Development and Validation of an AI-Powered OCR Educational Tool for Early Handwriting Instruction for Korean Early Elementary Students

This work developed and classroom-deployed the AI Hangul Writing Teacher, an artificial-intelligence–based optical character recognition tool designed to deliver scalable, individualized feedback in early handwriting instruction. Paper worksheets were photographed by teachers and processed on the server; students in the feedback condition received prompt post-submission reports showing overall accuracy, whereas the control condition operated in a measurement-only mode in which accuracy was computed but not displayed during practice. In a 10-day, class-level randomized classroom study with 91 first-grade students, handwriting recognition accuracy in the feedback group increased from 65% to 95%, compared with 66% to 73% in the control group. Post-test measures of learning flow and academic motivation also favored the feedback group (p < .001). These results are consistent with the value of prompt, system-generated feedback for strengthening early handwriting performance and engagement while preserving a paper-based workflow. The study further indicates that well-designed recognition systems can provide effective and scalable support for foundational literacy instruction. Limitations include the single-school setting, the 10-day duration, and provision of overall accuracy only (no per-character highlights). Future work will broaden the evaluation across schools and grade levels, examine more granular formative feedback and long-term retention, and optimize the capture pipeline and teacher-facing tools.