CreaRobot - Improving creativity using a social robot

Supporting children’s creativity in educational settings remains a key challenge, which generative AI-powered social robots may help address. This paper presents CreaRobot, a study in which children performed a standardized drawing task while interacting with a social robot driven by a vision-language model. We compared two conditions: a generic feedback condition, in which the robot produced pre-recorded generic compliments, and a contextualized feedback condition, in which the robot engaged in structured interaction phases grounded in a real-time visual analysis of each child’s drawing. Results from 49 participants show that contextualized feedback significantly improved the value of the creative product (M = 21.52 vs. M = 13.85; Cohen’s d = 1.34), with a large effect size. These preliminary findings suggest that grounding a social robot’s feedback in real-time perception of the creative process is a promising direction for AI-supported creativity in educational settings.