Evaluating generative artificial intelligence products using fuzzy social network multi-attribute decision-making model: User perspective

The rapid expansion of Generative Artificial Intelligence (GAI) has transformed various industries, giving rise to an array of GAI products. However, inconsistent product quality complicates user decisions, jeopardizing both the sustainability of GAI technologies and their broader adoption. To address these challenges, this study proposes a user-centered evaluation framework that integrates fuzzy social networks with advanced multi-attribute decision-making (MADM) approaches. Grounded theory is first employed to establish a ''marketing-information-product-individual'' system of factors influencing GAI product adoption. Next, fuzzy social networks reduce semantic ambiguity and mitigate expert bias, while the Decision-Making Trial and Evaluation Laboratory (DEMATEL) method uncovers causal relationships among these factors. The DEMATEL-based Analytic Network Process (DANP) then quantifies the relative importance of each factor, followed by a modified VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) method to comprehensively evaluate representative GAI products. The findings reveal that product maturity exerts the strongest driving force, whereas perceived effect experiences the highest overall impact. Information reliability and the individual dimension carry the greatest weights. Moreover, current products display notable deficiencies in risk management, user service, and ease of use, all of which warrant developers' attention to enhance user satisfaction and adoption. In light of these results, the study proposes targeted optimization strategies for four distinct GAI products. By integrating fuzzy social networks and MADM methodologies, this framework offers a rigorous, systematic evaluation tool that significantly improves decision-making accuracy and promotes the sustainable development of GAI applications.