A three-wave study of human-AI fit and adaptive performance

Generative artificial intelligence (AI) is increasingly embedded in knowledge work, yet less is known about how employees develop effective working relationships with AI across repeated interactions. Drawing on a three-wave, time-lagged survey of 447 employees using DingTalk, this study examines how perceived AI adaptability and user proactivity shape human-AI fit and, in turn, adaptive performance. The study combines partial least squares structural equation modeling (PLS-SEM) with fuzzy-set qualitative comparative analysis (fsQCA). The PLS-SEM results show that human-AI fit partly mediates the relationships between perceived AI adaptability, user proactivity, and adaptive performance. The fsQCA results further identify two equifinal pathways to high adaptive performance: a technology-driven pathway combining high AI adaptability with high human-AI fit, and a human-driven pathway combining high user proactivity with high human-AI fit. Across both configurations, human-AI fit appears as a core condition. By integrating adaptive structuration theory, experiential learning theory, and person-environment fit theory, this study explains human-AI collaboration as a temporally ordered process in which technological responsiveness and human initiative jointly support relational alignment. The findings contribute to research on human-AI collaboration and highlight the need to design AI-enabled work systems that support both adaptive performance and critical human oversight.