Situated AI judgement: A teaching and assessment model for generative AI in higher education

Abstract

This conceptual design article proposes situated AI judgement as a teaching and assessment model for responsible generative AI use in higher education. It argues that students need to learn not only how to use AI tools, but also how to evaluate, challenge, revise, and refuse AI outputs in relation to disciplinary standards, evidence, and ethical boundaries. Building on evaluative judgement and assessment scholarship, the article presents a four-stage model: pre-AI interpretation, structured AI dialogue, critical comparison, and reflective accountability. The model is developed through an anthropology and heritage studies activity in which students analyse public nomination files submitted under UNESCO’s 2003 Convention for the Safeguarding of the Intangible Cultural Heritage, especially the nomination of Momoeria, New Year’s celebration in eight villages of Kozani area, West Macedonia, Greece. The article shows how this case can help students examine institutional language, community participation, safeguarding, consent, visibility, and decontextualisation while judging the limits of AI-generated summaries. It contributes to applied learning and teaching by offering an assessment design that makes AI-mediated judgement visible and may be adaptable across disciplinary contexts, with a brief inclusive education example illustrating how the same logic might be adapted beyond heritage studies. The article operationalises responsible AI use not as disclosure or compliance alone, but as an assessable capacity to interpret, verify, critique, and ethically limit AI use within a specific field.