JISARA

Abstract: The advancement of Artificial Intelligence (AI) has created numerous opportunities to enhance the self-learning experience. Automated quiz generation has attracted attention as a way to support self-learning, particularly in digital education environments. However, existing systems often face key limitations: they frequently lack references, fail to provide meaningful feedback, and do not adapt content to individual needs. These shortcomings hinder the systems' effectiveness in promoting deep understanding and engagement. This paper introduces QuizAI, a novel AI-powered quiz generation system designed to address these issues. QuizAI can generate multiple-choice questions (MCQs) and open-ended questions based on user-provided sources, including PDF files and web pages. By utilizing Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs), the system ensures that questions are grounded in their source material and accompanied by supportive feedback. While QuizAI demonstrates comparable latency in generating dynamic content, it has produced a 50% duplication rate in five test documents, particularly when processing a larger number of questions per request. Despite these limitations, QuizAI achieves moderate to strong performance in terms of response time and question diversity, effectively complementing existing systems in the field. The proposed system enhances learning outcomes, supports recollection, and helps reduce learners' anxiety in self-paced educational settings.

Download this article: JISARA - V19 N3 Page 68.pdf

Recommended Citation: Nguyen, T., Chung, S., (2026). AI-Powered Study Assistant for Exams: QuizAI. Journal of Information Systems Applied Research and Analytics 19(3) pp 68-81. https://doi.org/10.62273/FHRO6233