Abstract
This research discusses the development of Japanese learning materials using Large Language Model and Generative Artificial Intelligence or LLM-GAI. The materials are created using a custom Generative Pre-trained Transformer or GPT named “Nihongo Navigator,” which can generate content (main text, vocabulary explanations, sentence pattern descriptions, and drill exercises) tailored to the learner’s level. Additionally, a free conversation system called “GPT-Talk” has been developed, combining speech recognition and synthesis technologies, enabling practical conversational practice. Future challenges include reducing the complexity of prompts to build a more user-friendly interface and advancing practical research on how to use these materials in actual educational settings.
