A ChatGPT-supported QIVE model to enhance college students’ learning performance, problem-solving and self-efficacy in art appreciation

The study proposed a ChatGPT-based learning approach (GPT-LA), which guide students to interact with ChatGPT following the QIVE (i.e. Question, Interaction, Verification, and Examination) stages, for art appreciation. To evaluate the proposed approach, a quasi-experiment was implemented to compare the art appreciation capability, problem-solving tendency, self-efficacy, and cognitive load of college level students using the ChatGPT-based learning approach to those adopting the conventional learning approach (C-LA). The experiment involved 53 university students from two different classes. One class of 27 students comprised the experimental group adopting the GPT-LA. The other class, consisting of 26 students, served as the control group and utilized the C-LA (referencing learning materials and web-based search). The results revealed that the GPT-LA significantly enhanced learners’ art appreciation capability and self-efficacy, and lowered their cognitive load compared to the C-LA. Moreover, students perceived higher usefulness of the GPT-LA than that of the C-LA. Suggestions for future application of ChatGPT in instruction are provided.