The shifting landscape of assessment in STEM education in the age of generative AI
TL;DR
A review of recent relevant studies identifies three emerging themes that scholars may further explore in this article collection: how AI has become entangled in assessment systems, including its effects on the validity of assessment evidence, the redesign of assessment practices, and institutional policies concerning students' and instructors’ understandings of acceptable AI use.
Abstract
This article focuses on the impact of generative AI (GenAI) on assessment in STEM education. Guided by the assessment triangle as a conceptual framework, we take an ecosystems perspective to review the extant STEM education literature and identify how GenAI affects what is assessed (cognition), how evidence of learning is elicited (observation), and the inferences drawn from that evidence (interpretation). Our review of recent relevant studies identifies three emerging themes that scholars may further explore in this article collection: (1) how AI has become entangled in assessment systems, including its effects on the validity of assessment evidence, the redesign of assessment practices, and institutional policies concerning students’ and instructors’ understandings of acceptable AI use; (2) how AI is being used to analyze assessment evidence and provide feedback; and (3) the need for AI-generated instructional materials to be evaluated and validated for use in learning and assessment. We conclude by broadly discussing potential directions for future research.