Reimagining Leadership: Visual Biases in AI-Generated Portraits

This study examines how generative AI systems reproduce bias in leadership representations. Despite the use of neutral prompts, analysis of 71 images produced by eight different models revealed clear patterns of bias related to gender, ethnicity, and age. Strong leader figures were predominantly represented as young, Western, and male, while women and ethnic minorities were more often depicted as passive or “visionless.” Findings suggest that generative AI systems do not simply reflect data, they replicate and reinforce cultural norms. The research emphasizes the need to move beyond evaluating AI based solely on technical performance, calling instead for assessments based on representational justice. The study argues that generative AI has the capacity to subtly reproduce existing social biases. To ensure diversity and equity in representation, AI training datasets must be developed with inclusive and fair principles. Otherwise, today’s biases risk becoming tomorrow’s digital myths through the uncritical deployment of such technologies.