Communicative roles in human-anchored and AI-presented news commentary short videos: a multimodal intertextuality analysis.

INTRODUCTION: Despite the growing integration of artificial intelligence (AI) into news production, little is known about the intertextual differences between human and AI anchors, the communicative roles constructed through such differences, and their multimodal realization in short-video news commentaries. METHODS: This study approaches multimodal intertextuality from a social semiotic perspective and employs a quantitative multimodal content analysis to compare human-anchored and AI-presented news commentary short videos. RESULTS: The findings reveal that though both programs rely predominantly on verbal intertextuality, human-anchored videos employ a more diversified multimodal intertextual repertoire, particularly through commenting and evaluating and genre intertextuality, whereas AI-presented commentaries depend heavily on direct quotation and make comparatively limited use of other intertextual techniques. These contrasting intertextual configurations construct distinct communicative roles. DISCUSSION: Human-anchored news more frequently assumes the role of an interpretive mediator, recontextualizing information through evaluative, explanatory, and genre-based intertextual practices. By contrast, AI-presented commentaries primarily functions as an institutional information relay that emphasizes fidelity to source materials and authoritative information transmission. Moreover, the multimodal analysis indicates that AI-presented commentaries exhibit a narrower range of verbal-visual intertextual interaction than their human-anchored counterparts, suggesting that technological sophistication in presentation does not necessarily translate into greater intertextual diversity. These findings point to an emerging division of communicative labor in contemporary journalism, highlighting the importance of role differentiation and collaboration in the development of human-AI news production.