43 sources · 2020–2026 · Last updated: April 2026

A personal wiki covering the emerging literature on AI-assisted qualitative research — empirical benchmarks, methodological frameworks, epistemological debates, and practitioner ethics. Built to think with, not just to catalog.

Core question: Who holds interpretive authority — the researcher or the model? Every source in this wiki takes a position on this question, explicitly or not.


Concept map

Empirical benchmarks κ 0.72–0.82 on structured data Cultural depth: ~30% accuracy 58% quote fabrication (no constraints) Tool choice encodes epistemology Bijker · Bennis · Prescott · Salazar Sakaguchi · Hamilton · Fischer Jowsey (PLOS) · Ayik et al. Methodological frameworks CALM — human as interpretive ground truth CAAI — dialogue replaces coding AbductivAI — surface the unexpected AI-in-the-loop — deepen commitments GAITA — researcher as reflexive leader NITA — narrative replaces coding (PERFECT) Carlsen & Ralund · Friese · Costa Wise et al. · Nguyen-Trung · Xu Wheeler · Brailas The rejection debate Jowsey et al. — categorical rejection 419 signatories incl. Braun & Clarke De Paoli — philosophy ≠ methodology Greenhalgh — four governance Qs Friese et al. — assemblage/distributed cognition 100+ counter-signatories jowsey-et-al · de-paoli greenhalgh · friese-et-al-beyond-binary Critical perspectives & ethics Epistemic flattening — dominant patterns amplified Positivism creep — tools impose post-positivist norms QDAS marketing vs. qualitative epistemology Global South: infrastructure and equity gaps Data ownership, consent, sustainability Brailas · Chatzichristos · Paulus & Marone Dahal · Davison · Montrosse-Moorhead Reeping · Fischer · Sakaguchi Practitioner reality 13/14 experts use AI; many initially denied it Accepted: transcription, translation, writing Contested: coding, theming, analysis Refused: fieldwork, embodied ethnography "Illusion of meaning" — key practitioner risk Dellafiore et al. · Zhang · Dahal Christou · Nicmanis & Spurrier Hamilton · Perkins & Roe

Standing pages

These five pages are updated with every new source ingested. Start here for synthesis:

Page Focus
qualitative-ai-methods Living taxonomy: AI as coder, discoverer, dialogic partner, assistant, pipeline
epistemology Post-positivist · Interpretivist · Critical · Pragmatist · Post-humanist stances
human-ai-collaboration CALM, CAAI, AbductivAI, GAITA, AI-in-the-loop, Reflexive TA frameworks
validity-trustworthiness Reliability vs. validity gap; trustworthiness criteria; documented failure modes
contested-claims Ten genuine disputes with best evidence on each side
empirical-findings All empirical studies synthesized: benchmarks, tool comparisons, hallucination findings, practitioner behavior

Key empirical findings

Finding Evidence See
Reliable for structured small-q coding κ 0.72–0.82; Jaccard ≈ 1.00 under ideal conditions bijker-chatgpt-qca-2024, bennis-ai-thematic-analysis-2025
Fails at cultural/linguistic depth 80% descriptive vs. 30% culturally embedded (Japanese) sakaguchi-chatgpt-japanese-2025
Hallucination is design-dependent 58% fabricated quotes with no constraints; near-zero with structure jowsey-frankenstein-ai-ta-2025
Tool choice encodes epistemology ATLAS.ti/ChatGPT → post-positivist; MAXQDA/QInsights → interpretivist ayik-et-al-2026-human-vs-ai-ta-tools
Experts use AI covertly 13/14 Italian experts use AI; shame culture produces under-reporting dellafiore-et-al-2025-expert-interviews
Low-frequency codes systematically missed AI misses rare but analytically important items salazar-gpt4-qualitative-2025, prescott-ai-thematic-analysis-2024

The rejection debate (2025–2026)

jowsey-et-al-2025-we-reject — 419 signatories including Braun, Clarke, Lupton, Fine — argues GenAI is categorically incompatible with Big-Q reflexive research on three grounds: AI cannot make meaning; reflexive research must remain distinctly human; environmental and social justice costs are unacceptable.

