Into the Zone: Evaluating AI-Augmented Personalized Learning Through a ZPD-Aligned Metric

• In Event: AERA Poster Session Friday 7:45 am
  In Poster Session: Learning Sciences Poster Session

Abstract

This study examines how artificial intelligence (AI) can help operationalize Vygotsky’s Zone of Proximal Development (ZPD) in adaptive learning systems. Although ZPD has been a cornerstone of educational theory for decades, translating this dynamic, socially mediated vision of learning into scalable practice remains challenging. We introduce time to first fail as a process-oriented metric for evaluating how quickly systems identify a learner’s optimal challenge level. Comparing rule-based versus AI-augmented recommendation engines, we found that LLM-enhanced personalization reduced time to first fail from 30.9 to 11.5 minutes while maintaining engagement. This suggests that AI can help realize responsible, developmentally appropriate instruction at scale. By bridging foundational learning theory with emerging technology, this work contributes to constructing more equitable educational futures.

Authors

• Amanda Siebert-Evenstone, Age of Learning

• Hee Jin Bang, Age of Learning