Pedagogy 2025

Abstract

Large language models (LLMs) like ChatGPT have great potential as educational tools. They can strengthen teacher efficiency and efficacy in lesson planning. Still, tools such as ChatGPT do not provide contextual information relating to specific children or nuanced cultural perspectives. So, even when supported by generative AI, teachers need to augment materials and choose concepts applicable to their unique environment to develop culturally sustaining lesson plans. Teacher training programs play a critical role in helping student-teachers develop the skills to refine ChatGPT outputs. Thus, as a teacher-educator at a university in Singapore, I asked the question: How can I offer assignments that aid preschool teachers in collaborating with ChatGPT on culturally sustaining lesson plans? To respond to the research question, I analysed data from 59 undergraduates enrolled in an early childhood education elective, ECE370 Introduction to Edible Education for Children. As part of their course assessment, students were asked to prompt ChatGPT to generate a lesson plan, then adapt that plan to the needs of their context, culture and community. Through review of students’ ChatGPT-generated lesson plans and associated documents, I discovered that to design helpful assignments, I should begin (a) including definitions and examples; (b) providing guidance on both prompts and lesson plans; and (c) focusing more on conceptual than observable changes. The wider implications are that student-teachers have the potential to be highly effective in adapting and contextualizing lesson plans for the populations they work with. To develop this skill, all they need is assignments providing appropriate support.