Rethinking Young Children's Social Information Processing in the Era of Large Language Models

Children's social information processing (SIP) has been framed within human-to-human interactional contexts. However, rapid diffusion of conversational AI introduces socially responsive nonhuman agents whose outputs are not shaped by lived experience, personal stakes, or cultural situatedness. We argue that children's SIP will be reshaped by these interactions. We begin by describing the changing nature of social input in this current age. We then show how AI assistants function as conversation partners: leaning structurally toward validation, constraining content but not relational dynamics, and generating confident outputs that may rest on sparse or false knowledge. We trace anticipated effects across five SIP mental stages: encoding, interpretation, goal clarification, response construction, and response decision. Finally, we discuss implications for theory, research, and practice.