Beyond AI Adoption: A Responsible and Resource-Aware Framework for AI Integration in Global South Higher Education

Artificial intelligence (AI) is rapidly reshaping higher education through applications in teaching, assessment, research, student support, and institutional decision-making. However, AI integration strategies are often shaped by assumptions of reliable connectivity, sufficient computing infrastructure, financial capacity, technical expertise, and established governance mechanisms. These assumptions may not adequately reflect the operational realities of many higher education institutions in resource-constrained settings. Evidence from Nigerian higher education points to persistent challenges involving digital infrastructure, connectivity, technical capacity, institutional support, affordability, and responsible-use concerns. Consequently, expanding access to AI tools alone may not translate into sustainable or equitable AI integration. This study introduces an emerging Responsible and Resource-Aware AI Integration Framework (RRAIF) for higher education. Grounded in the Nigerian tertiary education context, the framework is designed to support AI integration in resource-constrained settings across the Global South. RRAIF is built around five key dimensions: resource readiness, responsible governance, accessibility and inclusion, local relevance, and institutional capacity. Together, these dimensions consider whether institutions have the infrastructure, skills, policies, safeguards, and local conditions required to integrate AI responsibly and sustainably. Rather than treating AI adoption as a binary choice, RRAIF proposes four context-sensitive pathways Adopt, Adapt, Restrict, or Defer based on the interaction between institutional readiness, expected educational value, resource requirements, and responsible-AI risks. For example, an institution may adopt a low-resource AI application with strong governance safeguards, adapt a computationally demanding system to local infrastructure, restrict applications involving sensitive student decisions, or defer deployment where minimum readiness conditions are unmet. The framework will be empirically refined through a multi-institutional study involving academic, administrative, and technology stakeholders across Nigerian universities and polytechnics. The study will examine AI usage, infrastructure readiness, institutional support, governance, accessibility, local relevance, and perceived barriers to sustainable adoption. By treating resource constraints as design parameters rather than merely adoption barriers, this work aims to contribute a context-sensitive approach to responsible AI integration and inform more inclusive and sustainable AI ecosystems in higher education across the Global South.