An Adaptive Framework for Human-AI Collaboration in Generating Trustworthy Educational Feedback

This project addresses the challenge of trust in AI-generated educational feedback by developing a system that dynamically adapts to subject context and user needs. It integrates a novel AI trust calibration framework to align system behaviour with student and educator expectations. Through controlled experiments, surveys, and iterative design, the project aims to enhance the credibility and emotional safety of AI feedback, particularly for non-native English speakers. The outcomes include academic publications, an open-source plugin for educators, and broader applications across industries, supporting TDS’s mission to foster trustworthy digital content and community empowerment.