An AI Method to Score Celebrity Visual Potential.

This article focuses on developing and validating a machine learning model to predict Celebrity Visual Potential (CVP)—the likelihood that a person’s facial features convey charisma and persuasiveness associated with celebrity status—beyond mere physical attractiveness. Using a dataset of 22,000 celebrity and noncelebrity faces, the authors identify 11 facial features linked to inferred personality traits such as dominance, warmth, and competence, which mediate CVP. The deep learning model achieves 95.92% accuracy in distinguishing celebrities from noncelebrities and produces a CVP score that correlates positively with human judgments and predicts real-world outcomes like Instagram follower growth and LinkedIn executive status, independent of attractiveness. The research offers theoretical insights into the complex relationships between facial features, personality inferences, and celebrity potential, and suggests practical applications for marketing, human resources, and AI-generated virtual personas, while noting ethical considerations and limitations related to privacy and occupational variability.