Governing Classroom AI: Transparency, Integrity; and Learning in TEFL/EAP

Abstract

The rapid uptake of generative AI in TEFL and EAP has outpaced course-level governance, particularly with respect to transparent AI use, academic integrity, and data protection. This study aims to develop and evaluate a human-centred AI Classroom Contract as an auditable micro-policy that bridges the gap between guidance and classroom practice. Over eight months, we conducted a design-based study combined with a quasi-experimental comparison across classes with three collection points: baseline, mid-course, and end. Data included per-assignment AI transparency forms, learning-management-system logs on revision iterations, time on task, and punctuality, draft–revision artefacts with writing and speaking rubric scores, integrity audits, and surveys on perceived fairness and transparency. Analyses used multilevel modelling and mediation tests. Findings indicate higher completeness of transparency statements, fewer mis-citation and fabrication incidents, more iterative revision, improved punctuality, and moderate gains in task performance. Perceived fairness and transparency mediated effects on compliance and outcomes. Implementation was feasible under limited connectivity without meaningful privacy breaches. We conclude that the AI Classroom Contract is an effective micro-policy instrument that connects human-centred principles to TEFL and EAP practice, yielding a replicable package of the contract, templates, and a fact-check rubric to support quality and accountability in language learning.

Ähnliche Arbeiten