Quality of Human Expert vs. Large Language Model Generated Multiple Choice Questions in the Field of Mechanical Ventilation

Abstract Background Mechanical ventilation (MV) is a critical competency in critical care training, yet standardized methods for assessing MV-related knowledge are lacking. Traditional multiple-choice question (MCQ) development is resource-intensive, and prior studies have suggested that generative AI tools could streamline question creation. However, the effectiveness and reliability of AI- generated MCQs remain unclear. This study evaluates whether MCQs generated by ChatGPT are non-inferior to human-expert (HE) created questions in terms of quality and relevance for MV education. Methods Three key MV topics were selected: Equation of Motion & Ohm’s Law, Tau & Auto PEEP, and Oxygenation. Fifteen learning objectives were used to generate 15 AI-written MCQs via a standardized prompt with ChatGPT o1 (model o1-preview-2024-09-12). A group of 31 faculty experts, all of whom regularly teach MV, evaluated both AI-generated and HE-generated MCQs. Each MCQ was assessed based on its alignment with learning objectives, accuracy, clarity, plausibility of distractors, and difficulty level. The faculty members were blinded to the provenance of the MCQ questions. The non-inferiority margin was predefined as 15% of the total possible score (−3.45). Results AI-generated MCQs were statistically non-inferior to expert-written MCQs (95% upper CI: [-1.15, ∞]). Additionally, respondents were unable to reliably differentiate AI-generated from HE-written MCQs (p = 0.32). Conclusion AI-generated MCQs using ChatGPT o1 are comparable in quality and difficulty to those written by human experts. Given the time and resource-intensive nature of human MCQ development, AI-assisted question generation may serve as an efficient and scalable alternative for medical education assessment, even in highly specialized domains such as mechanical ventilation.