An Intelligent Agent for Teaching Reflection on Critical Thinking: Model Construction and Empirical Study

Providing students with appropriate reflective support is a key mechanism for fostering critical thinking, but traditional reflection methods remain limited in terms of personalized support and face challenge from “pseudo-reflection”caused by generative artificial intelligence. Teaching reflection agents have brought new opportunities for reconstructing reflection mechanisms, yet their impact on critical thinking remain to be thoroughly explored. Accordingly, based on critical thinking and GIBBS reflective theory, this paper constructed a teaching reflection agent model oriented toward critical thinking. Using a quasi-experimental design, the study conducted an empirical study with two classes from University S in Central China taking the “Internet Thinking and Digital Autonomous Learning”course to verify the effectiveness of the teaching reflection agent. The results showed that the teaching reflection agent significantly improved students’overall critical thinking levels. There were notable differences in the enhancement effects between high- and low-performing groups: the high-performing group exhibited greater improvement in critical thinking, while the low-performing group benefited more in terms of reflective interest and competency development. This study aimed to provide theoretical and practical references for the targeted instructional design of intelligent reflection tools and optimization strategies for critical thinking.