A higher-education ‘AI integrity risk ecology’ model grounded in assessment ethics

The paper introduces the Higher-Education AI Integrity Risk Ecology (HE-AIRE) model, which aims to address ethical, validity, and governance issues related to AI-powered assessment decisions. With the integration of tools like generative AI, auto-grading systems, feedback mechanisms, proctoring software, and AI detectors into higher education, assessment decisions are becoming increasingly algorithm-dependent, which can have ethical implications for fairness, transparency, accountability, and trustworthiness. Through an interdisciplinary framework encompassing assessment ethics, educational measurement, socio-technical systems, risk ecology, and global AI governance, the paper posits that AI-driven integrity risks are not exclusively technological or personal. These risks are, instead, a result of a socio-technical phenomenon involving teaching strategies, system architecture, institutional policies, and cultural dynamics. The HE-AIRE model identifies the major risk categories, such as validity, bias/equity, opacity, data/surveillance, pedagogical agency, accountability, and socio-cultural marginalisation, and arranges them in a four-tiered ecosystem, namely, pedagogy, technology/ethics, institutional operations, and socio-cultural environment.