When robots malfunction: children’s and teachers’ responses in Korean preschool classrooms

This study investigates how children and teachers in Korean preschool classrooms respond to naturally occurring robot malfunctions during interactions with AI-powered humanoid robots. Over a three-month period, two robots were integrated into naturalistic play sessions with five-year-old children. Drawing on lag-1 sequential analysis, the study identified immediate interactional patterns surrounding robot breakdown episodes. Children’s verbal and embodied reactions – such as repetition, adjusted phrasing, name-calling, emotional disengagement, and laughter – suggest that children actively attempted to manage or restore interaction during robot malfunction episodes. These responses indicate that children treated robots as responsive social actors and adjusted their interactional behaviors during breakdown situations. Robot malfunctions also often involved peers and teachers in managing breakdown situations, turning individual breakdown episodes into socially shared interactional events rather than isolated technical failures. The findings highlight how robot malfunctions become embedded within classroom interactional dynamics. This study contributes to research on classroom-based human–robot interaction by examining the ecological and sequential organization of robot breakdown episodes in naturalistic preschool settings.