AI Exploratorium: A Gamified Interactive Exhibition for Developing Critical AI Literacy

Abstract

The AI Exploratorium is an interactive, gamified learning environment set in a physical space, designed to develop AI literacy among high-school students. Through hands-on challenges, students train AI models, analyse their outcomes, and reflect on ethical implications. AI Exploratorium introduces key AI concepts for selected areas of AI (e.g. machine learning for image recognition), in the context of real-world use cases. It builds on experiential learning, adding gamification principles, and a spatial design of an interactive exhibition including digital and analogue materials. It is based on a structured AI literacy framework, developed and refined iteratively. The implementation is based on self-directed exploration and problem-based learning, and includes four main stations each presenting a challenge (an "AI puzzle"). In the first challenge participants train an image recognition AI model for a simulated autonomous car. They choose and label training data and test their model in a simulated test drive, competing to be the winning team. Next challenges add new insights; e.g. a deepfake detection card game introduces AI image generation. The "catch me if you can" challenge addresses AI in public surveillance, deepening the topic of ethical issues in AI use. In the final challenge participants develop their own AI application (no coding skills needed) for a personally meaningful purpose. Reflection quizzes after each challenge reinforce learnings acquired. A projected visualization of a "black box of AI" displays key learnings, initially concealed and gradually revealed with each solved challenge. A pilot test evaluation included semi-structured observation, short interviews, and a questionnaire assessing engagement and knowledge acquisition. Preliminary findings indicate that the AI Exploratorium effectively enhances students' understanding of AI concepts while fostering critical reflection on ethical considerations. This approach emphasizes the role of an exhibition-like learning environment in developing critical AI literacy, and aims to spark discussions on making AI concepts experienceable and AI literacy programmes implementable in different learning environments.