Exploring Key Dimensions of AI-powered Digital Human Live Streaming

AI-powered digital human live streaming is emerging as a significant marketing format in the e-commerce sector. however, academic understanding of its core characteristics remains confined to conceptual transposition and theoretical deduction, lacking empirical evidence of actual perceptions from consumers and multiple stakeholders. Through in-depth interviews with 19 multiple stakeholders from China (Generation Z consumers, e-commerce practitioners, and academics), this study employed thematic analysis with manual coding throughout the process to systematically identify and define three core dimensions of AI-powered digital human live streaming: anthropomorphism, intelligent interactivity and personalized recommendation, while further revealing the multi-layered internal structures of each dimension. The findings reveal that anthropomorphism comprises three levels: visual anthropomorphism, Behavioral anthropomorphism and emotional anthropomorphism; intelligent interactivity encompasses three elements: response immediacy, response relevance and conversational coherence; and personalized recommendation consists of three dimensions: content relevance, perceived personal attention and interaction customizability. This study provides an empirically grounded, user-language-based feature dimension framework for the field of AI-powered digital human live streaming, bridging the conceptual gap between macro-theoretical concepts and micro-level user perceptions, and laying a solid conceptual foundation for subsequent scale development and quantitative model testing.