Development and Evaluation of an AI-Powered Adaptive Learning Platform for Introductory Programming Education

Introductory computer science courses face persistently high failure rates, with traditional pedagogical approaches unable to accommodate the diverse learning needs of undergraduate students. While intelligent tutoring systems have been developed to address this challenge, most lack empirical evidence of effectiveness and remain constrained to specific institutional contexts. This paper presents the design, implementation, and preliminary empirical evaluation of an AI-powered adaptive learning platform built with Django and the OpenAI API for introductory Python programming education. The platform integrates a 14-chapter mastery-based curriculum with a 5-gate progression system requiring 80% proficiency thresholds, supported by a comprehensive suite of large language model-driven tools including a debug lab, code review engine, and context-aware study assistant. An empirical pilot study was conducted using this platform with CS I students at a state university in the southeastern United States. Although the small sample size limited the statistical power of the formal hypothesis tests, the observed effect sizes were medium-to-large for quiz scores (d = 0.63) and large for open-ended Q&A items (d = 1.02), and structural equation modeling of survey data revealed theoretically coherent relationships among perceived usefulness, self-efficacy, engagement, learning effectiveness, and satisfaction. These findings provide preliminary evidence that the platform is a promising approach and that generative AI-based adaptive learning systems can meaningfully enhance programming education outcomes, while the platform’s free, open-access design offers a model for democratizing AI-augmented instruction.