Scaling Assessment Innovation: From Immediate Feedback MCQs to Automated Question Generation with LLMs

Abstract

Managing assessment in large, multi-campus computing cohorts is challenging, especially with reduced contact hours and growing demand for distance learning. We describe a trial of a modified Immediate Feedback Assessment Technique (IFAT), combining automatically marked multiple-choice questions (MCQs) with peer review. This improved consistency, supported professional-style learning, and was well received by staff and students, but required extensive effort to author hundreds of MCQs. We present a classroom trial with a first student cohort, demonstrating technical feasibility and promising engagement. Subsequent reflection on how LLM-Assisted assessment design can make innovative practices like IFAT more sustainable and scalable in computing education is also introduced. We then extend this approach with a novel open-source command-line tool that leverages large language models (LLMs) to automatically generate Moodle-ready GIFT files directly from course materials such as PDFs or Powerpoint slides; including images. This approach was trialled with a second student cohort, though outcome data here is still emerging.