Discipline-Driven AI: Training a GPT to Qualitatively Code for Technical and Professional Communication Research

This project reports on the process used to train an AI model for qualitative coding and lessons learned for AI utilization in discipline-specific qualitative research. Interdisciplinary scholars are assessing the role of AI in qualitative coding processes, analyzing tools such as Cody [1], PaTAT [2] and ChatGPT3 [3]. Early findings suggest that emerging AI technology may contribute to qualitative coding by providing new insights [2] and increasing intercoder reliability [1], among other benefits. Though important, the sustained attention to the outcomes of AI-assisted coding provides little insight into the processes, successes, and failures during the training process. By focusing on process, researchers can better formulate strategies for AI-assisted coding. This project highlights researchers’ experiences training an AI with a disciplinary-specific knowledge base to support qualitative coding.