Beyond Hype: Empirical Evidence and Guidelines on Generative AI in Qualitative Analysis

Objective: this article explores the integration of generative artificial intelligence (GAI) into qualitative research, comparing its results with human-led analysis and traditional AI techniques. It provides empirical evidence on the strengths and limitations of GAI in thematic analysis while addressing ethical concerns and research integrity. Theoretical approach: the study situates itself within the debate on GAI in qualitative research, weighing utilitarian arguments such as efficiency and scalability against ethical concerns, such as the loss of human interpretive depth. It extends discussions on human–machine collaboration by incorporating a three-way comparison: human analysts, traditional AI (e.g., topic modeling), and GAI. Method: a qualitative secondary analysis (QSA) compares human-led analysis, traditional AI, and GAI using a dataset on women’s participation in the game development industry, aligning with Sustainable Development Goal (SDG) 5 on gender equality. The study evaluates outputs based on thematic accuracy, interpretive depth, and practical utility. Results: findings suggest that GAI excels in speed and scalability, particularly for deductive methodologies with predefined coding labels. However, human analysts outperform GAI in interpretive depth and theoretical connections. Traditional AI offers structured insights but lacks GAI’s adaptability or human researchers’ nuanced understanding. Conclusions: while GAI enhances efficiency and reduces costs, its limitations in replicating human interpretive capacity call for cautious integration. This study provides guidelines for leveraging GAI’s strengths while preserving human expertise, contributing to the broader discourse on technology’s role in qualitative research.