Engineering education is undergoing rapid transformation as AI technologies reshape how students learn and interact with instructional support. This study proposes a retrieval-augmented generation (RAG)-based intelligent learning companion that is embedded into a human–AI collaborative teaching model to support students during physical analog circuit laboratory sessions. By combining a curated course knowledge base, learner profiles, and a locally deployed large language model, the system delivers highly accurate, real-time and personalized guidance when students encounter difficulties in experiments.

In a controlled experiment with 30 undergraduate students, the researchers compared traditional instructor-led guidance with instruction supported by the intelligent learning companion. The results show that, although the system had limited impact on knowledge acquisition and emotional attitude, it significantly improved students’ practical skills as well as key dimensions of flow experience, including immersion, time transformation, and autotelic enjoyment. These findings suggest that intelligent learning companions are most effective when positioned as complements rather than replacements for human teachers—handling routine procedural and low-level knowledge questions, while teachers focus on conceptual scaffolding and emotional support. This work offers practical implications for designing learner-centered, practice-oriented instructional models in the era of intelligent education.

The work titled “Investigating the Impact of an Intelligent Learning Companion on Learning Effect and Experience in Analog Circuit Laboratory Instruction”, was published on Frontiers of Digital Education (published on January 5, 2026).
DOI:10.1007/s44366-026-0079-x