The GUIDE Framework: A Theory-Driven Approach to AI-Generated Feedback in Higher Education

Effective formative feedback remains difficult to scale and personalize in higher education, particularly for neurodiverse, multilingual, and otherwise diverse learners. Although generative AI offers promising automation capabilities, current applications often lack grounding in feedback research and insufficiently address inclusivity.This paper introduces a theory-driven approach to AI-mediated feedback that synthesizes established frameworks—including Medals and Missions, feedforward principles, WISE tone guidance, and reflection scaffolding—into a cohesive model optimized for large language model (LLM) implementation. The model is operationalized through a pilot web-based application that features configurable parameters for educational level, learner profiles, assignment context, and feedback preferences.The approach demonstrates how pedagogical rigor can be maintained while leveraging AI capabilities for personalization and scalability. This work contributes to research on AI-enhanced assessment by offering an evidence-based methodology that supports the delivery of equitable, inclusive, and actionable feedback at scale.