Mapping AI Literacy Development Through Minecraft: An Epistemic Network Analysis Study

Abstract

As Artificial Intelligence (AI) rapidly reshapes societal, economic, and technological systems, developing AI literacy among secondary school learners has become an essential educational priority. Yet, little is known about how students construct AI-related understanding within immersive, game-based environments that mirror authentic computational systems. This study addresses this gap by investigating how students engage with AI concepts while completing constructionist tasks in Minecraft and by modelling the structure of their conceptual connections using Epistemic Network Analysis (ENA). Six secondary school students were selected purposively, participating in a series of AI-focused challenges involving automation, data tracking, and ethical reflection, with discourse and digital artefacts captured and coded across three themes: (i) AI Concepts, (ii) Collaboration, and (iii) Problem-Solving. Classification into “advanced” and “novice” groups was determined by a pre-study computational thinking diagnostic and teacher-recorded performance in prior technology elective modules, thereby enhancing methodological transparency and ensuring that group comparisons were grounded in observable prior competence indicators. ENA visualizations suggested that, within this small sample, advanced learners generated dense, triangular epistemic networks indicating tightly integrated conceptual engagement, whereas novice learners demonstrated more fragmented networks that became increasingly cohesive over time. Post-task analyses further demonstrated a strengthened co-occurrence between algorithmic thinking and ethical considerations across all participants. The study’s small sample size and reliance on a single digital platform limit the generalizability of the results, and qualitative coding decisions may introduce interpretive bias. Future work should expand the participant pool, explore additional game-based or simulation environments, and examine the longitudinal trajectories of AI literacy development to understand better how conceptual, collaborative, and ethical dimensions evolve through sustained engagement.