scieee AI-readable full text Open interactive document viewer

Data Snack - Beyond ChatGPT: AI Systems for Qualitative Research

Fuchs, Nele

Abstract

Qualitative research enables the in-depth exploration of social phenomena and the complexity of human interactions. Non-numerical data is created – often in the form of interviews, observation protocols, or visual materials. However, both data preparation, such as transcribing audio files, and data analysis prove to be extraordinarily time-intensive. In the digital age, qualitative researchers with programming backgrounds are increasingly developing innovative software solutions to support their colleagues in these labor-intensive processes. The integration of AI-supported tools into qualitative research thereby leads to important controversial discussions within the research community. While epistemological debates about the impact of digital methods on the research process remain essential, this Data Snack focuses on concrete tools that are based on AI models or incorporate them for support. The aim of this session is to highlight possibilities that exist beyond ChatGPT. While the open-source nature of these tools allows for code review and modification, researchers remain responsible for implementing appropriate data protection measures, evaluating algorithmic bias, and maintaining methodological transparency in AI-assisted analysis workflows.

Full text

Get your coffee ready, we will start in a few minutes. WELCOME TO OUR Beyond ChatGPT: AI Systems for Qualitative Research Contact Nele Fuchs, M.A. Data Scientist for the Humanities at the DSC [email protected] LinkedIn 04.12.2025 AI Systems for Qualitative Research 7 ▪Qualitative research… o…involves more than just working with non-numerical data. o…generates hypotheses. o…is open to new and unexpected findings during the research process. o…often involves personal and sensitive data, for which the GDPR applies. Definition: Qualitative Research Designed by Freepik http://www.freepik.com/ 12 ▪The qualitative research community is having a significant debate about whether using 'AI' in research is justified. ▪Open letter, October 2025: "We reject the use of generative artificial intelligence for reflexive qualitative research“ (Jowsey et al., 2025) oGenAI as simulated intelligence is incapable of meaning making. oQualitative research should remain a distinctly human practice. oThe established manifold harms of GenAI, especially to the environment and workers in the Global South. ▪In summary, there are ethical, methodological, epistemic concerns. Controversial Debate about 'AI' AI Icon by Microsoft Power Point 13 To have a more nuanced discussion, we need to develop stronger technical AI literacy. Purpose of this talk 14 Deconstructing 'AI' “ 'AI' is not a technology but a set of imaginaries and promises.” (Jürgen Geuter, 2024) 17 Deconstructing 'AI' ▪Today's AI systems can be divided into two groups: oPattern recognition oPattern generation ('genAI') oLarge Language Model (LLM), e.g. GPT5 (ChatGPT), Claude Sonnet 4.5 (Claude) etc. oText-to-image model, e.g. Stable Diffusion ▪Two key terms: machine learning (ML); natural language processing (NLP) Unlocking smartphone with fingerprint or Face ID Spam filter in email inbox Autocorrection soun soon […] Icons by Microsoft Power Point 20 Deconstructing 'AI' AI system AI model E.g. a large language model. ➢Like the engine in a car. The application is built around one or more AI models. ➢Like the car body. The infrastructure on which it runs. High Performance Computing for large models. Own device for smaller model. ➢GPU recommended Icons by Microsoft Power Point 21 Deconstructing 'AI' AI system AI model E.g. a large language model. ➢Like the engine in a car. The application is built around one or more AI models. ➢Like the car body. The infrastructure on which it runs. High Performance Computing (HPC) ➢proprietary AI systems = no control Icons by Microsoft Power Point 31 Conclusion ➢There are no easy answers to complex ethical and technological questions. ➢The qualitative research community should develop a solid digital and AI literacy. ➢Collaboration between computer science and qualitative research is essential. ➢In order to support this collaboration, we need data competency centers. AI 32 AI meets Qualitative Methods An interdisciplinary network for research with AI AIQM | ➢Network of PhD candidates and postdoctoral researchers. ➢The goal is to build competencies in the (responsible) use of AI. ➢Regular digital get-togethers; every five to six weeks. New members are always welcome! More information 33 Jowsey, Tanisha and Braun, Virginia and Clarke, Victoria and Clarke, Victoria and Lupton, Deborah and Fine, Michelle (October 20, 2025). We reject the use of generative artificial intelligence for reflexive qualitative research. Available at SSRN: https://ssrn.com/abstract=5676462 or http://dx.doi.org/10.2139/ssrn.5676462 Geuter, Jürgen; House of Competence. (June 17, 2024). Jürgen Geuter (tante): Ethische Nutzung von KI. YouTube. https://www.youtube.com/watch?v=wnPBmDoMGBc Slides available: https://llm-literacy.de/wpcontent/uploads/2025/02/Ethische_Nutzung_von_KI.pdf References