Human–AI interactions reshape the self and our social networks

Large language models (LLMs) have rapidly entered everyday life as companions, advisors and sources of knowledge. Unlike earlier forms of digital media, they do not merely transmit information: they adapt to prompts, simulate empathy and provide interaction that feels personalized and reciprocal¹. This shift requires urgent attention in how social practices are learned and propagated: habits formed in interaction with LLMs do not stay on-screen but diffuse through alignment-driven exchanges, potentially reshaping human behaviour, mental well-being and social connection at different scales.