Whose Judgement Counts? An Appraisal-Informed Critical Discourse Analysis of Educational AI Communications

Abstract

Public guidance on generative AI asks students to exercise judgement while institutions and platforms define the conditions of use. This study examines how these texts distribute authority over writing, evidence and assessment. The purposively selected corpus comprises eight official webpages from three UK universities and Google and captions from three educational videos, retrieved in October 2026. Appraisal-informed critical discourse analysis examines 30 decision-focused units: passages addressing a particular choice or responsibility. Analysis also considers modality and legitimation, meaning the reasons offered to justify a practice. The main finding is boundary maintenance: texts distinguish what fluent writing, technical capability or disclosure can establish about learning, reliability and permission. University guidance and Gemini promotion both value learner participation, although they serve different purposes. Staff guidance also encourages discussion of assessment criteria. These findings suggest that teachers should explain acceptable assistance, student responsibilities and opportunities to question criteria. The small documentary corpus supports interpretation of published communications, not conclusions about classroom practice or AI effectiveness.