University of Phoenix Proposes 16-Stage AI Model for Online Education

As universities scramble to adopt AI-enabled learning tools, implementations often treat predictive analytics and generative AI as siloed technologies. The new framework seeks to bridge this gap by positioning these tools within an adaptive socio-technical ecosystem. By combining systems theory with the learning analytics cycle, the authors outline a process where data ingestion, predictive modeling, and generative feedback function as a continuous loop under ethical governance.

"AI in education should begin with the learner experience," said Darbyshire. The model relies on predictive signals—such as declining performance or late submissions—to identify students in need of assistance. Once a risk tier is identified, generative AI can suggest personalized study plans or formative quizzes. Critically, the framework mandates that these outputs undergo instructor review, ensuring that technology serves to augment, rather than replace, the human relationship in the classroom. This approach emphasizes transparency, bias monitoring, and the preservation of faculty discretion as core components of a responsible digital learning environment.