The Power to Question the Answer : Development of a Critical Action Learning Instructional Model Based on Generative AI Response Type Classification
TL;DR
This study examines how the rapid diffusion of generative artificial intelligence may weaken the cycle of critical questioning and reflection at the core of Action Learning in university education and proposes a tool-independent framework applicable across changing AI systems.
Abstract
This study examines how the rapid diffusion of generative artificial intelligence may weaken the cycle of critical questioning and reflection at the core of Action Learning in university education. Drawing on Revans’ Action Learning theory and Sweller’s cognitive load theory, it identifies how the passive acceptance of AI-generated outputs diminishes germane cognitive load. On this basis, it classifies generative AI responses into three types-definitive, reasoning-transparent, and multiple-alternative- according to the assertiveness of conclusions, the visibility of reasoning, and the plurality of alternatives. For each type, this study derives a corresponding set of learner analysis activities, such as evidence tracing, premise identification, and context-based prioritization. These are integrated into a four-stage cyclical model of AI use, critical analysis, reconstruction, and contextual application, accompanied by classroom scenarios and a process-oriented assessment rubric. For liberal education, this approach reframes the threat that AI poses to critical thinking as an occasion to strengthen it by turning AI outputs into material for critical analysis. By proposing a tool-independent framework applicable across changing AI systems, this study repositions AI outputs not as endpoints of learning but as starting points for critical inquiry and reconstruction.