From AI assessment to AI evaluative judgement: evidencing human judgement in AI-supported vocational learning

Artificial intelligence has often been discussed in education as a tool for assessing students, generating feedback and supporting learning analytics. Generative AI creates a different assessment problem: learners must also evaluate machine-generated outputs. This focused narrative mini review conceptualises the shift from AI assessment to AI evaluative judgement in vocational and professional learning. Building on general evaluative judgement and its extension to generative AI, it presents AI literacy as the broader competency context and AI evaluative judgement as a task-level analytic lens for examining situated judgement over AI-supported work. The review proposes the AI Evaluative Judgement Loop, in which a learner evaluates a current artefact against vocationally relevant criteria, accepts, revises or rejects and re-prompts, verifies the outcome, justifies the decision and re-evaluates where necessary. The Loop attends to changing objects of judgement, artefact provenance and iterative transformation. Process traces such as interaction records, version histories and rationales provide observable evidence from which aspects of judgement may be inferred; they do not reveal cognition directly. Domain criteria, including aesthetic appropriateness where relevant, remain context-specific. The framework is analytic and requires further exploration through triangulated studies of learner judgement in authentic tasks.