AI assessment changes human behavior
Jonas Goergen, Emanuel de Bellis, Anne-Kathrin Klesse
Abstract
AI is increasingly replacing human decision-makers across domains. AI-based tools have become particularly common in assessment decisions, such as when recruiting employees or admitting students. Calls for transparency and new legislation require organizations to disclose the use of AI assessment tools, thus making people under assessment aware of its use. We investigate whether this shift from human to AI assessment affects people’s behavior during the assessment. We propose that people emphasize their analytical characteristics and downplay their intuitive and emotional ones under AI (vs. human) assessment, a phenomenon we label “the AI assessment effect.” Twelve studies (eight in text and four in the Supporting Information; N = 13,342) document the AI assessment effect and its underlying mechanism: the lay belief that AI prioritizes analytical characteristics in its assessment. Whereas prior work has studied perceptions of AI assessment tools and their productivity gains, the current research demonstrates systematic behavioral changes because of AI assessment. The findings offer theoretical contributions to the psychology of AI and practical insights for organizations using AI assessment.