Different Logics, Similar Intentions: Heterogeneity in How Employees Evaluate AI Adoption

Research on artificial intelligence (AI) adoption at work has documented heterogeneity in user attitudes and adoption profiles, but less is known about whether employees differ in how they weight adoption-relevant information. In a within-subject structured-profile experiment, 446 employees provided 7,136 intention judgments across 16 profiles that varied AI adoption benefits, barriers, replacement vulnerability, and anticipated work harm. Multilevel regression mixture model analysis identified four latent decision-policy classes with different estimated weights on opportunity and threat cues; class mean intentions were not statistically distinguishable. Drawing on multiple-cue judgment theory, we interpret this pattern as heterogeneity in cue utilization rather than simply in attitude levels. Explaining and managing AI adoption therefore requires attention to latent decision policies, because aggregate effect estimates and profiles based on mean levels can conceal differences in cue responsiveness that shape how employees evaluate AI.