THE USE OF ARTIFICIAL INTELLIGENCE IN DEVELOPING RUBRICS FOR STUDENT ASSESSMENT IN HIGHER EDUCATION

Abstract

The rapid diffusion of generative artificial intelligence (GenAI) is reshaping assessment practices in higher education and is creating new opportunities for the design of transparent, scalable, and learning-oriented assessment instruments. This conceptual research article examines how artificial intelligence can support the development of analytic rubrics while preserving teacher agency and assessment validity. The study uses an integrative literature review and design-oriented conceptual analysis of research on scoring rubrics, AI-supported assessment, generative AI, and international guidance for responsible AI in education. Rather than treating AI as an autonomous grader, the article conceptualizes AI as a rubric-design assistant operating within a human-in-the-loop model. The proposed framework comprises seven stages: learning-outcome analysis, task analysis, AI-assisted criterion generation, descriptor generation, expert validation, pilot calibration, and iterative refinement. The analysis identifies four principal opportunities: reducing rubric-development workload, improving criterion coverage, facilitating differentiated descriptors, and supporting alignment between learning outcomes and assessment tasks. At the same time, the study identifies significant risks, including generic descriptors, construct underrepresentation, bias, over-reliance on AI-generated language, and inappropriate transfer from rubric generation to autonomous grading. Recent evidence suggests that supervised AI-rubric systems can achieve useful levels of consistency in particular contexts, while broader evidence on AI-based grading remains mixed and context dependent. The article therefore argues that the educational value of AI lies not in replacing professional assessment judgment but in augmenting teachers' capacity to design, test, and refine assessment instruments. The proposed framework provides a practical and theoretically grounded basis for responsible AI-assisted rubric development in higher education.