Abstract
In an era of rapid digitalisation and artificial intelligence (AI) diffusion, higher education faces a critical challenge: how can institutions ensure that AI-enabled innovation supports inclusive learning rather than amplifying inequality? While existing research highlights AI’s pedagogical affordances and ethical risks, less attention has been paid to the structural mechanisms through which institutional conditions shape divergent student outcomes. This developmental paper advances a structural conceptual model explaining how institutional AI capacity influences students’ digital capital, which in turn shapes trust calibration patterns in AI use, subsequently producing differentiated critical thinking outcomes and perceptions of pedagogical inequality. The model builds on Phase 1 qualitative findings from focus groups (n = 35) and interviews (n = 20) conducted across public and private universities in Mauritius, a small island developing state (SIDS) characterised by institutional asymmetries in digital infrastructure and policy maturity. The qualitative phase revealed institutional disparities in AI infrastructure and training, uneven AI literacy, and contrasting reliance patterns ranging from calibrated engagement to overreliance. Integrating digital capital theory with trust calibration research, the proposed model conceptualises AI-enabled pedagogical inequality as a structurally mediated process rather than a simple access gap. A subsequent quantitative phase will test the model using SmartPLS structural equation modelling. By linking institutional innovation capacity to inclusion outcomes, this paper contributes a theoretically grounded and empirically testable framework for understanding how higher education institutions can foster more inclusive AI-enabled learning environments.