Storycrafting With Constraints: Sámi Storytelling and Generative AI Workflows

Generative AI workflows potentially have much to offer in enhancing the pipelines of game-making from visual graphics to systems development. However, in the case of game-making from Indigenous or other minorities’ perspectives, representation in large generative AI models is often incorrect, insensitive, or missing altogether. Through the development of a Tabletop Roleplaying Game (TTRPG), we explored bespoke generative AI tooling for improving representation in visuals, cultural context, and language integrations in AI workflows for the Indigenous Sámi people located throughout the High North regions of Norway, Sweden, Finland, and Russia. Through this development, we sought to find the areas in which AI is already useful, needs improvement, or is not useful in the creative pipeline for crafting a story-based game centered around Sámi stories and issues. We utilized various fine-tuning training tools to enhance Sámi representation in base models Stable Diffusion 1.5 and ChatGPT 4o. In turn, we also used those tools for teaching creativity in Indigenous Sámi communities in game design, digital art, and storytelling. We believe the development of such tools can benefit game designers who want to create generative assets and can also benefit educators creating teaching materials for cultural storytelling and language.