Rethinking Student Assignments in Higher Education in the Age of Generative AI: From Product Assessment to Process-Oriented Learning

Abstract

Generative artificial intelligence (GenAI) can contribute to idea generation, content organisation, language editing, and product revision, making it difficult for teachers to judge students’ knowledge and understanding, cognitive engagement, and independent contribution solely from the final assignment. This study employed a focused integrative literature review and conducted structured extraction, source-type appraisal, thematic coding, and cross-text comparison across 37 materials: 24 GenAI-related studies, eight classic works on assessment theory, and five policy and guidance documents. The findings indicate that final products remain valuable for assessing students’ overall performance, but their explanatory power as a single source of learning evidence has diminished under conditions of human-AI co-production. Drawing on the literature synthesis, this article proposes a six-stage process-oriented assignment framework comprising task interpretation, independent attempt, human-AI interaction, information verification, outcome revision, and reflective disclosure. The framework is guided by alignment with learning outcomes, student agency, visibility of cognitive engagement, transparency and responsibility, and proportionality. It links sustainable assessment with long-term evaluative capability, responsible AI use, equitable participation, and manageable evidence requirements. As a literature-informed conceptual proposal, its effectiveness, scope of application, and implementation burden require examination across disciplines and course contexts.