An innovative artificial intelligence (AI) model named QUADL, is set to revolutionize online course assessment by generating questions remarkably similar to those crafted by human instructors. QUADL’s unique ability to identify key concepts from instructional texts and subsequently generate relevant questions has been hailed as an effective tool in achieving course learning objectives. This new development heralds a potential shift in educational courseware creation, enhancing the pedagogical effectiveness and efficiency of online learning.
QUADL: Redefining Course Assessment
QUADL, a creation by researchers including Noboru Matsuda, an associate professor of computer science at North Carolina State University, is designed to perform two crucial tasks. Firstly, it analyzes instructional texts to pinpoint key terms and concepts. It then leverages this understanding to devise questions centered around these critical elements. The integration of courseware content and the curriculum’s learning objectives into QUADL allows it to generate questions aimed at fulfilling specific learning objectives.
Meeting the Challenge of Effective Question Formulation
The development of effective questions that assess student progress towards course objectives has been identified as a challenge for both instructors and courseware developers. QUADL’s capabilities, as suggested by the study, offer a practical solution to this issue, proving to be a beneficial tool for those responsible for course creation and instruction.
QUADL’s Performance: A Comparative Evaluation
To assess QUADL’s effectiveness, the team employed existing online courseware known as the Open Learning Initiative (OLI). Five instructors using the OLI were enlisted to evaluate a series of questions, some of which were developed by QUADL, others by the current leading question-generating AI model, Info-HCVAE, and some were already in use within the OLI courses. The instructors were kept unaware of the origin of the questions and asked to rate each on its pedagogical value. Interestingly, QUADL’s questions received scores mirroring those assigned to human-created questions, outperforming those generated by Info-HCVAE.
Future Prospects and the PASTEL Suite
Upcoming studies are set to integrate QUADL’s generated questions into undergraduate classrooms to gauge their impact on student learning. This effort is an essential step to validate QUADL’s potential in real-world settings.
QUADL is part of a larger collection of AI technologies, PASTEL, being developed by Matsuda and his team. PASTEL aims to streamline educational courseware development, tackling various aspects from question generation to quality assurance of courseware effectiveness.
The research findings will be presented at the 24th International Conference on Artificial Intelligence in Education (AIED 2023), with the researchers actively seeking research and educational partners for the development and application of these generative AI technologies.