AI-Assisted General Education for Information Literacy: Designing a Dual-Verification Mechanism

Students without an information technology background often rely heavily on large language model (LLM)–based AI tools in general education courses on information literacy. However, such tools may generate content with erroneous sources, sycophantic responses, or outdated data. These limitations make it difficult for students to evaluate information reliability and can even lead to incorrect citations due to overreliance. To address this issue, this study proposes a dual-verification learning mechanism grounded in generative AI. The system consists of a retrieval agent that gathers and exchanges information with a verification agent, followed by teacher validation and annotation. This process not only produces a reliable retrieval-augmented generation (RAG) dataset for students but also ensures both accuracy and pedagogical support. Findings indicate that the proposed dual-verification model is feasible in the context of general education. Beyond providing students with teacher-verified resources, it also enhances instructional efficiency and teaching quality, offering valuable insights for the design of future interdisciplinary information literacy curricula.