ScholarWorks@숙명여자대학교

초록

This study aimed to enhance the practical applicability of essay-type assessments for pre-service teachers' teaching aptitude and personality, complementing the limitations of existing self-report measures. However, the viability of such assessments has been historically challenged by the time, cost, and reliability issues inherent in human scoring. To address this, the study adopted an integrated approach: first, developing a scenario-based assessment focused on educational problem-solving skills, and second, validating the application of generative AI (ChatGPT-4) for its automated scoring. A study with pre-service teachers revealed that while automated scoring was more lenient than human scoring, it demonstrated higher internal consistency and stability. The developed assessment showed discriminant validity through low correlations with self-report measures and convergent validity through its correlation with academic grades. These findings suggest the practical applicability and efficiency of an integrated assessment system that combines a scenario-based essay test with AI-powered scoring.

키워드