The nonlinear relationship between artificial intelligence development and income inequality: the moderating effect of education level

Purpose The inequalities emerging from AI development have garnered significant attention. From the perspective of AI’s dynamic development, this study aims to explore the relationship between AI and income inequality as well as the moderating role of education level on this relationship. Design/methodology/approach We use panel data on 274 cities in China from 2012 to 2021 and calculate the AI development index using the entropy method. Then we employ a regression model with two-way fixed effects for analysis. Findings The results indicate a significant U-shaped relationship between AI and income inequality, with the turning point occurring at an AI development level of 0.112. Specifically, when the AI development level is below 0.112, AI tends to narrow income inequality; while above this threshold, it progressively exacerbates income inequality. Education exerts a significant negative moderating role in this relationship, with higher education levels shifting the U-shape to an inverted U-shape. Originality/value First, we develop a multidimensional indicator system to measure AI development, addressing the limitation of a single indicator in existing research. Second, we analyze AI’s dynamic trajectory and its relationship with income inequality across different stages. More importantly, we introduce education as a moderating variable, a factor overlooked in prior research that can mitigate the income inequality. This study provides direction for future research on AI and social inequality in various contexts and offers empirical evidence to help policymakers regulate AI through improved education levels.