AI-Enhanced Economic Simulations for International Business Education: Impact on Learning Outcomes in Colombian Universities

This study examines whether generative-AI-supported economic role simulations improve conceptual understanding in undergraduate International Business education. The manuscript was revised to clarify the educational problem: International Business students must use economic literacy to reason about market entry, pricing, investment risk, supply chains, and competitive structure, yet many struggle to connect abstract economic concepts to realistic cross-border business decisions. A quasi-experimental longitudinal design compared 79 Colombian International Business students in intact class sections over two academic semesters. The experimental group completed structured ChatGPT-supported simulations, while the comparison group completed the same scenarios through instructor-led discussion and peer role-play. Outcomes included economic concept assessment scores, coded argumentation quality, and computational indicators of interaction quality. Mixed linear models indicated larger learning gains in the AI-supported group than in the comparison group. Mediation analysis suggested that improvements in economic argumentation partly explained the intervention effect, while dialogue turns, TF-IDF-weighted lexical diversity, and LDA C_v topic coherence were associated with stronger posttest performance among students using ChatGPT. Because the sample was modest and assignment occurred at class-section level, the findings are interpreted as promising but not definitive.