Generative Scaffolding in Pre-Trip Games: Effects of a GenAI-Chatbot on Inquiry-Based Learning

To address the issue of “travel without learning” in study tours caused by insufficient preparation, digital game-based learning (DGBL) serves as an engaging pre-trip solution. However, traditional DGBL’s static feedback limits open-ended inquiry and personalized learning. To overcome this, we proposed the DGBL-GAI approach to provide dynamic scaffolding, and developed a “Field Trip Preparatory Game” as its specific application. A preliminary quasi-experiment involved 83 seventh-grade students: the experimental group (n = 41) used the game guided by the DGBL-GAI approach, while the control group (n = 42) received traditional script-based static prompts. Results indicated that the DGBL-GAI approach was associated with better learning achievement and self-efficacy. Furthermore, an exploratory subgroup analysis based on post-intervention self-efficacy suggested a preliminary association: the achievement advantage of the DGBL-GAI approach appeared more evident among students with lower self-efficacy. Finally, behavioral pattern analysis provided supportive process-level evidence suggesting that, within the current sample, the experimental group appeared to engage in a more proactive and strategic learning pattern, using AI as an information integration hub and showing greater persistence, whereas the control group followed a more linear and passive learning pathway.