The making of Qualitative Data Analysis with AI

Our new book is out: Qualitative Data Analysis with AI: Theory, Methods, and Practice Edited by Susanne Friese and David L. Morgan

Together with David Morgan, I have edited a new volume, Qualitative Data Analysis with AI: Theory, Methods, and Practice

sagepub.com/shop/buy-a-book/qualitative-data-… , published by SAGE. The book is now listed on the publisher's website, and after many months of working with our contributors, it is rather satisfying to see it become a real book. The idea originated at the International Congress of Qualitative Inquiry in 2024. David raised the possibility of putting together an edited volume on AI in qualitative research, and we began discussing what such a book should look like.

We knew we did not want another collection of general statements about how AI will revolutionise research, nor a book devoted primarily to warning researchers about its dangers. There is already plenty of both. What interested us was something more concrete:

What happens when experienced qualitative researchers begin working with generative AI? How do they adapt their methods? What works, what doesn't, and what do they learn along the way?

We invited researchers from different methodological traditions, with different experiences of AI, to contribute chapters grounded in their own analytical work. The result is a collection of eleven chapters that do not all arrive at the same conclusions. And I think that is one of the book's strengths.

Thirty years after the first CAQDAS debates There is also a historical connection that I find particularly meaningful. In 1995, Udo Kelle edited Computer-Aided Qualitative Data Analysis: Theory, Methods and Practice, also published by SAGE. At that time, qualitative researchers were debating whether computers had any legitimate place in interpretive research. Would software support analysis, or would it mechanise a process that depended on human understanding?

I remember those debates well. They accompanied much of my early professional life in qualitative research software.

Thirty years later, we find ourselves asking some remarkably similar questions. But the technology has changed fundamentally. Traditional CAQDAS software helps researchers organise, retrieve, compare, and systematically work with their material. Generative AI can produce summaries, suggest interpretations, recognise semantic relationships, and engage with researchers in natural language.

That opens possibilities we did not have before. It also creates new methodological problems. And it means we cannot simply transfer our established software practices to AI and expect everything to work as before.

Why AI literacy matters One of the central arguments David and I develop in the introduction is that qualitative researchers need AI literacy. I do not mean that everyone needs to become a programmer or understand the mathematics behind neural networks. But we do need to understand enough about the technology to make informed methodological decisions.

For example, what does it mean when a chatbot claims to have analysed forty interview transcripts? Did it really have access to all the material? How does it handle a large dataset? Can we trust the quotations it provides? And what about the frequency counts that appear so confidently in its answers?

Researchers also need to understand why identical prompts may produce different responses, how language models can reinforce our assumptions, and why an apparently convincing interpretation may have little empirical grounding. These are not merely technical details. They affect the credibility of our findings.

In the introduction, we make the argument that AI literacy is an ethical competence. Responsible AI use is not achieved simply by following a checklist or declaring that a human remained involved. It depends on understanding the tools sufficiently well to evaluate their outputs, protect our data, challenge their suggestions, and remain accountable for our interpretations.

I sometimes compare this to learning to drive. You do not need to know how to build an engine, but you do need to understand how the vehicle behaves, its limitations, and the rules of the road. The same applies to generative AI in qualitative research.

What you will find in the book The eleven chapters approach the subject from quite different directions. Sharlene Hesse-Biber introduces the idea of the AI-cyborg researcher, bringing feminist praxis and human-centred inquiry into the discussion. Christina Silver revisits the Five-Level QDA method and asks how we should teach and learn qualitative analysis in the generative AI era.

Stefan Rädiker and Udo Kuckartz examine how AI is being integrated into MAXQDA, while Jessica Parker, Veronica Richard, and I report on an experiment testing whether consumer chatbots can reliably code open-ended survey responses. The results highlight substantial problems with consistency and auditability.

In my own chapter, Beyond Coding: Conversational AI for Qualitative Analysis with QInsights, I explore a different direction: moving away from coding as the organising principle of analysis and towards an iterative dialogue with the data. I introduce my CA^AI framework and show how this approach can work in practice.

Jonas Wibowo and Hendrik Wiese also describe their experiences with QInsights in an actual research project, including the methodological adjustments they made when their initial approach did not work as expected.

Other chapters explore hybrid interpretation with multiple language models, the use of AI in grounded theory, and modular prompting within the documentary method. Jessica Lester and Trena Paulus introduce the MERIT framework for responsible methodological innovation, and David Morgan closes the book by examining how innovations spread and what this might mean for the future adoption of AI in qualitative research.

What I particularly appreciate about these contributions is that they go beyond abstract arguments. The authors describe concrete methods, experiments, difficulties, and decisions. Several chapters show how an initial approach had to be revised because the technology did not behave as expected. This is precisely the kind of methodological learning our field needs.

Steering rather than surrendering If there is one idea that runs through our introduction and connects many of the contributions, it is that the role of the researcher is not disappearing. But it is changing. Generative AI can help us locate relevant material, explore comparisons, challenge initial interpretations, and consider alternative explanations. It can also produce superficial summaries, invent evidence, and confirm what we already believe. The difference often lies in how we work with it.

I increasingly see productive AI-supported analysis as a process of steering rather than delegating. We provide the analytical direction, question the outputs, return to the data, identify contradictions, and decide what an interpretation means in the context of our study.

This is not one-click analysis. Nor is it simply doing traditional coding faster. It requires us to rethink some of our analytical practices while preserving the methodological principles that make qualitative research valuable. And that, ultimately, is what this book is about.

A word of thanks Editing a volume like this is a substantial undertaking, particularly in a field where the technology changes so quickly. I am grateful to David for initiating the project and for our collaboration, and to all our contributors for their willingness to experiment, reflect critically, and share what they learned.

The book does not claim to provide the final word on AI in qualitative research. I doubt such a final word is even possible at this stage. What I hope it offers is a useful account of where we are now, what we are learning, and how we might move forward with methodological curiosity and critical judgment.

Qualitative Data Analysis with AI: Theory, Methods, and Practice is published by SAGE. See the book, chapter contents, and further details on the SAGE website

sagepub.com/shop/buy-a-book/qualitative-data-… .

I hope it will contribute to more informed discussions about AI in qualitative research — discussions grounded not only in what we believe these technologies might do, but in what happens when we put them to work.