Team Members: Steph Buongiorno
When learning materials are accessible, students with reading disabilities can effectively demonstrate their understanding in ways comparable to their peers. AI agents, powered by large language models (LLMs), present a unique solution for providing universal access support to students with reading disabilities by offering personalized learning experiences tailored to individual needs. This personalization enhances learning without compromising the generalizability of the AI agents’ design, allowing a single agent design to adapt to diverse scenarios.
Our research aims to advance AI agent design and deploy these agents in middle school classrooms to promote equity and better prepare students for technology-driven careers. These AI agents will also support teachers by offering data-driven insights into student progress, making it easier to track and address diverse learning needs. This research integrates game-based AI agents–driven by LLMs–into an educational version of Minecraft to personalize learning experiences, discover optimal metrics for supporting students with reading disabilities, and provide teachers with summaries and analytics on student progress.
Our work is divided into two agent types: Personalized Education Agents and Teacher Support Agents. Integrated into Minecraft, these agents assist students through personalized feedback and game scenarios tailored to their reading needs, while helping teachers understand student progress through accessible summaries and visualizations.