Friese, Nguyen-Trung, Powell & Morgan (2026) — Beyond Binary Positions

TL;DR: A direct, theoretically grounded response to the Jowsey et al. (2025) open letter, signed by 100+ scholars, arguing that the categorical rejection of GenAI in qualitative research rests on a philosophically contestable claim — that meaning-making is exclusively human — and that a binary accept/reject framing forecloses responsible, critically reflexive GenAI integration. The paper offers four theoretical frameworks to show that meaning is relational and distributed, not located in human minds alone.

Problem

jowsey-et-al-2025-we-reject was an unusually direct intervention: 419 signatories (including Virginia Braun and Victoria Clarke, the architects of reflexive TA) collectively rejecting GenAI for Big-Q reflexive qualitative research on three grounds: (1) GenAI cannot make meaning; (2) reflexive qualitative research must be distinctly human; (3) GenAI’s environmental and social justice costs are unacceptable. The letter presented these as settled conclusions, not hypotheses.

Friese et al. accept the seriousness of the concerns but contest both the epistemological foundations and the practical consequences of categorical rejection. Their target is not the Jowsey letter’s identification of risks — those are real — but its move from “these risks are real” to “therefore categorical prohibition.” That move, they argue, requires an unexamined philosophical premise about where meaning-making is located, and that premise is contestable.

Conflict of interest note: Stefanie Friese is co-founder of QInsights.ai; Jeffrey Powell is co-founder of Causal Map. The authors disclose this prominently. The co-authorship of David Morgan — a foundational qualitative methodologist with no commercial stake — and Khuong Nguyen-Trung serves partly to address the COI concern. Whether the COI undermines the philosophical argument is a question readers must judge independently.

Three Counter-Arguments

The paper is organized around Jowsey’s three reasons, offering a counter-argument to each.

Counter to Reason 1 (GenAI cannot make meaning): The claim that meaning-making is exclusively human is a philosophical position, not an empirical finding. Four theoretical traditions challenge it:

1. Assemblage thinking (Deleuze & Guattari): Meaning emerges from configurations of human and non-human elements; it is not located in individual human minds but in relational assemblages. AI tools are components of assemblages in which meaning can emerge.
2. Distributed cognition (Hutchins): Cognitive processes — including those that generate meaning — are distributed across people, artifacts, tools, and environments. The boundary between “inside” and “outside” human cognition is empirically porous, not philosophically fixed.
3. Posthumanist knowledge practices (Barad): Meaning is produced through intra-action — material-discursive entanglements — not through pre-existing human subjects encountering the world. AI tools are participants in these entanglements.
4. Sociomaterial entanglements (Orlikowski): In organizational and practice-based research, meaning is co-constituted by humans and technologies; clean separation is analytically impossible and practically false.

These are not fringe positions. They are core theoretical frameworks in qualitative, STS, and organizational scholarship. The Jowsey letter’s “exclusively human” premise is simply incompatible with them — or rather, it requires rejecting them, which the letter does not acknowledge.

Counter to Reason 2 (reflexive qualitative research must be distinctly human): Friese et al. point to the existing scholarship on responsible human-AI collaboration — friese-caai-framework-2026, costa-abductivai-2025, wise-et-al-2026-ai-not-the-enemy, nguyen-trung-gaita-2025, among others — as empirical demonstrations that GenAI can be integrated without displacing the researcher’s reflexive engagement. Whether these frameworks actually achieve what they claim is a methodological question to be evaluated empirically, not foreclosed by principle.

They also note a tension internal to the Jowsey coalition: Braun & Clarke have consistently described reflexive TA as flexible, with no single right way to proceed. Their signature on a letter declaring categorical prohibition appears to contradict this flexibility claim. Friese et al. do not dwell on this, but the tension is real and the authors flag it.

