Discernment in generative AI assisted instructional design: An evaluation map and design heuristics for learning experience designers

Abstract

Existing research shows that generative artificial intelligence (GenAI) increases instructional design efficiency. However, there are concerns about pedagogical quality, accuracy and content bias. While prior studies address student passivity in evaluating AI content, the potential for designers to be passive in creating instructional materials remains largely unexplored. This article addresses design safeguards for AI-assisted course and learning-material development through two complementary tools: the Discernment Map, an evaluation checkpoint for instructional designers, and a set of design heuristics. The study is conceptual and design-oriented: both tools were drafted from the literature on AI fluency, automation bias and schema theory, and then refined through structured consultations with six experienced instructional designers in higher education. Though yet to be empirically validated, the practice-informed tools work at two levels: the Map operationalises human oversight of AI outputs across Product, Process and Performance dimensions for each AI-generated artefact, while the heuristics extend discernment into the design of the learning experience itself, so that learners must evaluate rather than accept AI outputs. Both are agile micro-governance tools that integrate critical judgment directly into design workflows. The approach is intended to protect long-term instructional quality against the collective erosion that unchecked AI adoption risks, and to support intentional engagement with AI that reduces automation bias.