Contextualizing a Proposed AI Governance Policy Framework within the Communication Program of St. Paul University Manila

Abstract

This study investigated the dualistic role of artificial intelligence (AI) in shaping critical thinking within educational contexts, specifically focusing on the Communication Program at St. Paul University Manila. Employing a qualitative, interpretive research design, the study utilizes comparative discourse analysis of contradicting YouTube transcripts alongside a synthetic analysis of scholarly article by Brian Bantugan that proposed a policy framework on AI governance in St. Paul University Manila. The analytical framework integrates Cognitive Hierarchy Theory, Sociocultural Theory, and Cognitive Offloading Theory to move beyond binary perspectives of advocacy or resistance. Findings reveal two primary paradigms: AI as "cognitive augmentation," where the technology reconfigures intellect toward high-level task stewardship and information verification, and AI as "cognitive substitution," where overreliance leads to "passive consumption" and cognitive atrophy. Results indicate that while AI can enhance higher-order analysis, it risks "short-circuiting" the foundational mental effort required for deep learning. Consequently, the study proposes a progressive "Learn First, Augment Later" pedagogical model. In Basic Education, AI should serve as a limited "cognitive scaffold" to protect foundational skill formation. In Higher Education, AI transitions into a "cognitive partner" for advanced synthesis, provided learners maintain active agency. These insights culminate in a modified AI governance framework for St. Paul University Manila, emphasizing stage-based integration, process-based assessment, and the transition from output-oriented to cognition-centered evaluation. Ultimately, the study concludes that AI’s impact is determined not by the technology itself, but by the developmental timing of its integration, ensuring students remain active seekers of knowledge rather than passive recipients of machine-generated content.