The Hidden Inequality of “Free” AI: Why Skills, Not Access, Drive AI Adoption in Higher Education

This study provides the first empirical test of whether van Dijk's (2005) sequential access model transforms into skills-mediated pathways in cloud-based AI environments where physical access barriers are neutralized. Surveying 586 post-secondary students using hierarchical OLS regression, multinomial logistic regression, and bootstrap mediation, we found skills access dominated AI usage intensity (β = 0.626, p < .001) while physical access became nonsignificant (β = 0.049, n.s.). Skills mediated 54–72% of the relationship between access factors and usage, transforming sequential pathways into skills-mediated ones. We introduce algorithmic capital — cumulative AI competencies that replace infrastructure as the determinant of equitable adoption.