Four responses:

  • de-paoli-reject-rejection-2026 — The letter conflates philosophy of mind with methodology. Human exceptionalism is a philosophical position (Searle, Chinese Room), not a methodological claim. Withdrawal cedes AI tool design to those without qualitative training.
  • greenhalgh-2026-beyond-the-binary — The binary is the problem. Right question: does AI displace, obscure, or constrain the researcher’s reflexive engagement? Some uses do; others don’t. Four governance questions replace the for/against binary.
  • wise-et-al-2026-ai-not-the-enemy — Maps LLM architectural properties (attention, long-context, pre-training) to five interpretivist commitments. AI can deepen interpretive work — not by replacing the researcher but by making thoroughness achievable.
  • friese-et-al-beyond-binary-2026 — The most theoretically dense counter-response. Four authors (Friese, Nguyen-Trung, Powell, Morgan) with 100+ co-signatories. Deploys four philosophical frameworks simultaneously — assemblage theory (Deleuze & Guattari), distributed cognition (Hutchins), posthumanism (Barad), sociomateriality (Orlikowski) — to contest the “exclusively human” claim at its foundations. Also notes that Braun & Clarke’s signature on a categorical rejection letter contradicts their own prior insistence that reflexive TA is flexible with no single right way.

The underlying philosophical question — whether meaning-making is irreducibly human — cannot be resolved empirically. The counter-response is consolidating: three independent scholarly responses now contest the Jowsey position on philosophical grounds, while the empirical literature continues to develop practical frameworks for responsible integration. See contested-claims Claims 9 and 10.


Concept pages


All 43 sources

Foundational nelson-computational-grounded-theory-2020 · carlsen-ralund-computational-grounded-theory-2022 · anis-french-ai-qualitative-research-2023

Empirical benchmarks bijker-chatgpt-qca-2024 · bennis-ai-thematic-analysis-2025 · prescott-ai-thematic-analysis-2024 · salazar-gpt4-qualitative-2025 · sakaguchi-chatgpt-japanese-2025 · jowsey-frankenstein-ai-ta-2025 · hamilton-ai-qualitative-2023 · yang-gpt4-qualitative-guide-2025 · ayik-et-al-2026-human-vs-ai-ta-tools · fischer-llm-qda-2024

Frameworks friese-caai-framework-2026 · costa-abductivai-2025 · nguyen-trung-gaita-2025 · nguyen-trung-nita-2026 · wise-et-al-2026-ai-not-the-enemy · ubellacker-academiaos-2024 · sinha-gpt4-grounded-theory-2024

Practical guides xu-ai-thematic-analysis-2026 · naeem-chatgpt-ta-steps-2025 · goyanes-chatgpt-protocol-2025 · nicmanis-spurrier-ai-guide-2025 · christou-ta-through-ai-2024 · christou-ai-resource-2023 · perkins-roe-genai-inductive-2024 · zhang-ai-qualitative-research-2025 · yang-gpt4-qualitative-guide-2025

Critical / epistemological brailas-ai-qualitative-research-2025 · chatzichristos-ai-positivism-2025 · williams-ai-paradigm-shifts-2024 · paulus-marone-qdas-discourse-2024 · wheeler-technological-reflexivity-2026 · dahal-genai-qualitative-nepal-2024 · dellafiore-et-al-2025-expert-interviews

The rejection debate jowsey-et-al-2025-we-reject · de-paoli-reject-rejection-2026 · greenhalgh-2026-beyond-the-binary · friese-et-al-beyond-binary-2026

Cross-disciplinary andrews-progress-or-perish-2026

Ethics and evaluation davison-ethics-genai-2024 · reeping-llm-quality-framework-2025 · montrosse-moorhead-ai-evaluation-2023