Designing with Generative AI: Ethical and Creative Redesign of Visual Assets in a Statistics Game

Abstract

In game-based learning, visual elements play a crucial role, particularly in mathematics education, where they often represent numerical information and support conceptual understanding. Nevertheless, copyright constraints may limit the reuse and adaptation of educational games outside their original context, posing both practical and ethical challenges for teachers and researchers. This paper presents the redesign of the educational game Monster Samples, a team-based activity aimed at supporting students’ understanding of descriptive statistics, particularly mean and median through reverse reasoning tasks. In the game, learners must construct samples of monsters that satisfy predetermined statistical conditions, thereby engaging directly with the structure of these measures. The original version relied on illustrated monster cards whose licensing conditions did not allow their reproduction in new academic and dissemination contexts. To ensure legal compliance while preserving pedagogical integrity, Generative Artificial Intelligence tools were used to create a new set of visual assets. The paper describes how the new images were created with the support of AI, detailing the formulation of prompts, the successive cycles of refinement, and the need to ensure coherence with the mathematical structure of the game. Maintaining statistical variability across visual features was essential in order to preserve both the gameplay and the intended learning outcomes. The process also exposed clear limitations of Generative AI. In particular, generating monsters with more than three eyes consistently proved challenging, requiring repeated corrections and highlighting the difference between producing visually convincing images and ensuring quantitative accuracy. The paper further discusses practical challenges, affordances, and ethical considerations related to authorship, transparency, and responsible AI use in educational design. By documenting both opportunities and limitations, it illustrates how Generative Artificial Intelligence can function as a practical design partner while requiring careful pedagogical and ethical oversight. The contribution offers a transferable example of AI-driven content development that supports sustainable innovation in mathematics game-based learning.