ForumAId: Enhancing Student-Teacher Interaction through GenAI in LMS Environments

<p dir="ltr">The adoption of generative artificial intelligence (GenAI) in educational technology is rapidly redefining the landscape of teaching and learning. This paper introduces an innovative application, ForumAId, designed to integrate with Learning Management Systems (LMS) via Learning Tools Interoperability (LTI) 1.3 standards. The application leverages large language models (LLMs) to enhance student-teacher interactions by providing advanced tools for summarizing student comments and discussions. This functionality empowers educators to efficiently extract key insights from large volumes of student-generated content, reducing workload and enabling more focused pedagogical interventions. Furthermore, the application allows instructors to input custom prompts, facilitating targeted analyses of discussions to identify specific areas of concern or interest, such as misconceptions, engagement gaps, or topics requiring further elaboration. Traditional LMS platforms provide discussion forums as a medium for student engagement, but they lack built-in mechanisms for efficiently processing and summarizing large-scale discussions. Instructors often face significant challenges in manually reviewing extensive student interactions, making it difficult to extract meaningful insights or identify key discussion trends in a timely manner. This limitation leads to delayed feedback, reduced instructional efficiency, and an increased cognitive load on educators. Additionally, identifying patterns across discussions such as common misunderstandings or emerging topics of interest requires substantial manual effort, limiting the overall effectiveness of forum-based learning activities. A key feature of ForumAId is its ability to generate concise, context-aware summaries of student discussions while maintaining their original intent and meaning. Instructors can also specify a date range for posts to generate summaries, enabling focused analysis of discussions over a particular period. By allowing instructors to dynamically interact with the data through customizable prompts, the tool supports personalized teaching strategies that adapt to student needs. This fosters a more responsive and iterative teaching process, leading to improved learning experiences. The integration of this application within existing LMS frameworks ensures seamless adoption, making it accessible across diverse educational settings. The technical design of ForumAId emphasizes interoperability through LTI 1.3 standards and leverages state-of-the-art LLMs for natural language processing. A pilot study conducted with instructors demonstrates the system's potential to improve teaching outcomes and student engagement. Initial feedback highlights the tool's ability to streamline workflows, reduce cognitive load, and provide actionable insights into student learning behaviors. The application also addresses challenges such as ensuring data privacy, minimizing bias in AI-generated outputs through advanced prompt engineering techniques, and optimizing computational efficiency using LLM-based summarization. By tackling these challenges, the application aligns with broader goals of equitable and responsible AI deployment in education. This work shares our initial experience with ForumAId and explores its implications for improving student-teacher interaction. Future directions include incorporating Retrieval-Augmented Generation (RAG) to append subject-specific knowledge and provide real-time AI-generated feedback to student comments, further enhancing educational outcomes.</p>