Attitudes, fears and experiences of medical scientists towards using AI tools in their work routine

Abstract Background Artificial intelligence (AI) has experienced an explosive growth in the healthcare sector and the scientific field. Large Language Models (LLMs) are investigated as potential tool to be used in medical treatment, by e.g. producing doctor’s reports, supporting personalized medicine or educating patients. However, limitations are broadly discussed addressing patient safety and ethical concerns due to false and misleading content. Therefore, LLMs face criticism among users not only towards their integration in the patient-provider-interaction, but also in its use in medical research and health care education. In the ongoing EXPOLS (Exploring the Potential of LLMs in Statistical Consulting) study, we address the question ‘What are the attitudes, fears and experiences regarding the integration of AI tools in the clinical/scientific work routine?’ Methods We implemented an online survey that will be accessible in 07/2024. All employees at the University and Medical Center Freiburg will be asked to participate in the survey via online media. We expect a sample size of n = 600-1,200. The survey combines published measurements (e.g. TRI 2.0, result expectation measure), as well as a self-designed instrument with 15 items aiming to measure fears and attitudes towards AI. Latter was developed within a multidisciplinary team making use of ChatGPT as item generator. We will investigate its psychometric properties using exploratory factor analysis and describe distribution characteristics. Results At the time of the conference, the project will be completed and both psychometric insights and content-related results will be reported. The results could have important public health implications, since attitudes towards AI will steer its integration in the medical and clinical routine. Conclusions The EXPOLS study makes a current inventory among medical scientists in Freiburg following a modern approach in instrument development by including ChatGPT as collegial advisor. Key messages • This study explores attitudes, fears and experiences towards the integration of AI tools in the work routine from medical and scientific staff at the University of Freiburg. • The findings might be crucial in determining the integration of AI based tools in the medical and clinical work routine.