AI-Augmented Educational Assessment: Enhancing Feasibility, Explainability and Integrity

This thesis explores the feasibility, explainability and integrity of AI-powered automated assistance in educators’ assessment of student responses. It presents a workflow for investigating feasible AI-powered assessment, illustrates the alignment in human and AI-powered assessment, identifies effective AI-powered explanatory insights to augment educators’ assessment practices, and examines the capability of Generative AI in producing high-quality, human-like responses to assessment tasks.