Title: Mapping GenAI Literacy: Identifying Student-AI Interaction Patterns in Academic Writing via Epistemic Network Analysis

Abstract:

With Generative AI (GenAI) increasingly woven into educational frameworks, cultivating GenAI literacy has become essential. Nevertheless, existing evaluation methods predominantly depend on self-reported questionnaires, which fail to capture how this literacy is enacted within actual learning workflows. To address this limitation, our research employs Learning Analytics (LA). We gathered interaction logs from a cohort of 162 university students who completed an abstract writing task assisted by GenAI. By applying Epistemic Network Analysis (ENA), we modeled and contrasted the questioning approaches of students across different GenAI literacy levels. Initial findings highlight distinct interaction profiles: students with high literacy demonstrate iterative refinement and strategic inquiry, whereas those with lower literacy tend to depend on direct generation prompts. This study illustrates the capacity of process data to define GenAI literacy, thereby establishing a foundation for data-informed assessment and immediate, real-time interventions.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC