Developing and Implementing a GPT-API Chatbot for Language Classroom: Design and Preliminary Insights

• Published in: 멀티미디어 언어교육, 28(1) pp. 22 - 38
• Main Authors: 황요한, 임세범
• Format: Journal Article
• Language: English
• Published: 한국멀티미디어언어교육학회 01.02.2025
• Subjects: 영어와문학
• ISSN: 1229-8107; 2982-7302
• DOI: 10.15702/mall.2025.28.1.22

This study investigates the educational potential of GPT-API-based chatbots in elementary language classrooms, emphasizing their effectiveness, usability, and student perceptions in comparison to WRTN, a LLM-based chatbot. Recognizing the limitations of traditional rule-based and pre-programmed educational chatbots, this study explores how GPT-API chatbots, tailored to specific educational purposes, can address these challenges. A GPT-API chatbot was developed using OpenAI’s API, designed to guide students in creating narratives based on personal experiences. The study involved 22 elementary school students who used the both GPT-API chatbot and WRTN for two weeks during expressive writing activities. Findings from surveys and interviews reveal that the GPT-API chatbot excelled in aligning with course objectives, providing ease of use, and delivering accurate feedback, which enhanced its perceived educational value. In contrast, WRTN’s broader scope and occasional inaccuracies diminished its effectiveness in supporting focused learning tasks. Students rated the GPT-API chatbot positively for its learning support, adaptability, and future applicability, although it scored lower on human-like interaction. Qualitative data further emphasized the importance of purpose-driven design, with students preferring the GPT-API chatbot for its targeted and reliable support in writing tasks. Based on these findings, this study highlights the transformative potential of GPT-API-based chatbots in language and writing classrooms. KCI Citation Count: 0