As generative AI tools become more commonly used in qualitative research, their role in supporting thematic analysis (TA) calls for critical exploration. This study compares the thematic outputs of four AI tools—ChatGPT-4o, QInsights, ATLAS.ti AI, and MAXQDA AI Assist—against a validated human-coded analysis of K–12 STEM teacher data. Using Braun and Clarke’s six-phase TA framework, we examined how each tool’s outputs aligned with, extended, or diverged from human-generated themes. Findings showed partial convergence across tools, no hallucinated content, and substantial time savings. While none fully replicated the depth of human interpretation, all captured relevant insights. This comparison offers a deeper epistemological understanding of AI integration into qualitative inquiry, as well as practical guidance for researchers considering its thoughtful application in qualitative analysis.
Ayik et al. (Fri,) studied this question.