AI-scaffolded DC electricity: Simple engineering experiments with ChatGPT to foster higher-order thinking skills

Abstract

Low-cost laboratory environments often constrain the development of Higher-Order Thinking Skills (HOTS) and limit the provision of formative feedback. This study examines the effect of simple ChatGPT-assisted Direct Current (DC) engineering experiments on HOTS and elucidates the scaffolding mechanisms involved. Using a sequential explanatory mixed-methods design with a single-group pretest–posttest structure, thirty-four undergraduates completed two low-cost laboratory sessions where ChatGPT supported the validation of pre-experiment questions and the interpretation of measurement results. Quantitative data were obtained from constructed-response HOTS tests, complemented by structured interviews. Results show a substantial improvement (pretest M = 24.05 → posttest M = 56.87) and a sharp reduction in score dispersion (variance: 87.97 → 21.08; IQR: 12.20 → 3.20), indicating more homogeneous learning outcomes and a shift toward higher performance. Qualitative findings highlight two mechanisms: improved problem representation before experimentation and strengthened data–model reasoning afterward through rapid explanations and standardized procedures. Overall, the approach offers an affordable and scalable pathway for fostering HOTS through the integration of physical experimentation and Artificial Intelligence (AI)-based scaffolding. Its implication is that integrating low-cost physical experiments with AI support can serve as a practical approach to enhancing HOTS in physics learning.

Ähnliche Arbeiten