West Virginia University Researchers Examine Role of ChatGPT-5 Pro in Psychiatry Education
Dr. Wanhong Zheng, a professor at the West Virginia University School of Medicine, is at the forefront of a new study examining how artificial intelligence is reshaping mental health care and education. As part of the Department of Behavioral Medicine and Psychiatry, Zheng recognizes that AI tools have become deeply embedded in daily life. This shift means that medical training must evolve to address how patients increasingly rely on AI for mental health support. To address this, a WVU research team tested ChatGPT-5 Pro, aiming to determine if the AI could generate training materials for psychiatry students. The initiative seeks to bridge a significant knowledge gap created by the growing use of AI in mental health care.
The team focused on how AI chatbots influence the training of future psychiatrists. Their research centered on four key mental health conditions: schizophrenia spectrum disorder, anxiety disorders, mood spectrum disorders, and major depression with psychosis. Using ChatGPT-5 Pro, the researchers created patient scenarios that include detailed symptoms, medical history, and realistic chatbot interactions. These vignettes provide trainees with opportunities to practice diagnostic reasoning and decision-making. Each scenario also includes multiple-choice questions with clear explanations to support learning objectives.
AI Generates Valuable Training Materials
Findings revealed that ChatGPT-5 Pro could produce high-quality, realistic training materials. The AI generated scenarios with strong language, accurate diagnostic details, and substantial educational value. Trainees found the examples valuable for developing their skills. However, the study also highlighted the limitations of relying solely on AI. Human oversight remains essential, particularly regarding patient safety and ethical considerations. Faculty must lead discussions on these AI-generated cases to ensure trainees address risks properly. It’s also crucial for future psychiatrists to understand how patients use AI chatbots and to apply best practices in their clinical work.
Gangqing “Michael” Hu, an associate professor and co-leader of the study, identified a critical concern. AI chatbots might inadvertently reinforce harmful beliefs in patients. For instance, if a chatbot validates a patient’s unusual or delusional thoughts, it could exacerbate those beliefs. Hu cited reports of patients developing delusions after prolonged chatbot interactions, though the direct role of AI in these outcomes remains unclear. He explained, “A chatbot can be warm and agreeable, but that can be risky. It can keep validating a patient’s unusual beliefs.” This finding underscores the need for careful design and use of AI tools in mental health care to avoid potential harm.
Complexities of AI Interactions
The study also demonstrated that chatbots like ChatGPT-5 Pro have extensive knowledge bases. However, managing how these tools interact with users is complex. Hu emphasized the importance for doctors to understand these interactions and assess whether AI is genuinely helping or worsening a patient’s condition. He added, “It’s important for the physicians to understand the patients’ use of AI tools as well.” This means that medical professionals must remain actively involved in overseeing AI use, rather than relying entirely on automated systems.
The research team evaluated the AI-generated scenarios across four domains: language quality, diagnostic accuracy, patient safety, and educational value. The AI performed well in most categories, though the team emphasized the need for human guidance. Zheng noted, “We want the cases to match teaching goals and objectives.” This ensures the scenarios reflect real-world challenges. The study, published in npj Digital Medicine, offers insights into integrating AI into medical education while maintaining ethical standards.
Prioritizing Patient Safety
Despite promising results, Zheng and his team stressed the importance of prioritizing safety. When patients who use chatbots experience severe depression, it is critical to assess for risks such as suicidal or homicidal ideation. Zheng said, “We still have concerns about safety.” He added that AI must be used alongside human expertise to ensure ethical and effective care for patients. This balance is essential for protecting patients and maintaining the integrity of medical practice.
The research team believes that AI tools can play a valuable role in psychiatric training. However, their use must be handled with care. Zheng and his colleagues advocate for a balanced approach in which AI contributes to education while human expertise and ethical considerations remain central. Their work highlights the potential of AI as a training aid in psychiatry, but only when used responsibly and in collaboration with expert medical professionals.