Unpacking media channel effects on AI perception: A network analysis of AI information exposure across channels, overload, literacy, and anxiety among Chinese users

Artificial intelligence (AI) has developed rapidly and attracted growing public attention. While online information regarding AI promotes technological awareness, it may also trigger anxiety. In China, where AI adoption is accelerating, the psychological effects of different media channels for AI information communication remain underexplored. This study employed a cross-sectional survey across mainland China (n = 1371), measuring online AI information scanning, information overload, AI literacy, and AI anxiety. Network analysis using the EBICglasso method examined associations among these variables and identified central and bridging nodes. Results revealed diverse associations across media channels. While most media channels were positively associated with both AI literacy and AI information overload, the strength of these associations varied. Short-video and social media were the most significant in promoting AI literacy, whereas online audio exposure showed the strongest positive association with AI information overload. AI literacy (as an inhibitor) and AI information overload (as a stressor) played opposing roles in mediating the relationship between media exposure and dimensions of AI anxiety, with literacy mitigating and overload exacerbating anxiety. The associations between specific media channels and different dimensions of AI anxiety also differed. Furthermore, centrality analysis identified sociotechnical blindness anxiety as the most influential node and AI learning anxiety as a key bridge across communities. This study is the first to apply network analysis to examine dimensional interrelations among AI information scanning, overload, literacy, and anxiety. Findings highlight the channel-specific impacts of AI communication and offer practical implications for mitigating public AI anxiety through targeted media strategies.