Rethinking teachers’ diagnostic skills in AI-supported formative assessment: from diagnosis to meta-diagnosis

This paper explores how integrating AI-supported formative assessment transforms the requirements for teachers’ diagnostic skills in adaptive teaching. AI-supported diagnostic tools broaden the scope of what teachers may attend to: in addition to monitoring and interpreting students’ learning processes, teachers need to consider AI-generated diagnostic inferences when making instructional decisions. Rather than simply increasing the amount of information available for diagnostic judgment, AI-supported assessment introduces a qualitatively different type of diagnostic evidence: AI-generated inferences based on student data. Contrastive case examples illustrate how this shift transforms diagnostic processes into a form of meta-diagnosis, in which teachers evaluate AI-generated inferences in relation to their own diagnostic judgments. Based on these insights, we present a conceptual perspective on teachers’ diagnostic skills in AI-supported settings. We argue that teachers’ diagnostic judgment increasingly involves (a) integrating different types of diagnostic evidence—teachers’ own diagnostic judgments of student learning and AI-generated diagnostic inferences—and (b) monitoring students’ learning processes as they interact with AI systems, adapting instruction accordingly. This conceptualization identifies meta-diagnosis as a key aspect of diagnostic thinking in AI-supported formative assessment. Finally, the paper offers directions for research on teachers’ diagnostic skills and their professional development in AI-supported formative assessment, emphasizing teachers’ continued centrality as pedagogical decision-makers.