Implementing Generative AI Agent Game to Support Reading of Classical Chinese Literature: A Needs Analysis

This study investigated the challenges faced by middle school students when engaging with "Study with Confucius," a generative artificial intelligence (GenAI) agent game designed for learning classical Chinese reading. Utilizing the framework proposed by Groff and Mouza (2008), the research aimed to conduct a comprehensive needs analysis across three key dimensions: student-related, teacher-related, and technology-related aspects. Data were collected from 29 students in mainland China through video recordings of their gameplay. Challenges were defined based on their duration and students' responses that contradicted predefined correct answers, resulting in the identification of 29 challenges. Findings indicated that students encountered student-related challenges including linguistic misinterpretation, distraction, cognitive overload, student misbeliefs, and inappropriate attitudes; teacher-related challenges such as lack of support and inadequate access to teaching and technological resources; and technology-related challenges including external and internal malfunctions. These insights contribute to understanding the complexities as well as learning opportunities of integrating GenAI agent games into reading education.