ChatGPT as a direct formative assessment tool in stoichiometry: correlation with achievement and insights into pre-service science teachers’ perceptions

Abstract Generative artificial intelligence (Gen AI) has become increasingly important in education. However, its application for direct formative assessment represents an emergent area of inquiry. This mixed-methods study investigated the use of ChatGPT for formative assessment in a stoichiometry unit with 38 pre-service science teachers. We examined the relationships between ChatGPT assessment scores (scores produced when students prompted the AI to assess their own answers), final achievement scores, and instructor assessment scores. Spearman’s correlation revealed that ChatGPT assessment scores strongly correlated with final achievement scores ( r s = 0.670) and instructor assessment scores ( r s = 0.882). To explore the mechanisms underlying the achievement correlation, a qualitative analysis of interviews ( N = 13) explored pre-service science teachers’ perceptions. Findings indicated that ChatGPT was perceived as beneficial for providing immediate feedback during the learning process, pinpointing weaknesses requiring attention, and prompting adjustments in learning strategies – a key process in self-regulated learning – potentially explaining the link to achievement. These findings suggest ChatGPT’s potential as a supplementary formative assessment tool, suitable for practical integration into chemistry classrooms.