Using LLMs to support teacher reflections on using questions to deepen learning and promote student engagement

Abstract

While much attention has been focused on improving student outcomes, there is growing interest in supporting teachers' reflection on their teaching practice using Generative Artificial Intelligence (GenAI) technologies. This paper examines the application of Large Language Models (LLMs) such as GPT-4 to automate the analysis of secondary school teachers' teaching practice in Singapore, specifically within the context of one of the teaching areas identified by the Singapore Teaching Practice model: using questions to deepen learning. We aimed to demonstrate the effectiveness of LLMs in analyzing classroom lessons in this teaching area. The methodologies employed in this study included the collection of classroom data and their analysis, both manually and using LLMs. Specifically, this involved transcribing the classroom lessons and analyzing each question using LLMs, with the results compared to a ground-truth dataset created through manual analysis. The findings suggest that LLMs are effective in providing, forming the basis for future teacher reflection and the potential for automated self-reflection tools in Singapore schools.