Persuasion by an AI-Generated Presenter: Perceived Credibility, Message Quality, Engagement, Realism, and Positive eWOM Intention

Consumers increasingly encounter product endorsements delivered by AI-generated video presenters rather than human influencers, yet whether established source and message predictors of persuasion operate similarly when the source lacks a human identity remains unclear. Drawing on source credibility theory and persuasion research, this study develops the AI Presenter Persuasion Model, linking perceived informational credibility, message quality, and engagement with an AI-generated presenter to consumer attitude and, through attitude, positive electronic word-of-mouth intention. A sample of 428 Taiwanese social media users viewed one AI-generated presenter video before completing a survey. All three predictors were positively associated with attitude, which was positively associated with sharing intention. Prior AI experience was associated with stronger credibility-attitude and engagement-attitude relationships, whereas perceived synthetic realism the extent to which the presenter appeared natural and believable was associated with a stronger message quality-attitude relationship. Because all variables were measured in a single post-exposure survey and no human-presenter comparison was included, the findings indicate that credibility, message quality, and engagement remain meaningful evaluative dimensions when the communicator is AI-generated, but do not establish that AI-mediated persuasion replicates or operates independently of mechanisms documented in human-influencer research.