AI-based task development in teacher education: an empirical study on using ChatGPT to create complex multilingual tasks in the context of primary education

作者：Taha Ertuğrul Kuzu, Thomas Irion, Wolfgang Bay · 发表于：Education and Information Technologies · 年份：2025 · DOI：10.1007/s10639-025-13673-8 · 研究领域：Online Learning and Analytics、AI in Service Interactions、Intelligent Tutoring Systems and Adaptive Learning

Abstract Research on AI systems like ChatGPT indicates that there are not only obstacles but also potentials of using AI systems on the learner-level, e.g., for fostering learning processes. On the teacher-level, however, there is not much research yet and an important possibility to use AI systems like ChatGPT could be for, e.g., the development of complex tasks, which could help teachers in lesson planning processes. This study provides explorative and qualitative insights into such a task development process with the help of ChatGPT and for that purpose, n = 17 prospective teachers’ interactions with ChatGPT and post-interaction reflections were examined. They used ChatGPT with the assignment to develop complex tasks with the specific subject-matter parameters ‘Operative principle’, ‘Geoboard’ and ‘Multilingualism’. Methodologically, interpretative interaction analyses are conducted to gain abductive insights into the transcribed data and the analysis indicates that teachers, who were trained in the specific subject-matter as well as in prompt engineering, could use ChatGPT in a way that highly viable tasks could be generated whereas non-trained teachers mostly generated non- or less-viable tasks. This leads to the assumption of a threshold for the viable usage on teacher-level: Teachers who are subject-matter- and content-wise proficient and who know about effective prompting can use ChatGPT efficiently and evaluate it as a good tool for helping them in developing tasks t...