AdaPT uses LLMs to adapt lesson plans for diverse classrooms

Researchers from academia have introduced AdaPT (Adaptive Lesson Plan Transformer), a system that leverages large language models to help teachers quickly adapt existing lesson plans for different regional contexts and diverse student profiles. Instead of generating content from scratch, AdaPT takes an existing lesson plan and applies transformations—adjusting language, examples, difficulty levels, and pacing—while providing transparent explanations for each change. The system features an interactive interface where teachers can input student profiles, view a structured representation of the lesson, and iteratively refine the output in a teacher-in-the-loop workflow. This approach addresses a critical gap in educational technology: most tools focus on generating new content, but teachers often need to modify existing plans to fit new classrooms.

The team evaluated AdaPT through a user study with 9 teachers and an expert evaluation with 3 education specialists. Results indicated that AdaPT supports teachers' natural workflows by reducing the cognitive load of manual adaptation while maintaining instructional quality. Teachers reported that the system's explanations helped them understand and trust the transformations, and the iterative refinement allowed them to retain control over the final lesson. The study highlights AdaPT's potential to promote educational equity by enabling high-quality lesson plans to be quickly repurposed for underserved regions or differentiated instruction—without requiring teachers to start from zero. The paper is available on arXiv as a preprint.