The Augmented Qualitative Researcher: Using Generative AI in Qualitative Text Analysis

Abstract

Large language models (LLMs) such as ChatGPT are fueling academic discussions about generative artificial intelligence's role in research. In this article, we view LLMs as technological tools to support qualitative research and we aim to contribute to the long-standing history of articles that provide qualitative researchers with recommendations that can help them incorporate technological developments in their research practice while maintaining the interpretive, hands-on, and human-led approach that characterizes qualitative research. We present two approaches to using ChatGPT and integrating it into the qualitative research pipeline. We outline four main benefits of incorporating generative AI in qualitative work: 1) streamlined and expedited research; 2) enhanced transparency in researchers' ideas and clearer concept and category development; 3) identification of additional categories and themes beyond researchers' findings; and 4) increased reliability as generative AI support or challenge researchers' coding and analysis. We offer a practical framework for iterative prompt construction and evaluation for thematic and index coding, along with considerations for further analytical queries. These guidelines assist qualitative sociologists in conducting comprehensive, systematic, transparent, and reproducible research using generative AI. We also recognize generative AI’s limitations and underscore the indispensable role of human researchers throughout the process.