Generative AI-assisted reflective writing for improving students’ higher order thinking: Evidence from quantitative and epistemic network analysis

The emergence of Generative AI such as ChatGPT—a cutting-edge pre-trained language model—has garnered unprecedented attention from educational researchers worldwide. Its remarkable performance in knowledge generation and natural language dialogue has sparked global scholarly interest in its implications for education. Research on higher-order thinking skills (HOTS) has predominantly been conducted in K-12 education. However, despite our aging society, there has been scant investigation into students’ HOTS regarding supportive caregiving for elderly adults, particularly from the perspective of Embodied Cognition Theory (ECT). This study examined the potential of ChatGPT to enhance students’ reflective writing skills. We investigated how ChatGPT assists students in exploring “Conflicts of Roles in Home Care for the Elderly” through reflective writing to understand their perceptions of caregiving and analyze their HOTS. A quasi-experiment was conducted to compare learning attitudes and empathy between an experimental group and control group. Students’ reflective writing was coded into six categories. The results indicated that students using the ChatGPT-integrated ECT learning method demonstrated significantly better learning attitudes and empathy. Although there was no significant difference in HOTS scores between the two groups, the Epistemic Network Analysis (ENA) results revealed that the experimental group exhibited a more diverse and interconnected conceptual understanding. This study integrated ECT with ChatGPT and employed ENA, contributing to theoretical and practical adjustments in learning environment frameworks based on an analysis of students’ understanding of the material.