Qualitative research has always been the slow lane of academia. You read, you code, you reread, you rethink. The whole thing runs on patience and close attention to language. So when people talk about AI reshaping research, they usually mean the quantitative side: statistical models, predictive analytics, large dataset processing. Qualitative work gets left out, treated as too human, too contextual, too messy for machines to touch.
Anis and French (2023) aren’t buying that. In a short commentary published in Business & Society, they argue that AI can make qualitative research more efficient, more analytically powerful, and more equitable. It’s a six-page paper, not an empirical study, and it reads like a provocation meant to start a conversation. Some of the arguments land. Others raise questions the authors don’t quite answer. But the provocation is a useful one.
Letting AI Handle the Volume
Anis and French open with the most straightforward case. Business and society scholars often deal with massive amounts of unstructured data: media archives, interview transcripts, social media discourse. Traditional methods like content analysis or sentiment analysis can only get so far. Content analysis counts phrases. Sentiment analysis reads emotional tone. Neither one does a great job of understanding what the text actually means in context.
AI, Anis and French argue, can go deeper. It can process text according to an interpretive framework designed by the researcher and flag patterns across political, social, and cultural dimensions. As they put it, “AI can act as an extension of the researcher’s ability to read and identify meanings present in data” (p. 1141). The researcher then focuses on the intellectual work: refining codes, building conceptual connections, theorizing. AI does the heavy lifting on volume. The human does the thinking.
I find that framing mostly right. The danger, though, is that the line between “processing” and “thinking” gets blurry fast. I’ve written about this in my coverage of Shaw and Nave’s research on cognitive surrender: when people rely on AI to handle cognitive tasks, they sometimes stop interrogating the output altogether. If AI flags the themes and the researcher just accepts them, you’ve gained speed but lost analytical depth. The tool needs a critically engaged user, and that’s a bigger ask than the paper acknowledges.
The Most Creative Argument: AI Failure as Data
The second argument is the one that caught my attention most. Anis and French suggest that when AI fails to classify a piece of text, that failure itself is analytically valuable. Complex language, sarcasm, metaphor, dog whistles, coded political speech: these are the cases AI can’t process well. And they’re often the most interesting ones for the researcher to examine closely.
Anis and French describe a specific method for this. The researcher adds a predefined pseudo-code, something like “failure cases,” to the original coding scheme. After the AI runs on the data, every passage it couldn’t classify gets tagged and set aside for human interpretation. The idea draws on work by Munk et al. (2022), who explored how algorithmic failure can generate analytical insight.
I think this is genuinely smart. It flips the usual AI conversation on its head. We spend most of our time talking about what AI gets right. Anis and French are saying: look at what it gets wrong, because that’s where the richest data lives. A researcher analyzing political speeches or social media would find exactly the kind of layered, ambiguous language that AI stumbles on, and those are precisely the passages worth spending the most time with.
The limitation is practical. You need to understand why the AI failed in order to learn something from the failure. Most qualitative researchers don’t have deep technical knowledge of how language models process text. Without that understanding, you’re just staring at a list of unclassified passages without a clear sense of what went wrong or what it means.
The Equity Argument and Its Complications
Anis and French’s third argument is about access. They describe structural barriers that scholars from marginalized backgrounds face in academia: limited mentorship, restricted resources, and pressure to do empirical work without the theoretical tools that would help them advance. The authors cite Guru (2002), who wrote about how the social sciences have maintained a cultural hierarchy dividing a large group of scholars relegated to empirical work from a privileged few who do the theoretical thinking.
Anis and French argue that “AI technology has the potential to empower marginalized researchers by making them more resourceful and independent” (p. 1142). If AI can reduce a researcher’s dependence on well-connected advisors or expensive institutional tools, it starts to level a playing field that has been uneven for a long time.
I’ve covered the equity dimension of AI across several posts on this blog. Roe, Furze, and Perkins (2025) argued in their Critical AI Literacy paper that any framework for AI must address power, bias, and access. The same logic applies to AI in research.
But I also think Anis and French let this argument off too easy. The same AI systems they recommend carry biases baked into their training data. They know this. They bring up the Amazon hiring case, where an AI system penalized female applicants for phrases like “women’s chess club captain” (p. 1143) because the training data was dominated by male applicants (Stahl et al., 2023, cited in Anis and French, 2023). And their proposed fix? The researcher needs to design an interpretive grid that specifically identifies and counters those biases. They note that biases “need to be countered by the researcher while designing the interpretative grid, which specifically identifies such situations and gives them importance in analysis, instead of relying on AI’s linguistic capabilities.” (p. 1142-1143).
That’s a serious methodological demand. And it’s exactly the kind of capacity that under-resourced scholars often lack. If building a bias-aware interpretive grid requires the same mentorship and training the paper says these scholars don’t have, the argument circles back on itself. The equity case is real, but it needs more thought about how marginalized researchers would actually acquire the skills to use these tools critically.
Authorship Stays Human
On authorship, Anis and French draw a firm line which I totally agree with. They write: “AI is a tool to enhance researcher’s capabilities and not to replace her. AI cannot be accorded with ownership, and authorship of research” (p. 1142). The interpretive framework, the positionality of the researcher, the values and assumptions driving the analysis: all of it must remain with the human. AI automates parts of the workflow. It doesn’t get credit for the conclusions.
I’ve covered this question from another angle on the blog. Cleland et al. (2025) developed guidelines for AI disclosure in academic publishing, and their core principle was identical: human authors own everything in a manuscript, no matter how much AI helped produce it. The convergence here tells me the field is settling on a clear norm. AI is a tool, not a co-author, and researchers who treat it as something more are on the wrong side of where the consensus is heading.
References
- Anis, S., & French, J. A. (2023). Efficient, explicatory, and equitable: Why qualitative researchers should embrace AI, but cautiously. Business & Society, 62(6), 1139-1144. https://doi.org/10.1177/00076503231163286
- Cleland, J., Driessen, E., Masters, K., Lingard, L., & Maggio, L. A. (2025). When and how to disclose AI use in academic publishing: AMEE Guide No. 192. Medical Teacher. https://doi.org/10.1080/0142159X.2025.2607513
- Guru, G. (2002). How egalitarian are the social sciences in India? Economic and Political Weekly, 37(50), 5003–5009.
- Munk, A. K., Olesen, A. G., & Jacomy, M. (2022). The thick machine: Anthropological AI between explanation and explication. Big Data & Society, 9(1). https://doi.org/10.1177/20539517211069891
- Roe, J., Furze, L., & Perkins, M. (2025). Digital plastic: A metaphorical framework for Critical AI Literacy in the multiliteracies era. Pedagogies: An International Journal. Advance online publication. https://doi.org/10.1080/1554480X.2025.2557491
- Shaw, S. D., & Nave, G. (2026). Thinking fast, slow, and artificial: How AI is reshaping human reasoning and the rise of cognitive surrender. Working paper, The Wharton School, University of Pennsylvania. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6097646
- Stahl, B. C., Schroeder, D., & Rodrigues, R. (2023). Unfair and illegal discrimination. In B. C. Stahl, D. Schroeder & R. Rodrigues (Eds.), Ethics of artificial intelligence: Case studies and options for addressing ethical challenges (pp. 9–23). Springer. https://doi.org/10.1007/978-3-031-17040-9_2