Human analysts or artificial intelligence? Source attribution in athlete performance evaluation.

Introduction: Artificial intelligence (AI) is increasingly used to support athlete performance evaluation, yet responses to AI-based analysis may depend on whether the evaluative task is perceived as requiring objective calculation or contextual human judgment. Drawing on task-dependent algorithm aversion, this study examined whether an identical athlete performance report would be evaluated differently when attributed to an AI-based analysis system or professional human analysts. Methods: A total of 220 participants were randomly assigned to review the same football performance report presented with either an AI-based system or professional human analysts as its source. Participants evaluated the report's message credibility, perceived diagnosticity, and consideration of the athlete's unique characteristics. A one-way multivariate analysis of covariance and follow-up univariate analyses were conducted while controlling for familiarity with the athlete. Results: Source attribution had a significant multivariate effect on the combined outcomes. The AI-attributed report received significantly higher evaluations of message credibility and perceived diagnosticity, whereas the human-attributed report was perceived as giving greater consideration to the athlete's unique characteristics. Discussion: The findings imply that algorithm appreciation and algorithm aversion can coexist within the same evaluative task. AI attribution was associated with more favorable evaluations on dimensions linked to systematic information processing, whereas human attribution was associated with greater perceived individualized and context-sensitive understanding.