Detect-Interpret-Respond: A Framework to Ground the Design of Student Inquiry into AI Systems

In this paper, we describe our experience developing a framework for understanding AI systems that we use to drive the design of AI learning experiences for elementary-aged youth in an informal, free-choice environment. This framework—detect/interpret/respond (DIR)—shows promise as a flexible and age-adaptable model for youth to connect across learning experiences and work toward a coherent understanding of AI. As an application of DIR, we describe our experience designing and testing learning experiences related to a cutting-edge AI system (a Virtual Human). Insights from our initial studies suggest that this framework can help unify youth encounters with various AI systems and provide a productive schema that accommodates increasing complexity as youth advance in their understanding. DIR, therefore, offers a heuristic for sense-making and inspection of AI systems that is both accessible to very young children and robust enough to be useful as they mature.