Evaluation of physicians’ opinions on receiving health information from ChatGPT; trustworthiness, value, and danger

This study assesses physicians' perspectives on the role of ChatGPT in providing health-related information, highlighting its reliability, contributions and potential risks. In a survey of 88 physicians who were selected using convenience sampling, ChatGPT's answers to three different health questions were evaluated, and reliability, benefits and potential hazards were rated on a scale of 1-10. The physicians were then asked about the rationale behind their evaluation scores. The qualitative assessments of the physicians were evaluated together with quantitative analyses of the scores. The results showed high trustworthiness scores across all questions (mean scores of 8.4, 8.2 and 8.6 over 10) and non-significant difference between questions, as well as significant benefits in promoting health literacy through general health advice and lifestyle recommendations (mean scores of 7.3, 7.9 and 8.3). Answer for lowering cholesterol was evaluated at at higher level of contribution than first question (p = 0.0004). However, concerns were raised about risks, particularly due to more direct answers including treatment recommendations, and the potential for misuse by patients was noted (mean scores of 2.6, 3.5 and 2.5). Second question (the treatment for atrial fibration) was evaluated at a higher level of risk (p = 0.002). While ChatGPT has the potential to be an effective tool for health education and promotion, this study underlines the importance of increasing users' health literacy and establishing guidelines to ensure safe and ethical use. The findings highlight the need for multi-stakeholder approaches to optimise the integration of ChatGPT into healthcare settings.