Enhancing Academic Performance with Generative AI-Based Quiz Platform

Abstract

Course examinations are among the most time-consuming instructional tasks in undergraduate teaching. We build and evaluate a generative-AI quiz platform that lets instructors upload course materials (PDFs, slides) and automatically synthesizes questions targeting multiple Bloom cognitive levels. Across two undergraduate courses we observe improved learning outcomes and substantially reduced instructor preparation time.

Problem & Motivation

Designing high-quality assessments that cover course objectives across multiple cognitive levels is one of the most time-consuming responsibilities for university instructors. Each curriculum revision forces a rebuild from scratch, squeezing the time available for actual teaching interactions and student support. Traditional item-writing also fails to adapt quickly to evolving learner progress, leaving a persistent gap between assessment items and the live state of instruction.

Method

We developed a generative-AI quiz platform that ingests instructor-provided materials (PDFs, slides) and produces questions aligned to declared learning objectives. The system was evaluated in two undergraduate courses — Python Programming (programming-task generation) and Introduction to Artificial Intelligence (cross-domain knowledge). We compared pre/post learning outcomes and collected student feedback questionnaires to assess the platform's impact on academic performance.

Findings

• Learning performance improved significantly after introducing the AI quiz platform.
• Generated items spanned multiple Bloom cognitive levels — from Remember through Analyze.
• Instructor item-writing time was substantially reduced, freeing time for interactive teaching.
• Students reported that immediate, on-demand quizzes helped them verify their own understanding and increased motivation.

Implications

Generative AI changes not only how questions are written, but how assessment is positioned in the course loop. Instructors spend less time producing items and more time analyzing student responses to surface misconceptions and adapt instruction. The combination of automated item generation and rapid feedback shows particular promise for programming and technology-oriented courses.

Citation

@inproceedings{chang2024ai_quiz,
  author    = {Chia-Kai Chang and Lung-Chun-Ta Chien},
  title     = {Enhancing Academic Performance with Generative {AI}-Based Quiz Platform},
  booktitle = {Proc. IEEE Int. Conf. on Advanced Learning Technologies (ICALT)},
  pages     = {193--195},
  year      = {2024},
  month     = jul,
  doi       = {10.1109/ICALT61570.2024.00062},
}