ChatGPT prompts for generating multiple-choice questions in medical education and evidence on their validity: a literature review

The article titled "ChatGPT Prompts for Generating Multiple-Choice Questions in Medical Education and Evidence on Their Validity: A Literature Review" explores the emerging role of ChatGPT in creating multiple-choice questions (MCQs) for medical education and critically examines the validity of these AI-generated questions. Given the increasing use of AI tools like ChatGPT in educational settings, the study addresses an urgent need to understand the application and effectiveness of such tools in generating assessment items, specifically MCQs.The authors conducted a rapid review of the literature, screening 1920 studies and identifying 23 relevant studies that focused on ChatGPT-generated MCQs. The review highlights the diversity of prompts used to generate these questions, noting that some prompts were aligned with recommended prompt engineering strategies, such as referencing specific exam styles or adopting certain personas. The study found that while some ChatGPT-generated questions were of comparable quality to human-written ones, there were significant variations in accuracy, difficulty, and discrimination levels. Notably, while ChatGPT significantly reduced the time required for question creation, the study emphasizes the importance of thoroughly reviewing AI-generated MCQs before using them in examinations.The review concludes that while ChatGPT has the potential to streamline the test development process, careful prompt design and rigorous review processes are necessary to ensure the validity and educational value of the generated MCQs. The article serves as a valuable resource for medical educators considering the integration of AI in assessment creation, providing insights into both the benefits and limitations of using ChatGPT for this purpose.