Abstract

The rapid adoption of generative Artificial Intelligence (GAI) tools such as ChatGPT has led to the widespread assumption that simply granting access to AI promotes educational equity and deeper learning. However, initial observational findings indicate that students and teachers experience unequal cognitive benefit from AI—even when using identical prompts. This paper examines this phenomenon as a form of quasi-development, wherein technological access is widespread but meaningful empowerment remains uneven. Drawing on a synthesis of peer-reviewed studies on digital inequality, AI literacy, cognitive scaffolding, and socio-technical learning environments, combined with empirical observations of AI response variation across devices, internet speeds, and account tiers, this paper argues that AI amplifies existing educational disparities rather than resolving them. The discussion proposes a new paradigm of critical AI literacy, moving from mere access to guided interpretation, prompt reasoning, and epistemic evaluation. The paper concludes that equitable implementation of AI requires infrastructural support, teacher mediation, and the integration of AI reasoning frameworks into formal instructional design.