Analysis of Adult Learners' Usage Preferences and Requirements for Agent-Based AI Teaching Assistants

With the rapid development of large language model technology, the applications of agent-based AI in various fields have received extensive attention. However, few studies have focused on the usage preferences and application requirements of adult learners regarding agent-based AI teaching assistants. In this study, direct-type and heuristic AI teaching assistant agents for the "Basic Computer Applications" course were developed based on the Coze platform, and applied in open education courses. The interaction data between adult learners and the AI teaching assistants are analyzed with inductive and statistical methods, and the research shows that these agents can provide students with continuous support for course knowledge anytime and anywhere. Findings indicate that adult learners primarily interact with the AI teaching assistants through WeChat, prefer direct-type AI teaching assistants for solving course test problems, and also express a need for emotional support. This study offers novel insights and recommendations for researchers and developers of AI teaching assistants for adult learners.