Benefit–Risk Perceptions and Public GenAI Acceptance: A Survey Experiment Based on the Emerging Technology Acceptance Model

Drawing on the Emerging Technology Acceptance Model, this study integrates perceived benefits and perceived risks into a unified analytical framework to examine public acceptance of generative artificial intelligence (GenAI). Using three scenario-based survey experiments, the study shows that perceived usefulness and perceived ease of use significantly promote GenAI acceptance, whereas perceived social welfare loss, personal property loss, and technical–ethical conflict all significantly reduce acceptance. The negative effects of these three risk scenarios also differ in magnitude, with personal property loss exerting the strongest inhibitory effect. Heterogeneity analyses further indicate that technology usage behavior moderates these relationships selectively rather than universally, with stronger evidence on the risk side than on the benefit side. In addition, lower confidence in regulatory oversight is more clearly associated with stronger risk sensitivity, whereas evidence for stronger benefit sensitivity under higher regulatory trust remains weak. These findings deepen understanding of the differentiated benefit–risk trade-offs shaping attitudes toward emerging AI technologies among Chinese internet users in an online survey context.