Comparing Human Insight and AI in Thematic Analysis of Nursing Education Reflections

Background A limitation in the advancement of interprofessional education is the large number of resources needed for student learning evaluation. Reflective assignments are used in interprofessional education for students to reexamine their learning while reflecting on past professional practice and their aims to incorporate interprofessional collaboration into future practice. However, analyzing the meaning of these reflections is time consuming. Method Using a qualitative descriptive design, this study reviewed the outputs of traditional independent coding and theming by humans versus computer-generated coding and theming by a large language model to evaluate interprofessional reflections. Results Study results indicate that currently, outputs of large language models are not identical to a human team for reflexive thematic qualitative analysis. Conclusion Large language models provide one method to efficiently review large data sets. However, there are limitations to these models. Large language models should be used in conjunction with other evaluation methods.