FROM PERMISSION TO RESPONSIBILITY

Abstract

Generative artificial intelligence (GenAI) has disrupted conventional university assessment, leading many institutions to oscillate between blanket prohibition and unrestricted use. Existing models, such as the Artificial Intelligence Assessment Scale, clarify different levels of acceptable AI involvement, while policy and ethics frameworks emphasise governance, training, privacy, fairness, transparency, and accountability. However, these strands are often treated separately, leaving students and educators without practical guidance that connects how much AI may be used with who is responsible and which ethical safeguards are required. This conceptual paper proposes a human-centred Ethical Responsibility Matrix for GenAI-supported university assessment. Drawing on a structured conceptual review and framework synthesis of recent peer-reviewed literature on AI assessment, AI literacy, academic integrity, and educational governance, the study integrates three dimensions: levels of AI involvement; the responsibilities of students, educators, and institutions; and ethical safeguards appropriate to each level. The matrix differentiates between no-AI assessment, AI-assisted planning, AI-supported feedback, human–AI collaboration, and AI exploration or full integration. At each level, it specifies requirements concerning disclosure and documentation, critical verification of AI outputs, data privacy, equitable access, and final human accountability. The proposed framework shifts the debate from detecting misconduct towards designing valid, transparent, and learning-centred assessments. It is intended as an internationally adaptable decision-making tool rather than a single universal policy. By aligning permitted AI use with explicit responsibilities and safeguards, the paper seeks to preserve academic integrity and student agency while enabling the pedagogically meaningful use of GenAI in higher education.