Decoding the duality of GAI anthropomorphism and its joint effects-a sequential mixed-methods approach

The advancement of anthropomorphic generative artificial intelligence, especially in large language models and multimodal capabilities, has developed its two dimensions: functional anthropomorphism and interactional anthropomorphism. Despite this progress, prior research has predominantly emphasized interactional anthropomorphism, neglecting a holistic understanding of the dual dimensions and their combined effects. This research utilizes a sequential mixed-methods approach, starting with qualitative interviews (n = 15) to explore the joint effects of dual anthropomorphism. The qualitative results were incorporated into a subsequent series of experiments aimed at testing the joint effects, their underlying mechanisms, and boundary conditions. By extending the Expectation Confirmation Model (ECM), this research integrates the dual anthropomorphic features of GAI into a dynamic process that links users' initial expectations-both cognitive and emotional-to their subsequent experiences, evaluations, and continuance intentions. This user-centered approach addresses the growing demand in IS research to focus on not only technological features but also on how these features influence user experiences. The findings provide practical recommendations to GAI service designers and deployers, offering strategies to enhance user experiences and improve the effectiveness of GAI applications.