Cognitive Offload Visibility: A design principle for instructional design and assessment in AI-mediated nursing education.

Abstract

Generative artificial intelligence (AI) is embedded in the everyday work of nurse educators and students, yet the boundary between learner cognition and AI cognition often remains invisible. AI-assisted work can circulate without explicit differentiation between the contribution of the tool and the human, thereby confounding the representation of cognitive work by artifacts of learning. This article introduces Cognitive Offload Visibility (COV), or the deliberate design of learning and teaching to make explicit which cognitive work belongs to the learner and which was offloaded to AI, as a design principle for AI-supported instructional design and assessment in nursing education. Grounded in Cognitive Load Theory, metacognition, and professional identity formation, COV positions disclosure as a formative practice rather than a compliance requirement. The article also introduces the Authentic Evidence Framework (AEF), which operationalizes COV across four layers of assessment evidence that AI cannot straightforwardly produce: process, judgment, context, and presence. Together, COV and the AEF complement the AI Assessment Scale by grounding assessment design in visible learner cognition. Application to three common nurse educator concerns (online discussion forums, scholarly writing, and high-stakes examinations) illustrates how the frameworks preserve assessment validity in AImediated learning environments. The article also addresses the time these designs require, arguing that COV redistributes instructional effort rather than reducing it and that workload policy determines whether it can be adopted at scale. The article argues that assessment in the AI era is a design task rather than a surveillance exercise and outlines implications for practice, research, and policy, with faculty development identified as central to its realization.