Harnessing Ambiguity in Challenge-Based Learning as a Practice of Deep Engagement with Large Language Models

Large language models (LLMs) epistemically and ethically disrupt established forms of student learning and teaching. Since real-world problems rarely yield clear solutions, the question of how students engage with ambiguity becomes central when integrating LLMs into education. While ambiguity is often framed negatively—as a lack of clarity provoking avoidance—I argue that ambiguity can also serve as a catalyst for deeper learning. To examine this potential, I propose an epistemic-ethical lens that addresses five categories of the sociotechnical educational system: ontology, aims, methods, practices, and environment. Applied first, the lens reveals how ambiguity aversion may exacerbate epistemic and ethical disruptions when students use LLMs. Applied again, it shows how pedagogical designs that embrace ambiguity can counter these risks. Challenge-based learning (CBL) exemplifies such a design. In CBL, students collaborate on real-world problems presented by external challenge owners who themselves lack solutions—even with access to advanced AI systems. These challenges remain complex and open-ended, ensuring that ambiguity drives authentic inquiry and exploration. By positioning ambiguity as a productive force rather than a barrier, CBL creates the conditions to address epistemic and ethical concerns of LLM-supported education at a practical level. The article contributes in three ways: advancing the epistemic-ethical debate through a novel lens, reframing ambiguity as a catalyst in the age of LLMs in education, and demonstrating how CBL can operationalize this insight in educational practice. Limitations and directions for future research and practice to harness ambiguity for deeper student engagement with LLMs are discussed.