Beyond access: Engagement with conversational AI-assisted feedback shapes cognitive load in digital game-based learning

To examine the cognitive consequences of conversational AI-assisted feedback in complex digital learning environments, this study employed a between-subjects longitudinal field design to investigate how both access to and actual engagement with conversational AI-assisted feedback were associated with students' perceived extraneous and germane cognitive load during a five-week business simulation game. Because conversational AI-assisted feedback represents an interactive form of instructional support whose effectiveness depends on learner-related and contextual characteristics, the study further examined the moderating roles of simulation progression (time), intrinsic cognitive load, and metacognitive self-regulation. The findings indicate that regular engagement with conversational AI-assisted feedback did not uniformly reduce extraneous cognitive load; rather, its effects depended on learners' intrinsic cognitive load and the progression of the intervention. In contrast, students who regularly engaged with the conversational AI-assisted feedback reported significantly higher germane cognitive load than both students without access to the chatbot and students with access who did not use it. Overall, the findings advance understanding of human-AI interaction by demonstrating that the cognitive effects of conversational AI-assisted feedback are shaped by dynamic patterns of learner engagement, individual learning characteristics, and temporal changes during extended AI-supported learning interventions.