By Maria J. Palmeri, Ph.D., and Robert M. Gordon, Psy.D.
Artificial intelligence (AI) was not a formal or informal part of my graduate school training, yet AI is increasingly utilized in present research, clinical work, and healthcare. While it brings excitement and possibility, it also elicits feelings of uncertainty about ethics, biases, and overreliance in the field and beyond.
The term “artificial intelligence” was first used in 1955 at the Dartmouth Research Conference, organized by John McCarthy, the goal of which was to “proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it” (McCarthy et al., 1955). Current definitions of AI refer to the system or digital interface that is created to mimic human intelligence to perform tasks.
Since 1955, forms of AI have been integrated into the field of science and medicine, ranging from a Stanford’s research group’s computer system built on “if-then” technology, to recommending antibiotics for various infectious diseases in the 1970s, to the development of electronic health record tools to improve efficiency and decision support in the 1990s and early 2000s, and recently developed deep learning and generative models that we know today (e.g., recurrent neural networks, language learning models [LLMs] (Alnattah et al., 2025).
Benefits of AI in Rehabilitation
With the seismic boom of AI in the modern age, it is easy to forget that it expands beyond LLMs such as ChatGPT, Meta, and Grok. For example, computer or app-based cognitive training programs that utilize adaptive algorithms, track performance, and provide real-time feedback can be incorporated into cognitive rehabilitation, allowing for additional practice and consolidation outside of in-person sessions.
AI-based assessments and screening technology for cognitive and psychological disorders are being integrated into some healthcare systems for data collection. Furthermore, AI is often used to create disability-friendly technology (e.g., voice-recognition software, hearing assistance, predictive text, digital support tools).
LLMs in particular are being used for clinician benefit, such as streamlining note writing (models can be taught to construct notes/mimic writing style), aiding in writing code or cleaning data in research, building algorithms for searching and summarizing literature, and teaching patients to break down complicated tasks or problem-solve.
Navigating the Pros and Cons of AI
As I venture through my fellowship in rehabilitation psychology, attending numerous seminars on AI-focused education, encouraging us to take advantage of its utility. I find myself increasingly drawn to the novelty and potential benefits, but worry about its proclivity for reinforcing socio-cultural and medical biases of underrepresented individuals (Hansen & Kerkhoff, 2025). For example, LLMs were shown to produce images that narrowly overrepresent “disability” as physical (e.g., wheelchair users), show outdated assistive technology, or depict disabled individuals as sad/lonely (NYC Bar, 2025).
Other concerns include clinicians or students using unencrypted patient-sensitive information on “everyday GPTs,” thus making it AI’s “property,” or the overreliance of using AI to write clinical notes or come up with neuropsychological test batteries and treatment plans, contributing to the “critical thinking epidemic” (i.e., societal decline of analytical, independent thought formation and problem-solving skills). A colloquial phenomenon, it alludes to the idea that increased dependence on AI tools is suppressing educational gains and independent decision-making (i.e., “cognitive offloading”) (Jose et al., 2025; Tian & Zhang, 2025).
I currently provide cognitive rehabilitation with adults who have experienced Traumatic Brain Injury (TBI), concussion, stroke, and long COVID, who sometimes use OpenAI/ChatGPT to do the cognitive work for them. I try to balance these concerns by building the patient’s active collaboration in establishing treatment goals and the development of compensatory strategies that can be generalized to real-life situations.
Artificial Intelligence Essential Reads
Personal Reflections and Key Lessons
In reflecting upon my career goals from the perspective of what healthcare systems, supervisors, mentors, and colleagues encourage when it comes to AI, three key lessons stand out:
- What goes into an LLM remains there, so if you upload anything, including your CV, case information, or manuscript, you are essentially “giving it” permission to use that information to build further models.
- LLMs and other AI-based programs are designed to keep you needing it, and it “wants” to help you—it will literally make up information and package it in a way that sounds true.
- AI-based programs are not search engines and should not replace actual research and fact-checking.
Early career and training psychologists have the advantage of curated environments that allow us to equip ourselves with knowledge on the ethical use of AI, and it is our ethical duty to do so.
Tips to Navigate AI
Undoubtedly, AI can be a useful tool but it can also inadvertently lead both clinicians and patients down slippery slopes. Despite how we as a professional community feel about AI in psychology, it does not seem to be going away. I try to keep in mind some golden rules for incorporating any AI into patient care, clinical and administrative work, or research:
- The clinician is always needed to make thoughtful judgments and final decisions and proofread for accuracy.
- The clinician values knowing better first by equipping ourselves with as much information about AI as possible from multiple fields such as medicine, law, and education.
- It is critical to view and process information through a critical, cautious, skeptical, and ethically-grounded lens when implementing any form of AI into practice to mitigate unwanted outcomes and repercussions.
The use of AI into psychological practice highlights continuity of tradition and innovation: “Tradition…is best served by a framework that balances continuities with discontinuities, preservation with change, and gratitude with an openness to moving on” (Mitchell, 1993, p. 8).
Training Resources for Students and Clinicians
Lastly, I want to share some resources for AI usage that leaders in the field have expertly crafted:
Maria J. Palmeri, Ph.D., received her Ph.D. from Ferkauf Graduate School of Psychology, Yeshiva University. She completed her Predoctoral Internship at NYU Langone Health–Rusk Rehabilitation. She is currently a Postdoctoral Psychology Fellow in Clinical Rehabilitation Research & TBI through the Advanced Rehabilitation Research Training Program.
Robert M. Gordon, Psy.D., is a Clinical Associate Professor at NYU Grossman School of Medicine and former Director of Intern Training at NYU Langone Health-Rusk Rehabilitation. He is a member of the Medicine & Addictions workgroup (established by 14 divisions of the American Psychological Association) that sponsors this blog.
