Construction of an evaluation index system for graduate students’ effective generative AI utilization ability: a meta-intelligence perspective

The incorporation of generative AI into higher education has created new opportunities and challenges for graduate students' research and learning. This study developed an evaluation index system for graduate students’ effective generative AI utilization ability based on meta-intelligence theory and literature analysis. A two-round Delphi consultation with 17 experts was conducted to refine the initial indicators, followed by a questionnaire survey of 478 graduate students. Exploratory and confirmatory factor analysis were conducted to examine the structural validity of the system. The results supported a four-dimensional structure comprising analytical ability, creative ability, practical ability, and wisdom ability, including 19 secondary indicators and 46 specific items. The four-factor model demonstrated satisfactory fit, with standardized factor loadings exceeding 0.7 and strong reliability and validity evidence. Subjective and objective weighting methods were further integrated to determine indicator weights. This study provides a theoretically grounded and empirically supported basis for evaluating and cultivating graduate students’ effective generative AI utilization ability.