Reflective use of AI in university Courses

Findings from a pretest at the end of the 2026 summer Semester

Authors

  • Tanja Mühling Technical University of Applied Sciences Würzburg-Schweinfurt, Faculty of Applied Social Sciences Author
  • Sven Warnke Technical University of Applied Sciences Würzburg-Schweinfurt, Faculty of Applied Social Sciences Author
  • Harald Bolsinger Technical University of Applied Sciences Würzburg-Schweinfurt, THWS Business School Author

DOI:

https://doi.org/10.83397/PAIHE.2026.03

Keywords:

generative AI, higher education teaching, student Survey, purposes of use, reflective use of AI, pretest

Abstract

This article reports on findings from the pretest of a student survey on the use of generative AI in courses where AI use is framed and reflected upon from a pedagogical perspective. At the end of the summer semester of 2026, 50 students from three degree programs at the Würzburg-Schweinfurt University of Applied Sciences (THWS) were surveyed. In these courses, many of them used AI applications beyond ChatGPT for the first time, particularly a locally operated language model (AnythingLLM), Claude, and NotebookLM. The most common use among students is to have content explained to them or to use AI to structure their own content; the majority use AI to generate texts for exams only occasionally, if at all. A principal component analysis distinguishes between a product-oriented, a reflective-dialogic, and an exploratory use of generative AI. About seven out of ten respondents report using AI more consciously and purposefully at the end of the semester than at the beginning; students with little prior experience rate their own development most highly in descriptive terms. Respondents predominantly describe their own use of AI as reflective, though they agree more strongly with the critical evaluation of AI results than with the conscious control of AI usage. In open-ended responses, respondents cited, above all, more targeted prompting, the critical examination of results, and the conscious decision not to use AI for certain tasks as key aspects of their AI-related learning progress over the course of the semester. Since the results are based on retrospective self-reports from a small, non-random sample, they should be understood as exploratory findings that help refine the survey instrument for the planned longitudinal study.

References

Bolsinger, H. J., Warnke, S., & Mühling, T. (2026). The practice of reflective teaching with AI: Future skills development with offline LLMs. Practice of Artificial Intelligence in Higher Education, 1(1), Artikel 2. https://doi.org/10.83397/PAIHE.2026.02

Dicks, A., Ehlert, M., Schulz, W., & Solga, H. (2026). KI-Nutzung am Arbeitsplatz: Neue Daten zeigen deutliche Bildungsunterschiede. WZB Mitteilungen, (193), 10–14. https://bibliothek.wzb.eu/artikel/2026/f-27779.pdf

Ehlers, U.-D., & Rauch, E. (2024). KI im Studium aus Studierendensicht: Nutzung, Fähigkeiten und Einstellungen Studierender zu KI. Ergebnisse aus Daten der DHBW-Studierenden-Panelstudie 2024. NextEducation; Duale Hochschule Baden-Württemberg. https://next-education.org/downloads/2024-11-13_Forschungsbericht_KI-Nutzungsverhalten_Studierender_der_DHBW.pdf

Horstmann, N., Hüsch, M., & Breiter, A. (2026). Künstliche Intelligenz im Studium – die Sicht von Studierenden im Wintersemester 2025/26 (CHE DatenCHECK 9/2026). CHE Centrum für Hochschulentwicklung. https://hochschuldaten.che.de/kuenstliche-intelligenz-im-studium-die-sicht-von-studierenden-im-wintersemester-2025-26/

Kuckartz, U., & Rädiker, S. (2022). Qualitative Inhaltsanalyse: Methoden, Praxis, Computerunterstützung (5. Aufl.). Beltz Juventa.

von Garrel, J., & Mayer, J. (2023). Artificial intelligence in studies—use of ChatGPT and AI-based tools among students in Germany. Humanities and Social Sciences Communications, 10, Artikel 799. https://doi.org/10.1057/s41599-023-02304-7

von Garrel, J., & Mayer, J. (2025). Künstliche Intelligenz im Studium: Eine quantitative Längsschnittstudie zur Nutzung KI-basierter Tools durch Studierende (2023 & 2025). Hochschule Darmstadt. https://doi.org/10.48444/h_docs-pub-533

Warnke, S., Bolsinger, H. J., & Mühling, T. (2026). Leistung ist nicht Lernen: Ein Minimalkonsens für die Gestaltung von Hochschullehre mit generativer KI. Practice of Artificial Intelligence in Higher Education, 1(1), Artikel 1. https://doi.org/10.83397/PAIHE.2026.01

Published

05.11.2026