Ensuring Equity and Inclusivity in AI-powered Educational Assessment

A Case Study of Policy and Practice in Nigeria

Abstract

Artificial intelligence (AI) and computer-based testing (CBT) are increasingly central to educational assessment in Nigeria, promising efficiency, transparency, and fairness. However, concerns about equity and inclusivity persist, especially for rural learners and students with disabilities. This study critically examines how Nigeria’s education policies and practices address fairness in AI-enabled assessment, with particular attention to alignment with international ethical standards. A qualitative desk review was conducted, and policy documents, annual reports, and empirical studies published between 2015-2024 were analysed. Ribble Digital Equity Framework (2012) and UNESCO AI and Education: Guidance for Policymakers (2021) provided the analysis, focusing on access, participation, empowerment, and ethical use. As results reveal, the Nigerian policy texts express general commitments to digital access but do not provide specific mechanisms for the equitable integration of artificial intelligence. Students in rural areas are disproportionately marginalised by infrastructure shortages, such as low internet access and computer illiteracy. Students with disabilities, especially the visually impaired, experience systemic marginalisation because of the uneven distribution of assistive technology, resulting in poorer admission performance. Compared with UNESCO standards, Nigeria’s regulatory frameworks lack algorithmic transparency, non-discrimination, and inclusive design. The paper finds that, although Nigeria's assessment system remains primarily CBT-based, existing disparities in infrastructure, digital literacy, accessibility, and disability accommodation present significant risks for future AI-enabled assessment adoption. The findings suggest that, unless accompanied by deliberate safeguards relating to inclusive design, algorithmic accountability, accessibility, and digital capacity building, future AI-enabled assessment systems may reproduce or amplify existing educational inequalities rather than mitigate them.