Generative AI and the profession of genetic counseling

Abstract

The development of artificial intelligence (AI) including generative large language models (LLMs) and software like ChatGPT is likely to significantly influence existing workforces. Genetic counseling has been identified as a profession likely impacted by advancements of LLMs in natural language processing tasks. It is important therefore to understand LLMs before using them in practice. We provide an overview of LLMs and the strengths, biases, risks, and potential uses in genetic counseling. We discuss how these models show promise for supporting certain tasks in genetic healthcare (e.g., letter writing, triage, intake or follow-up, decision aids, chatbots, and simulations). However, any interaction between LLMs and clients or clients' confidential information raises significant ethical, regulatory, and privacy concerns that are yet to be addressed. While LLMs may excel in information processing and are making unprecedented strides with regard to communication, we highlight aspects of psychotherapeutic encounters that require human interaction. Although LLMs/chatbots can provide information relevant to genetic tests and can mimic empathy, we postulate that these interactions cannot adequately replace the personalized application of counseling theory, skills, knowledge, and decision-making provided by a human genetic counselor. We propose that LLMs show great potential for use in aspects of genetic counseling practice. A continued, strengthened philosophical focus on the counseling process and psychotherapeutic goals of practice will be an essential aspect of genetic counselors' roles in the era of AI-supported counseling. Ongoing attention to the deployment of AI in clinical contexts and the relational elements of care will help ensure quality care for clients.