Reconceptualising academic reading-to-write in the age of generative AI: an AI-mediated knowledge construction framework

Abstract

Generative artificial intelligence (GenAI) research has largely focused on text generation, feedback, and writing enhancement, overlooking the cognitive and epistemic processes underlying academic knowledge construction. This paper proposes the AI-mediated knowledge construction (AMKC) framework to explain how GenAI may support graduate students' reading-to-write development. Integrating academic literacies, reading-to-write research, sociocultural theory, and dialogic approaches, the framework positions GenAI as a cognitive mediator and dialogic partner across dialogic reading, knowledge transformation, source integration, disciplinary meaning-making, critical reflection, and academic writing. Six theoretical propositions elaborate the mechanisms of AI-mediated literacy development, while pedagogical implications and a future research agenda address implementation and empirical validation. By shifting attention from writing assistance to knowledge construction, AMKC provides a theoretically grounded account of GenAI-mediated academic literacy development in higher education.