. Psychologists should critically evaluate LLM-generated insights and avoid letting them overshadow their professional judgment. By adhering to ethical guidelines and prioritizing patient well-being, psychologists can integrate technology into their practice ethically.
Conclusion
The principles of evolution have endowed humans with complex, adaptive, and situationally-responsive capacities central to the practice of psychotherapy. While LLMs represent significant advancements in technology, they are fundamentally limited by their lack of biological, psychological, and evolutionary grounding. Their ability to simulate the human experience through natural language processing can be impressive. However, the capacity to understand and address the adaptive complexities of human behavior, establish a genuine therapeutic alliance, and provide empathetic, nuanced care are qualities deeply rooted in human evolution and unlikely to be reproduced by LLMs.
References
Gavazzi, J. D. (2024). The Ethical Use of Artificial Intelligence in the Practice of Psychology. The Pennsylvania Psychologist, Volume 84(5), 30-31.
Horvath, A. O., & Symonds, B. D. (1991). Relation between working alliance and outcome in psychotherapy: A meta-analysis. Journal of Counseling Psychology, 38(2), 139–149. https://doi.org/10.1037/0022-0167.38.2.139
Kinlein, S. A., Wilson, C. D., & Karatsoreos, I. N. (2015). Dysregulated Hypothalamic-Pituitary-Adrenal axis function contributes to altered endocrine and neurobehavioral responses to acute stress. Frontiers in Psychiatry, 6. https://doi.org/10.3389/fpsyt.2015.00031
Levy, K. N. (2013). Introduction: Attachment Theory and Psychotherapy. Journal of Clinical Psychology, 69(11), 1133–1135. https://doi.org/10.1002/jclp.22040
Tooby, J., & Cosmides, L. (1992). The psychological foundations of culture. In J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture (pp. 19–136). Oxford University Press.
John D. Gavazzi, PsyD, ABPP
John D. Gavazzi, PsyD, ABPP
Board Certified in Clinical Psychology
Correspondence: john.gavazzi@gmail.com