Perceived AI Anthropomorphism and Brand Loyalty in Generative AI Services: The Roles of Confirmation, Brand Trust, Satisfaction, and Continuance Intention

Generative artificial intelligence services combine functional performance with humanlike interaction, yet it remains unclear whether perceived humanlikeness contributes directly to post-use satisfaction or is associated primarily with more proximal performance and relational appraisals. Drawing on the stimulus–organism–response framework, anthropomorphism theory and the Computers Are Social Actors perspective, the Expectation-Confirmation Model, and consumer–brand relationship research, this study examines perceived AI anthropomorphism (PAIA), confirmation, brand trust, satisfaction, continuance intention, and brand loyalty. Survey data were obtained from 555 ChatGPT users in Türkiye and analyzed using PLS-SEM. PAIA was strongly associated with confirmation (β = 0.624) and brand trust (β = 0.578), whereas its direct association with satisfaction was small (β = 0.081). Confirmation and brand trust were positively associated with satisfaction (β = 0.581 and β = 0.306), satisfaction was strongly associated with continuance intention (β = 0.705), and brand trust, satisfaction, and continuance intention were each positively associated with brand loyalty. Exploratory factor analysis, HTMT diagnostics, and confirmatory comparisons of alternative measurement structures showed substantial overlap between confirmation and satisfaction. The six-construct model fit better than the more parsimonious alternatives examined, but conclusions that depend on separating these two constructs are treated as tentative. Overall, PAIA was strongly associated with post-use and relational appraisals, but whether its association with satisfaction runs mainly through confirmation could not be established with confidence. Because the design is cross-sectional, the findings are interpreted as associations rather than causal effects.