Developing a Custom GPT for NGSS-PLUS-5E Lesson Planning in Elementary Science Education

Lesson planning is a critical yet time-intensive component of science teaching, especially when teachers must align instruction with the Next Generation Science Standards (NGSS) and the 5E instructional model. This study provides a practical guide for developing a custom Generative Pre-trained Transformer (GPT) to support NGSS-PLUS-5E lesson planning in elementary science education. The dataset included 264 qualified elementary life science lesson plans collected from preservice elementary teachers across five academic semesters, along with a standardized lesson plan template and an evaluation rubric. The custom GPT development process involved three phases: data preparation, model configuration and testing through prompt engineering, and user implementation. To examine the quality of the generated outputs, three lesson plans were produced and evaluated using the NGSS-PLUS-5E rubric by two independent raters. Results provided initial proof-of-concept evidence that the generated lesson plans received high scores across four dimensions: organization and quality, NGSS alignment, 5E model alignment, and instructional practices, although some limitations were identified in the depth of elaboration and the variety of instructional approaches. The findings suggest that a custom GPT can serve as a useful co-planning tool for generating structurally complete, standards-aligned lesson plans. However, teacher review remains essential to enrich content and adapt lessons to specific classroom contexts effectively.