audiotranskription Hintergrund

AI Research

Our Publications on AI

The use of generative AI in qualitative research cannot be assessed solely on the basis of the speed of its responses. What matters most is what comes after the assignment is given: reviewing the material, placing it in the context of the case, correcting incorrect suggestions, and documenting one’s own analytical decisions.

At audiotranskription, we explore these questions through our own experiments, workshop reports, and scholarly articles. We are interested not only in potential ways to streamline work, but also in the testing effort involved, methodological limitations, data protection, and the structural implications of commercial AI infrastructures.

On this page, we have compiled the resulting findings and publications.

Lecture Transcript: “Is AI Transforming Qualitative Research?!” "Between the Hope for Acceleration and the Human Process of Analysis (and Learning)"

The continuously updated lecture transcript, spanning about 30 pages, provides an overview of the use of AI throughout the qualitative research process.

The topics covered include automatic transcription, various usage strategies, hybrid interpreting, serendipity prompting, verification efforts, and privacy-conscious and local processing methods.

The script combines academic contributions with insights gained from our own pilot studies, the development of f4, and discussions with qualitative researchers in webinars and workshops.

Version History

The versions from October 2025, February 2026, and June 2026 show how our assessments have changed and become more nuanced over the course of just a few months.

This ensures that newly obtained results, changes in technical capabilities, and methodological refinements remain transparent. The older versions are therefore still available.

List of Publications

Here is a complete compilation of our publications on the use of generative AI in qualitative research.

The list includes methodological articles, workshop reports, teaching materials, and critical analyses. Published texts are linked directly. Articles without links have been submitted or are in preparation; the links will be added after publication.

Submitted & In Progress

  • Dresing, Thorsten (submitted): From “Can the LLM do that?” to “What do I actually want?” LLM Re-Analysis as an Opportunity to Elucidate Implicit Evaluation Practices. In: Balzer, Lars/Laupper, Ellen/Eicher, Véronique (eds.): Special Issue “Rethinking Evaluation.” Empirical Pedagogy, 2027.
  • Krähnke, Uwe/Dresing, Thorsten/Pehl, Thorsten (in preparation): Serendipity-Prompting: Large Language Models as Maieutic Catalysts in Qualitative Data Analysis.

Published

  • Pehl, Thorsten/Dresing, Thorsten/Krähnke, Uwe (2026): The hybrid interpretation group with dialogically integrated LLMs. On the Use of Generative AI for Interpretive Reconstructive Analysis. In: Friese, Susanne/Morgan, David L. (eds.): Qualitative Data Analysis with Artificial Intelligence: Theory, Methods, and Practice. Say. (The print edition will be published in August 2026.)
  • Pehl, Thorsten/Dresing, Thorsten/Krähnke, Uwe (2026): An Introduction to the Hybrid Interpretation of Text Data Using Multiple Large Language Models. In: Behrmann, Laura; Epp, André; Gras, Juliana; Nowak, Anna Christina; Panenka, Petra; Stamann, Christoph; Vock, Rubina; Weydmann, Nicole (eds.): Teaching Qualitative Research. Opladen: Barbara Budrich Publishing (UTB 6625), Chapter 6.8, pp. 287–292. Open Access: https://elibrary.utb.de/doi/epdf/10.36198/9783838566252
  • Dresing, Thorsten/Pehl, Thorsten/Krähnke, Uwe (2026): Data masking using pseudonymization and anonymization is not enough! Methodological, Ethical, and Data Protection Criteria for AI-Supported Processing of Qualitative Data. SSOAR Preprint. https://nbn-resolving.org/urn:nbn:de:0168-ssoar-108772-7
  • Dresing, Thorsten (2026): Analysis of Qualitative Data Supported by Local AI—Quality Assessment Through Retrospective Comparison. In: Balzer, Lars/Beywl, Wolfgang (eds.): *evaluiert*. 3rd, revised ed. ed. Bern: hep Verlag, pp. 268–269.
  • Dresing, Thorsten (2026): Is AI Transforming Qualitative Research?! Between the hope for acceleration and the human process of analysis (and learning). Transcript of a lecture at the University of Bremen, June 5, 2026. https://www.audiotranskription.de/ki-einsatz-in-der-qualitativen-forschung/
  • Evers, Jeanine/Pehl, Thorsten (2026): An Experiment with AI: Comparing Human Argumentation Analysis with an LLM. SSOAR. https://nbn-resolving.org/urn:nbn:de:0168-ssoar-109244-1
  • Evers, Jeanine/Pehl, Thorsten (2026): Thinking about Commercial AI and Its Consequences for Qualitative Research: Rule by Law versus Scientific Integrity. SSOAR. https://nbn-resolving.org/urn:nbn:de:0168-ssoar-109026-1
  • Krähnke, Uwe/Pehl, Thorsten/Dresing, Thorsten (2025): Hybrid Interpretation of Text-Based Data Using Dialogically Integrated LLMs: On the Use of Generative AI in Qualitative Research. SSOAR Preprint. https://nbn-resolving.org/urn:nbn:de:0168-ssoar-99389-7
  • Dresing, Thorsten/Krähnke, Uwe/Pehl, Thorsten (2025): Citing ChatGPT and Other AI-Generated Content in Academic Writing. A Practical Guide for Researchers. audiotranskription.de. https://www.audiotranskription.de/ki-richtig-zitieren
  • Dresing, Thorsten (2025): GDPR-Compliant Hybrid Interpretation. A hands-on report on local interpretation using four LLMs on a MacBook Pro. audiotranskription.de. https://audiotranskription.de/llm-lokal-und-dsgvo-konform-nutzen

The Underestimated Cost of Inspections [Workshop Report (Draft)]

AI outputs are generated in a matter of seconds. Their academic examination, on the other hand, may take considerably longer.

In qualitative research, verification often involves returning to the source material, contextualizing statements within the case, checking supposed evidence, and filtering out plausible but unsupported suggestions.

Based on its own test runs, this workshop report examines the relationship between generation time, testing effort, and the amount of revision required. It shows which work steps can save time and where quality control eats up a large portion of those savings.

Commercial AI and Its Implications for Qualitative Research [Paper, English]

The article by Evers and Pehl views commercial AI not merely as a tool, but as part of an economic, technical, and political infrastructure.

The discussion will focus on data capitalism, environmental costs, geopolitical dependencies, and potential impacts on scientific integrity and freedom of research. Thus, this article expands on the question “What can the system do?” by adding the equally relevant question: “Under what conditions does this performance arise?”

Reflection questions help researchers assess the use of commercial AI in light of the methodological and ethical requirements of their own project.