Asymmetric impacts of artificial intelligence on housing price valuation across education levels

This study examines the scenarios under which asymmetric estimations can arise from employing artificial intelligence (AI) models to assess housing prices. Advances in the pricing instrument effectively tackle the inherent nonlinearity issue in relevant datasets, and AI technologies have demonstrated superior predictive power compared with traditional approaches. However, AI models may penalize minority groups against the majority in a social manner. We empirically explore the potential negative externalities, specifically the asymmetric estimations that can arise across social groups when using machine learning technology to assess housing prices. Our findings highlight three notable observations. First, education levels are significantly and positively associated with housing prices. Second, AI models can appraise housing prices more precisely compared with conventional hedonic models. Finally, AI models tend to overestimate the housing prices of well-educated groups and underestimate those of less-educated groups. These results indicate that AI models improve the predictive power of price assessments; however, indiscriminate adoption and application of AI-based predictions may aggravate social inequality. Our findings provide insights into ways to alleviate inequality in urban areas; thus, policymakers can refer to our empirical evidence when designing initiatives to enhance social inclusion and coherence, and when considering strategies to realize balanced urban development.