Background
Technological advancements, including generative Artificial Intelligence (AI) like ChatGPT, are transforming education. AI has potential usefulness in nursing education, particularly in the development of test items that align with evolving learning objectives and competency frameworks.
Aim
This article examines the ability of generative AI to accurately answer faculty-developed nursing exam questions and explores its strengths and limitations in assessments of varying cognitive complexity.
Methods
Three nursing exams from two different undergraduate nursing programs were administered to the AI model, and responses were evaluated for accuracy, with patterns of performance reviewed across item types and difficulty levels.
Results
ChatGPT-4 demonstrated fair accuracy on faculty-developed test items but struggled with higher-order, complex, and multiple-response questions, highlighting limitations in its reasoning capabilities.
Conclusions
Results reveal differences in AI accuracy based on exam complexity, with implications for AI's role in test item development, assessment strategies, and curriculum design in nursing education. The findings suggest that generative AI may serve as a resource for educators to streamline item writing, enhance question rigor, and foster student engagement with challenging material.
扫码关注我们
求助内容:
应助结果提醒方式:
