Abstract
This study asks who turns to AI for advice and why. We investigate this question through the structure of individual social embeddedness. Using a national survey from South Korea (2023), we find that people’s social worlds are strongly associated with their choices. Those with many trusted confidants are less likely to consult AI—suggesting that close human ties can substitute for algorithmic guidance—whereas people embedded in more diverse networks are more likely to use AI, treating it as a complementary channel. Beyond networks, individuals who invest time in personally meaningful goals are more inclined to consult AI, pointing to a proactive, agentic pathway. Overall, AI consultation is not merely a function of demographics or tech attitudes; it is patterned by the social architecture of everyday life, implying that the diffusion of AI advice is likely to be socially patterned—concentrating among those whose close networks are thin and those whose diverse affiliations channel exposure to new tools—with implications for inclusive design and policy.