Psychiatrists’ Experiences and Opinions of Generative AI: An Exploratory Online Mixed Methods Survey in Germany (Preprint)

Abstract

<sec> <title>BACKGROUND</title> The integration of generative artificial intelligence (genAI) in psychiatric practice has gained attention due to their potential to enhance clinical workflows. However, the extent of their use and the perceptions of psychiatrists regarding their effectiveness remain underexplored, particularly in the context of the German healthcare system. </sec> <sec> <title>OBJECTIVE</title> This study aims to assess the experiences and attitudes of psychiatrists in Berlin and Brandenburg towards the use of genAI-based chatbots in clinical practice, focusing on their applications, benefits, and concerns. </sec> <sec> <title>METHODS</title> An online mixed-methods survey was conducted from September 19, 2024 to March 14, 2025, targeting psychiatrists in public psychiatric hospitals, psychiatric departments, and outpatient practices in Berlin and Brandenburg as well as members of the German Association for Psychiatry, Psychotherapy, Psychosomatics and Neurology (DGPPN). A total of 11,754 psychiatrists were invited, with 126 completing the survey (response rate: 0.011%). The survey included sociodemographic questions and assessed prior experience with genAI, their perceived impact on clinical work, and specific applications in documentation, diagnostics, and communication. Descriptive statistics were used for quantitative analysis, while qualitative responses were analyzed through summarizing content analysis. </sec> <sec> <title>RESULTS</title> Approximately half of the respondents (52.0%) reported using AI chatbots, with ChatGPT being the most utilized tool (50.0%). Participants primarily employed AI for writing doctor's letters (22%) and obtaining medication information (25%). Key themes emerged regarding the potential of genAI to reduce bureaucratic burdens and improve documentation efficiency, although skepticism about their practical benefits was noted. Concerns were raised about the limitations of AI in diagnostic and therapeutic decision-making, with calls for targeted training for clinicians. Language support was highlighted as a significant advantage for non-native German speakers.While our findings provide timely insights, the sample size limits the generalizability of results and warrants further large-scale studies across diverse clinical contexts. </sec> <sec> <title>CONCLUSIONS</title> Psychiatrists expressed cautious optimism about the use of genAI, particularly for reducing administrative burdens, while voicing clear concerns about their limitations in diagnostics and the need for targeted training. These findings highlight the importance of a balanced, evidence-based approach to AI integration in psychiatry, especially in light of the EU AI Act. Further research is essential to understand the long-term clinical and ethical implications of genAI use in mental health care. </sec> <sec> <title>CLINICALTRIAL</title> This study was reviewed by the Ethics Committee of the Brandenburg Medical School, which issued a waiver of ethical approval (protocol no. 228072024-ANF). </sec>

Ähnliche Arbeiten