AI-Driven Assessment

Abstract

Generative Artificial Intelligence (AI) is transforming higher education assessment yet challenging the validity and authenticity of assessment evidence. This chapter argues that the central concern extends beyond academic dishonesty to whether assessment evidence genuinely reflects students' learning, reasoning, and demonstrated competence. Drawing on Messick's unified theory of validity and principles of authentic assessment, the chapter examines how AI may introduce construct-irrelevant variance and weaken the authenticity of assessment evidence. It proposes a framework for preserving valid and authentic assessment evidence through process-based assessment, supervised assessment, multimodal and authentic assessment, and ethical AI-literate engagement. The chapter concludes that sustainable assessment in the AI era requires higher education to move beyond AI detection and prohibition towards assessment practices that generate trustworthy evidence of student learning and competence.