Abstract:

The emergence of publicly available generative artificial intelligence (genAI) in late 2022 has significant implications for education. Despite deficit views of AI as providing increased opportunities for students to engage in cheating and plagiarism, evidence is emerging that AI has the potential to assist and enhance teaching practice within all levels of education. This study is concerned with primary school mathematics education, a discipline that is typically lagging in its use of digital resources when compared to other disciplines such as language and science. Although it is widely agreed that digital resources have the potential to disrupt traditional teaching approaches, it is also widely acknowledged that there is a consistent lag between the emergence of new digital technologies and research that explores their implications for classroom practice and student learning.

Although research about the use of genAI tools in education has begun to emerge, there are clear gaps. For example, general literature that explores the challenges and potential of genAI for education are not specific to a particular level of education, and literature that specifically refers to mathematics education tends to focus on high school or tertiary education. There is a clear gap in current, empirically informed literature that addresses the practicalities and impacts of using genAI tools as a quality resource for primary mathematics educators.

This poster presentation will present emerging findings from a qualitative multiple case study conducted in nine Australian schools. The study explores how primary teachers use genAI to develop high-quality rich tasks along with associated resources such as assessment rubrics, reflection prompts, differentiation ideas, and supports for conducting explicit teaching within and around the task. Data were collected from 34 teachers who, in teams, designed three tasks using genAI. Lesson observations were conducted while the tasks were implemented, and student focus groups were held immediately following each lesson. The participating teachers were also interviewed prior to and immediately following the study. Transcripts of the task design prompts and AI responses were also analysed.

Data analysis was conducted using three theoretical frameworks: The Technology Acceptance Model (TAM), TPACK, and the Mathematical Knowledge for Teaching (MKT) model. Initial findings indicate the participant teachers were positve in relation to the perceived usefulness (PU) of genAI and all intended to continue using it to different degrees. It was evident the depth of MKT teachers draw on to plan and implement AI generated tasks varies greatly according to teacher expertise and is not necessarily related to their length of experience in the classroom. The level of teacher MKT is also a determinant for the quality of the designed task and associated resources. Wide variation in the implementation and adaptation of tasks were observed across classrooms and levels of student engagement also varied although tended to increase due to the contextualised nature of the designed tasks.

Implications of the findings are the need for teacher professional learning about prompt crafting, a need for teachers to improve understandings of how generative AI works, and continued development of teacher MKT. This study also has implications for resource developers as generative AI has the potential to diminish reliance on commercial resources.

Keywords:

AI, Technology, Mathematics Education, Generative AI.