Counter to Reason 3 (environmental and social justice costs): The authors accept that environmental harms from large-scale AI use are real and serious. But they argue that proportionality matters: the appropriate response to unequal costs is governance, harm-reduction policies, and attention to who bears costs — not blanket prohibition. Prohibition doesn’t reduce harm; it removes qualitative researchers from conversations about how AI tools should be designed and governed. The political consequence of withdrawal is that qualitative epistemological values disappear from AI development.

This argument echoes de-paoli-reject-rejection-2026's point about the political cost of withdrawal, and greenhalgh-2026-beyond-the-binary's refusal to sign on similar grounds.

The “Making Space” Frame

The paper’s organizing metaphor — “making space for critical and reflexive GenAI integration” — is strategic. Rather than defending AI use generally, Friese et al. are carving out a protected zone: research that is both critical about AI and uses AI reflexively. This isn’t a defense of unreflective AI adoption or the “AI as efficiency tool” framing that Jowsey rightly criticizes. It is a claim that the category of responsible, epistemologically sophisticated AI integration exists and deserves methodological space.

The signatories (over 100 qualitative researchers) include Stefano De Paoli de-paoli-reject-rejection-2026, Margrit Schreier, Antony Bryant, and others — representing a range of qualitative traditions that are not unified by any commercial interest.

Epistemological Stance

Pluralist. The theoretical frameworks enlisted — assemblage thinking, distributed cognition, posthumanism, sociomateriality — don’t form a single unified epistemology. What they share is the rejection of human-AI dualism and the relocalization of meaning-making from individual human minds to relational processes. The paper’s epistemological bet is that the philosophical traditions most relevant to qualitative research are precisely the ones that make the “exclusively human” claim hardest to sustain.

Rigor and Trustworthiness

This is a theoretical/argumentative paper, not an empirical study. The authors marshal existing scholarship and theoretical resources rather than producing primary data. Rigor here is argumentative quality: the counter-arguments are grounded in canonical frameworks, the COI is disclosed rather than concealed, and the engagement with the Jowsey letter is point-by-point rather than rhetorical.

Limitations

• The COI is real and cannot be fully neutralized by disclosure. Friese’s commercial interest in QInsights.ai means readers should weigh the commercial and scholarly arguments separately.
• The four theoretical frameworks (assemblage, distributed cognition, posthumanism, sociomateriality) are all deployed relatively briefly — each would require a full paper to establish their implications for AI in qualitative research. The paper gestures toward theoretical resources rather than developing them.
• The paper doesn’t resolve the methodological question: how, specifically, do you verify that a researcher using AI has maintained genuine interpretive authority? The frameworks explain why “exclusively human” is philosophically contestable; they don’t provide the operational criteria for distinguishing responsible from irresponsible AI integration.
• Morgan’s presence as a senior qualitative methodologist without commercial stake strengthens the scholarly credibility, but his contributions to the argument are not individually attributable.

Connections

• jowsey-et-al-2025-we-reject — the direct target; reading these together is essential for understanding the debate.
• de-paoli-reject-rejection-2026 — parallel philosophical rebuttal; De Paoli uses ANT (Latour), Friese et al. use a broader set of theoretical frameworks including Latour’s intellectual descendants.
• greenhalgh-2026-beyond-the-binary — same refusal to sign, similar reframing; Greenhalgh focuses on epistemic authority rather than assemblage theory.
• friese-caai-framework-2026 — Friese’s own CAAI framework is cited as an example of responsible integration; the beyond-binary response shores up the epistemological legitimacy of CAAI.
• wise-et-al-2026-ai-not-the-enemy — cited as empirical demonstration that responsible integration is achievable.
• contested-claims — Claim 9 (categorical rejection as philosophical dogma); this paper significantly strengthens the counter-rejection side with theoretical depth.
• epistemology — the four theoretical frameworks (assemblage, distributed cognition, posthumanism, sociomateriality) represent important epistemological traditions that the wiki should map.
• human-ai-collaboration — the “making space” frame and the emphasis on researcher-retained reflexive authority are contributions to the collaboration frameworks section.
• ai-research-ethics — the environmental justice counter-argument (governance over prohibition) adds a nuanced position to the ethics landscape.