Developing an AI Chatbot Platform for Teaching and Research

This project aims to upgrade traditional written examinations by building a hybrid AI-assisted grading and feedback system that digitalises paper scripts and combines rule-based pattern matching with generative AI to improve grading efficiency, consistency, and pedagogical value, especially in large-enrolment contexts (the testbed is COMP1021 Introduction to Computer Science, 500–700 students). The project’s scope includes creating a secure repository for scanned exam papers, an AI-powered grading engine (with confidence scoring and manual review pathways), an instructor analytics module (e.g., question difficulty, common errors), and a student portal that delivers personalised, rubric-based feedback and supports reflection, revision, and regrade requests; overall objectives are to reduce grading workload, increase fairness and transparency, and reframe exams from purely summative ranking tools into feedback-rich, formative learning opportunities, with evaluation via baseline comparisons (manual grading), surveys, interviews/focus groups, and deployment across two semesters with potential scaling to other disciplines using written exams.