GenAIprime: Generative AI for Remediation-Priming of Knowledge Gaps by Dynamically Creating Multimedia Learning Resources

Contemporary learners routinely acquire knowledge asynchronously from sources outside of formal learning settings, thus increasing the demand for instructional technologies to address the learners’ various knowledge gaps within STEM fields. Whereas post-assessment remediation is a high payoff target for such interventions, we investigate invoking Generative AI (GenAI) to ‘prime the pump’ for learners with any needed terminology, concepts, and technical-leveling prior to conventional human-delivered remediation. We have developed the GenAIprime framework that invokes GenAI Large Language Models (LLMs) to create personalized, multimedia-based remedial materials by prompting GenAI with the specific incorrect responses to each student’s digitized quiz/exam submission. By integrating with the Canvas Learning Management System (LMS), GenAIprime analyzes assessment content and extracts context using the OpenAI GPT 3.5 Turbo model. In addition to high relevancy, it also demonstrates a 2.15-fold execution speedup over the local Llama 3.1 model for more feasible extension to institutional-scale delivery. The platform offers students a detailed performance overview and targeted video recommendations to address misconceptions, while instructors can monitor progress, refine AI-driven suggestions, and provide contextual feedback. The potential for GenAIprime to enhance learning outcomes by focusing learners’ time and attention is described herein as a novel instructional technology that can be delivered and scaled up with minimal burden to instructional staff.