TokenVerse is an educational adventure game that invites players to explore the inner workings of large language models (LLMs) through interactive gameplay and immersive storytelling. Set inside a fictional AI system, players navigate different environments, interact with AI characters, and uncover the mechanisms and hidden challenges behind seemingly seamless human–AI interactions. By transforming abstract technical concepts into tangible spaces and playable experiences, TokenVerse aims to make AI more accessible, understandable, and engaging for non-experts.
The TokenVerse project investigates how game-based learning can support AI literacy by helping people develop more accurate mental models of AI systems. Rather than simply explaining how AI works, the game encourages players to learn through exploration, experimentation, and reflection. Each chapter introduces different aspects of LLMs, from the risks of human–AI interaction to the processes through which human language is transformed into machine-processable representations. Through playful interactions and critical storytelling, TokenVerse encourages players to look beyond AI’s user-friendly interface and critically examine the technologies, limitations, and sociotechnical systems underlying AI.
Current work on the TokenVerse project includes:
- Designing and developing interactive game chapters that translate complex LLM concepts into accessible gameplay mechanics and spatial metaphors.
- Investigating how game-based learning influences players’ AI literacy, mental models, and understanding of human-AI interaction risks.
- Conducting playtesting and participatory co-design workshops to explore how players interpret, experience, and learn from interactive representations of AI.
- Exploring how narrative and gameplay can encourage critical reflection on the technical, social, and ethical dimensions of AI systems.
View Website