AI-Powered Lesson Planning: A Web-Application for Interdisciplinary Teaching Using the 5E Model

Abstract

This study examines the technical and pedagogical performance of an AI-powered web application in assisting teachers planning their lessons. The application uses artificial intelligence and structured to meet research-based educational frameworks. It integrates constructivist principles, the 5E instructional model (Engage, Explore, Explain, Elaborate, Evaluate), Bloom’s taxonomy, and interdisciplinary design to generate coherent, standards-based lesson plans. In addition, the application integrates a UDL framework as a scalability feature. Unlike generic generative tools such as Chatgpt and others, the application embeds pedagogical constraints to ensure generated lessons are pedagogically sound and consistent over all parts from aligning standards to setting objectives and designing activities and assessments. The application was designed and evaluated through systematic analysis of its generated instructional products rather than user perceptions or student outcomes. A total of twenty-four lessons constituted the primary dataset, including eight AI-generated lessons using the 5E instructional model, eight AI-generated lessons using the UDL framework to examine framework consistency and fidelity, and eight human-authored lessons. In addition, one lesson generated using a generic AI system (ChatGPT) was produced as a baseline comparison and was not included in the primary dataset. Across three evaluators, the researcher, Copilot, and DeepSeek, AI generated lessons aligned with the 5E instructional model demonstrated consistently high compliance, scoring within a narrow range of approximately 90% to 98%. AI generated lessons aligned with the UDL framework showed similarly stable performance, ranging from approximately 94% to 97%, indicating strong framework consistency and scalability. Human authored lessons exhibited greater variability, with scores ranging from approximately 70% to 100%, reflecting strong pedagogical creativity and contextual adaptation alongside differences in explicit structural alignment. In contrast, a baseline lesson generated using a generic AI system (ChatGPT) achieved a substantially lower score around 47 %, highlighting limitations in instructional completeness, assessment architecture, and framework fidelity when compared with framework anchored application outputs. Findings reflect that enforcing instructional design frameworks into AI generation logic enhances structural completeness, reliability, and pedagogical coherence. At the same time, human-authored lessons demonstrated creativity and superior contextual adaptation. The results support the model of human-centered AI in which generative systems function as scaffolds rather than replacing teachers’ expertise and creativity. By doing so, AI can help reduce the workload of teachers while still allowing them to have control over the lessons and make any necessary changes. This approach can also help amplify the impact of human intelligence in education, making it a valuable tool for teachers and students alike.