Tracing GenAI Literacy: Uncovering Student-AI Interaction Patterns in Academic Writing through Epistemic Network Analysis

Researchers used learning analytics and epistemic network analysis to study how 162 university students interact with generative AI during academic writing tasks, revealing that high-literacy students employ iterative refinement and strategic questioning while low-literacy students rely on direct generation commands. This data-driven approach offers a new framework for assessing GenAI literacy beyond traditional self-reported scales.

This academic research addresses a growing gap in educational AI assessment by moving beyond subjective surveys to examine actual student-AI interaction patterns. As generative AI becomes embedded in university curricula, understanding how students leverage these tools has become pedagogically urgent. The study's use of interaction logs and epistemic network analysis provides quantifiable behavioral markers that distinguish proficient AI users from novices.

The findings reflect broader trends in AI literacy research where educators struggle to measure meaningful competency in rapidly evolving technology spaces. Traditional assessment methods fail to capture the nuanced decision-making processes that characterize effective AI tool usage. High-literacy students demonstrate metacognitive awareness through iterative refinement—asking follow-up questions, evaluating outputs critically, and requesting modifications. In contrast, low-literacy students show transactional behavior, treating GenAI as a one-shot generation tool rather than a collaborative interface.

For educational institutions and edtech developers, these insights enable more targeted interventions. Real-time dashboards monitoring interaction patterns could flag students adopting passive generation strategies and recommend pedagogical scaffolding. This research framework could inform curriculum design, helping educators teach deliberate questioning and critical evaluation skills alongside domain knowledge.

Looking ahead, this methodology establishes foundations for automated GenAI literacy assessment integrated into learning management systems. As institutions scale AI-assisted learning, process-based assessment becomes essential for identifying struggling students and validating whether AI integration improves learning outcomes or merely productivity. The work suggests that true GenAI literacy requires teaching students to engage AI as a thinking partner rather than an answer generator.

• →High-literacy students use iterative refinement and strategic questioning when working with GenAI, while low-literacy students rely on direct generation commands.
• →Interaction logs and epistemic network analysis can quantify GenAI literacy more reliably than self-reported survey scales.
• →Process-based assessment enables real-time identification of students struggling with effective AI tool usage.
• →This framework could guide curriculum design and inform automated literacy assessments in learning management systems.
• →Behavioral interaction patterns reveal that GenAI literacy involves metacognitive awareness and critical evaluation skills, not just tool familiarity.