Demonstration of Generative AI Assisted Inductive Coding: Exploring Sociotechnical Imaginaries for Qualitative Research

As Generative AI (GenAI) becomes increasingly embedded in research practices, its potential role in qualitative analysis remains under-explored and contested. This paper demonstrates a five-step human-in-the-loop framework for GenAI-assisted inductive coding, designed to harness GenAI’s capabilities while preserving the interpretive depth of qualitative inquiry. Drawing on the sociotechnical imaginaries (STIs) perspective, we argue that every use of GenAI enacts particular visions of how humans and machines should collaborate in the research workflow. Built on Locke et al.’s (2022) coding practices, our framework structures collaboration between human researchers and GenAI across five coding moments: initial sensemaking, segmentation, open coding, focused coding, and fit assessment. It provides both procedural guidance and design flexibility, enabling researchers to selectively delegate tasks to GenAI while maintaining interpretive control. Methodologically, it helps researchers manage large datasets, enhance transparency, and reduce coding bias. Conceptually, it makes the STI of GenAI in research visible by embedding points for human oversight, prompt design, and ethical reflection throughout the process. We offer this framework as both a practical tool and a boundary object that supports qualitative analysis while enabling researchers and institutions to co-construct more responsible and inclusive futures for GenAI in qualitative research.