Abstract GIFTS: Automated Quiz Generation Using Generative AI and QTI for Teaching Content Management Systems The rapid advancements in artificial intelligence (AI) are reshaping numerous fields, with education being a key beneficiary. Generative AI, known for its capability in reasoning and content creation, presents an innovative approach to enhancing digital learning environments. This study explores an automated methodology for generating quiz questions using AI-powered tools in conjunction with the Question and Test Interoperability (QTI) format. By integrating Generative AI with widely used teaching content management systems such as Canvas and Moodle, we demonstrate how AI-driven automation streamlines the creation of diverse assessment formats, including multiple-choice, fill-in-the-blank, and numeric/text input questions. To implement this approach, we first define structured input formats that guide the AI in generating quiz questions. We employ prompt engineering techniques to instruct the AI on the desired question complexity, format, and alignment with learning objectives. More specifically our approach is to feed a generative AI system the QTI format that we need output and feed it our learning goals in some instances or sample questions in others and ask it to generate (in QTI format) other questions. Then we review these questions for accuracy. We also explore how different prompting may affect the quality of the generated questions. For example, what if our prompt includes a website or pdf of evidence based guides for creating test questions such as Writing Test Items To Evaluate Higher Order Thinking by Thomas Haladyna. We explore whether this prompting strategy creates questions that better assess student learning. Our approach significantly reduces the time and effort required for educators to design assessments while ensuring variability, adaptability, and quality in quiz content. The system enables dynamic question generation, allowing for real-time adjustments to difficulty levels, and question structures. Additionally, the use of standardized QTI format facilitates seamless integration across different learning management platforms, ensuring interoperability and broad applicability. By leveraging the capabilities of AI, this method enhances both formative and summative assessment strategies, improving the overall efficiency of teaching, testing, and student evaluation processes. This efficiency leaves more time for the human to human interaction that is so essential to student motivation and learning. This work highlights the transformative potential of AI-powered quiz generation in educational technology. Utilizing this technology can pave the way for future work. One example is scaling this technique with different prompting strategies such as prompting AI with a full syllabus. Another is using the data from these assessments to automatically generate follow up questions that adapt to that particular student. The foundational techniques described in this paper are key to these long term objectives.
Sayginer et al. (Thu,) studied this question.