Teaching real lessons with fake worlds

Building realistic simulations for teaching helps students develop problem formulation and interpretation skills that AI tools can't automate. By creating synthetic worlds with authentic data patterns—including causal correlations, realistic distributions (normal, log-normal, power-law), and temporal rhythms—educators can provide safe environments where students learn to handle messy, real-world technical challenges. The approach uses emergent complexity from simple agent-based rules rather than manually coding every detail, producing datasets that mirror production systems while allowing students to experiment without real-world consequences.