Exploring the Feasibility of AI-Driven Educational Robots in Simulation Environments for Early Childhood Learning: A Pilot Study

Abstract

This paper presents a pilot study on using simulated socially assistive robots (SARs) integrated with ChatGPT for early childhood learning. The study explores whether simulated SARs can engage children and enhance learning outcomes similarly to physical SARs. In this exploratory study, five 4-year-old children and three 3-year-old children engaged in individual counting tasks during sessions as part of the research. Measures of engagement with the task, such as completion rates and attention, were recorded. The results indicate that while the 4-year-olds consistently performed well, the 3-year-olds showed varying degrees of engagement and learning improvement, with some requiring additional guidance. The study demonstrates the potential of scalable, AI-driven SARs in early education and suggests future directions for larger-scale applications.