Identifying leadership talent remains a critical challenge for organizations, with traditional methods often lacking predictive accuracy. This study explores whether artificial intelligence can enhance our ability to predict leadership emergence in teams. Using demographics, personality and other individual difference variables, motivational attributes, and peer evaluation data from 8,112 individuals participating in global virtual teams, we developed and tested several machine learning models for identifying emergent leaders and non-leaders. The best-performing model (XGBoost) achieved an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.87 and overall classification accuracy of 79%. The model showed balanced performance in identifying both leaders (F1-score: 0.76) and non-leaders (F1-score: 0.81). Analysis of feature importance revealed that teammate-assessed behaviors—particularly helpfulness with coordination, task-based effort, and communication frequency—along with English language competency were the strongest predictors of leadership emergence. Our findings demonstrate that machine learning approaches can significantly improve leadership identification accuracy compared to conventional methods, while offering insights into the relative importance of different predictive factors. This research advances both leadership theory and practice by validating AI’s potential for enhancing talent identification processes, particularly in identifying early-career leadership potential, and presents a methodological framework for future AI applications in leadership research.
