Pedagogical Equipoise: Navigating The Affordances, Risks, and Institutional Readiness of Generative and Agentic Artificial Intelligence in Higher Education

Abstract

Generative and agentic artificial intelligence (AI) tools are reshaping teaching, learning, assessment, and institutional practice in higher education, yet decisions about which tools to adopt and how to use them remain largely ad hoc. This paper introduces pedagogical equipoise as an organizing concept for evaluating AI tools: a deliberate, evidence-informed and provisional balance between the instructional benefits a tool affords and the cognitive, ethical, and institutional risks it introduces, rather than a default assumption that a tool is either beneficial or harmful. Using a structured qualitative document analysis of 27 widely used generative and agentic AI tools, spanning conversational, visual/video, audio, and agentic/development categories, the study examines pedagogical affordances and constraints alongside institutional readiness. The analysis draws on Constructivist Learning Theory, Mayer's Cognitive Theory of Multimedia Learning, and Critical Digital Pedagogy. The paper is grounded in a thematic synthesis of peer-reviewed and authoritative grey literature published between 2020 and 2025, following Braun and Clarke's (2006) six-phase approach to thematic analysis. Four interconnected themes emerged: conversational AI as a pedagogical partner; multimodal AI as a resource for representation and expression; agentic AI for workflow automation; and an overarching tension between efficiency and cognitive engagement. The paper concludes by synthesizing these themes and their theoretical grounding into a decision framework intended to help institutions and educators operationalize pedagogical equipoise in policy, professional development, and everyday practice.