Challenges in Parent-Supervised Setup of Simulated Socially Assistive Robots for Personalized Early Childhood Learning

This study investigates challenges parents face in configuring simulated Socially Assistive Robots (SARs) for early childhood education. The research focuses on usability barriers in task setup, personalized learning customization, and AI-driven adaptive feedback using ChatGPT within a simulated learning environment. Ten parents and eight children participated in configuring SAR tasks such as counting and letter recognition. Key challenges identified include complex initial setup (reported by 40% of parents), difficulty in interpreting AI-generated feedback (M =2.8, SD=0.92), and usability issues in speech recognition (M=3.5, SD=1.01, reported by 70% of parents). We analyzed the data using descriptive statistics and thematic analysis, focusing on usability patterns from both structured questionnaires and open-ended responses. The results emphasize the importance of having clear task templates, incorporating multiple forms of guidance, and using AI prompts that adapt to the context. Unlike previous studies on physical SARs, this research identifies key usability challenges that persist even in simulated environments, emphasizing the need for AI-driven adaptive assistance and user-friendly customization. The findings contribute novel insights into AI-powered SAR usability for non-expert users in home-based, simulation-driven education.