Dataset for AI Dependency Dual-Path Model

This study tests a dual-path model of generative artificial intelligence (GenAI) dependency and its associations with health information-seeking behavior and advice adoption. The model proposes that perceiving GenAI as a functional assistant relates to machine-heuristic perceptions, while perceiving it as a social companion relates to anthropomorphic perceptions. Both mechanisms are expected to predict intention to seek health information via GenAI, which in turn predicts willingness to adopt its health advice. A nationally representative survey of 1,200 South Korean adults was conducted in April 2025 using stratified cluster sampling. Partial least squares structural equation modeling (PLS-SEM) supported the proposed associations. Perceiving GenAI as an assistant predicted stronger machine-heuristic perceptions, while viewing it as a companion predicted higher anthropomorphic perceptions. Both pathways were positively related to intention to seek health information, which strongly predicted willingness to adopt GenAI-generated advice. These findings extend health information research by showing that GenAI functions not only as a source of knowledge acquisition but also as a potential advisor, with distinct cognitive and relational pathways of user dependency.