Innovating AI-Assisted Academic Literacy in Higher Education: Human Epistemic Control and Sustainable Human–AI Cognitive Partnership

Abstract

This study aims to (1) examine the redefinition of AI literacy through human-AI interaction in academic writing, (2) identify the activation of higher-order thinking skills (HOTS) when students evaluate and revise AI-generated content, and (3) explain how transparency in human-AI collaboration supports academic integrity. Twenty Tourism Study Program students in Bandung participated in in-depth interviews analyzed using NVivo. Nine themes emerged namely: efficiency and ease of AI use, transparency and reflection through the AI-interaction log, idea development and understanding, dependency and risk, paraphrasing and text personalization, critical thinking and information evaluation, ethics and academic integrity, ongoing AI learning, and the burden and effectiveness of the AI-interaction log. These themes were synthesized into six dimensions of AI-assisted academic literacy namely: instrumental, cognitive, textual-authorial, critical-evaluative, ethical-integrity, and reflective-regulatory. Students activated HOTS by comparing AI outputs with other sources, assessing accuracy and relevance, and making revision decisions based on their own judgment. The AI-interaction log enhanced transparency and accountability. The model positions AI as a cognitive partner while students retain epistemic control, authorship, and academic responsibility, ensuring that AI supports rather than replaces their thinking processes