In this study, we examine domain experts’ integration of artificial intelligence recommendations in the workplace. Although ethnographies reveal domain experts' openness, widespread non-integration persists. To understand this puzzle, we conduct a qualitative comparative meta- analysis combining qualitative meta-analysis with fuzzy-set qualitative comparative analysis. Drawing on studies in management and information systems journals, we analyze how sociotechnical and behavioral parameters coalesce to explain integration and non-integration. We identify theoretical patterns involving relational expertise, motivational alignment, scrutiny, and technology types. Our study offers perspective on AI opacity, tool versus counteractor views, and offers a new approach for synthesizing literature.
