Calcyon: An AI-Based Chatbot Framework for Calculus Learning to Support Computational Thinking Skill and Self-Regulated Learning

Abstract

Abstract* Background The integration of artificial intelligence (AI) into mathematics education has advanced rapidly, yet limited attention has been given to systematic AI based chatbot frameworks that integrate computational thinking (CT) and self regulated learning (SRL) scaffolding within university calculus. This study developed and validated Calcyon, an AI based chatbot framework designed to support differential calculus learning through coordinated CT and SRL scaffolding. Methods A Research and Development approach guided by the ADDIE model was employed. Expert validation (n = 3) was conducted across five quality dimensions using Aiken's V formula. A student pilot readability and feasibility test (n = 11) assessed practical usability across six dimensions. A design and development research approach guided by the ADDIE model was employed. Three experts evaluated the framework across five quality dimensions using a four point rating scale, with content validity assessed using Aiken's V. A formative pilot involving 11 undergraduate calculus students evaluated readability and feasibility across six dimensions. Results Expert validation indicated that Calcyon demonstrated strong validity across the evaluated content, instructional, technical, linguistic, and media design dimensions. Most validation items met the predefined validity criterion, while a small number were refined based on expert feedback. The student pilot further indicated good readability and feasibility, with responses showing consistently positive evaluations of the prototype. Conclusions Calcyon provides an empirically validated instructional design framework that combines AI mediated interaction with explicit CT and SRL scaffolding for university calculus. The findings support the framework's expert rated validity and preliminary readability and feasibility. Further experimental research is required to determine its effectiveness in improving CT, SRL, and calculus learning outcomes.