Generative artificial intelligence is increasingly reshaping qualitative research practices and the epistemic conditions under which interpretation, sense-making, and methodological judgment are produced. While recent methodological research has examined AI-supported qualitative analysis, considerably less is known about how researchers themselves experience, appropriate, and critically reflect on the use of generative language models within their own empirical work. This poster presents findings from a higher education study that investigates how novice researchers integrate AI tools into qualitative empirical research processes and how they negotiate questions of interpretative responsibility and methodological agency.

The poster contributes to methodological discussions in qualitative educational research by empirically illuminating how generative AI is appropriated, reflected upon, and bounded in research practice. It discusses implications for the further development of qualitative research methodology under conditions of increasing automation, with particular attention to reflexivity, transparency, and epistemic responsibility in processes of knowing and acting.

The study draws on empirical material from master’s students conducting qualitative research projects, including interview-based studies and qualitative content analysis. Within a clearly defined methodological and ethical framework, participants were encouraged to experiment with generative AI tools during selected phases of their research. Data consist of reflective written accounts and group discussions and were analyzed using qualitative methods. Conceptually, the study builds on recent work in human–computer interaction and qualitative methodology that highlights the ambivalent role of large language models: while they can support efficiency, coordination, and analytic scaffolding, they also introduce risks related to interpretative narrowing, automation bias, and premature analytic closure (Gao et al., 2023; Xiao et al., 2023).

The findings indicate that participants primarily experience AI as a supportive resource in specific methodological tasks, such as data handling, deductive coding, and the clarification of analytic procedures. These experiences resonate with empirical studies demonstrating that large language models can meaningfully assist deductive qualitative coding, while still requiring continuous human judgment and methodological oversight (Xiao et al., 2023). At the same time, the reflections reveal a pronounced awareness of the epistemic limits of AI-supported analysis. Participants emphasize the necessity of maintaining human interpretative authority, critically interrogating AI-generated outputs, and actively positioning themselves as responsible agents in the analytic process. In this respect, their reflections align with hybrid approaches to qualitative interpretation that conceptualize AI as a dialogical methodological partner rather than an autonomous analytic authority (Krähnke et al., 2025).

Gao, J., Choo, K. T. W., Cao, J., Lee, R. K.-W., & Perrault, S. (2023). CoAIcoder: Examining the effectiveness of AI-assisted human-to-human collaboration in qualitative analysis. ACM Transactions on Computer-Human Interaction 31.1. 1-38. https://doi.org/10.48550/arXiv.2304.05560 Krähnke, U., Pehl, T., & Dresing, T. (2025). Hybride Interpretation textbasierter Daten mit dialogisch integrierten LLMs: Zur Nutzung generativer KI in der qualitativen Forschung. Preprint. https://nbn-resolving.org/urn:nbn:de:0168-ssoar-99389-7 Xiao, Z., Yuan, X., Liao, Q. V., Abdelghani, R., & Oudeyer, P.-Y. (2023). Supporting qualitative analysis with large language models: Combining codebook with GPT-3 for deductive coding. In Proceedings of the 28th International Conference on Intelligent User Interfaces (IUI ’23 Companion). ACM. 75-78. https://doi.org/10.1145/3581754.3584136