Generative Artificial Intelligence in STEM Learning Assessment

A Technology Education Perspective

Abstract

In STEM education research, a primary concern among many researchers is the issue of assessment, particularly in evaluating interdisciplinary STEM competencies. For example, when assessing interdisciplinary problem-solving skills, there is a notable lack of appropriate assessment tools. STEM learning environments often involve complex problem designs, requiring students to deconstruct the context collaboratively during the problem-solving process. How to effectively apply knowledge from various STEM fields remains a challenge that existing assessment tools have not yet fully addressed. With the development of generative artificial intelligence, there are now more tools available to analyze students' discussions throughout their learning journey. The aim is to explore what do AI-based analyses of students’ group discussion discourse reveal about their interdisciplinary problem-solving processes in STEM projects. Besides, based on these findings, what are the potential limitations of using AI to assess students’ interdisciplinary problem-solving processes. By applying generative AI tools to analyze group discussion dialogues, the research aims to uncover critical areas for future focus in STEM education.