Trust in AI: supportive vs. autonomous roles across four domains

Abstract

Trust in artificial intelligence (AI) remains a central concern as autonomous systems become increasingly embedded in everyday life. This study investigates how users evaluate AI in two distinct roles—supportive decision aid vs. autonomous decision maker—across four societal domains: healthcare, finance, workplace decision-making, and education. Across all domains, participants reported substantially higher comfort with supportive AI, revealing a robust trust asymmetry in affective trust-related responses. We further examined whether individual differences, including personality traits, value orientations, demographics, and prior AI experience, predicted trust in either AI role. Experience with AI and openness-to-change emerged as meaningful predictors of trust in autonomous AI, suggesting that familiarity and value-driven dispositions shape acceptance of more independent systems. These findings underscore the importance of psychological mechanisms, such as exposure and value alignment, in shaping trust in AI and offer implications for the design of human-centered, trustworthy AI systems.