The rapid integration of Artificial Intelligence (AI) into instructional design has shifted the paradigm from static content delivery to dynamic, co-creative partnerships. This research critically analyzes the efficacy of AI-based instructional design models by synthesizing traditional frameworks, such as ADDIE (Analysis, Design, Development, Implementation, and Evaluation) and TPACK with emerging AI-Agent literacies. Through a multi-phase evaluation of an AI-Educational Development Loop (AI-EDL), we examined the impact of AI-generated instructional materials on student learning outcomes and teacher workload. Results demonstrate that while AI enhances scalability and personalization, its efficacy is contingent upon rigorous “Human-in-the-Loop” (HITL) oversight. We propose the AIA-PCEK framework (Artificial Intelligence Agent – Pedagogical Content Ethical Knowledge) as a standard for future instructional design, emphasizing that the synthesis of tech and teaching must prioritize pedagogical intentionality over algorithmic automation.