On the relationship between interpersonal trust and trust in artificial intelligence: empirical evidence, explanatory mechanisms, and policy implications

Abstract Artificial intelligence (AI) systems are becoming increasingly integrated into socio-technical infrastructures. However, their long-term deployment depends fundamentally on public trust. While existing research primarily focuses on system-level determinants of “trustworthy AI” (e.g., decision accuracy, reliability, transparency), less attention has been given to how broader societal trust structures influence trust in and acceptance of AI. This study examines the relationship between interpersonal trust and trust in AI across countries. Using cross-national survey data from the World Values Survey and a global AI study conducted by the University of Melbourne and KPMG, a matched sample of 38 countries was examined. Pearson ( r = −0.346, p = 0.033) and Spearman ( $$\rho=-0.406$$ , p = 0.011) correlation analyses revealed a statistically significant moderate negative association between interpersonal trust and trust in AI. Ordinary least squares (OLS) and Theil-Sen regression models confirm this negative relationship. Countries characterized by lower interpersonal trust tend to exhibit comparatively higher trust in AI, whereas countries with higher interpersonal trust often display more cautious attitudes toward AI systems and hence they trust AI less. These results support the substitution hypothesis that, in contexts of weak interpersonal trust, AI can serve as a compensatory mechanism for the perceived untrustworthiness of human actors and institutions. The paper presents four complementary explanatory mechanisms and derives policy implications for legislation, media discourse, and economic governance. The results suggest that AI governance strategies must be context-sensitive and embedded within broader societal trust structures.