Teacher Ratings of Machine-Generated Instructional Scaffolds for K12 Math Classrooms

This study investigates the potential of Large Language Models (LLMs) in generating instructional scaffolds comparable to human-crafted materials, focusing on the development of warmup tasks that review and activate prior knowledge. Addressing the challenges educators face in scaffolding instruction - primarily due to time constraints and a lack of appropriate tools - this research outlines a co-design process to create such models and evaluates their quality through human assessment. With the increased need for tailored, differentiated instruction in the post-pandemic era, this work aims to contribute a process and early results for improving the pedagogical rigor of LLMs.