AI supported doodling and gallery walk defence provide visible evidence of student reasoning in higher education assessment

The rapid integration of Artificial Intelligence (AI) into higher education challenges conventional assessment validity, as polished written outputs no longer reliably indicate students’ own reasoning or conceptual synthesis. This study examines an integrity-by-design assessment sequence intended to make student thinking visible and defensible by shifting evidential weight toward human synthesis and public accountability. A qualitative multi-cohort case study was conducted across three undergraduate Educational Technology course offerings at a public University, Oman (N = 76 students in 23 groups). We tracked cohort patterns using a standardised Visual Artefact Protocol to evaluate Conceptual Density and Relational Explicitness across all 23 hand-drawn artefacts. The sequence comprised four phases: (1) bounded AI-supported ideation; (2) collaborative, hand-drawn doodle synthesis; (3) a structured gallery-walk defence with rubric-guided peer evaluation; and (4) individual written reflection. Data from peer evaluations, doodle artefacts, defence traces, and reflections were analysed using reflexive thematic analysis. Four interrelated themes emerged: analogue constraint as sense-making, public explanation as accountability, AI as a scaffolder instead of a scribe, and friction as pedagogy. The study contributes an integrity-by-design framework that moves beyond binary accept/reject debates on AI. It demonstrates how strategically sequenced epistemic bottlenecks, analogue doodling and public defence can recentre human judgement as the non-negotiable core of assessment validity, thereby offering a thematically consistent and pedagogically coherent alternative to post hoc AI detection methods. The results indicate that the sequence supported integrative understanding of course concepts while generating auditable evidence of student reasoning, presenting a potentially transferable model that positions AI use transparently at the ideation stage, while preserving human synthesis and peer defence as central evidence of learning.