The Influence of AI-Generated Visual Content and Voice in Affiliate Promotional Videos on Consumers' Purchase Intention on TikTok Shop

Abstract

Artificial intelligence (AI) increasingly enables affiliate creators to produce synthetic visuals and voice-overs, yet these modalities may not influence consumers in the same way. This study examines the separate and joint predictive effects of AI-generated visual content and AI-generated voice on consumers' purchase intention on TikTok Shop. A quantitative cross-sectional survey was administered online using purposive sampling. Of 112 responses received, 100 met the inclusion criteria: respondents were at least 18 years old, active TikTok users, familiar with TikTok Shop, and had viewed affiliate videos containing AI-generated visual and/or voice elements. Three constructs were measured using 15 five-point Likert items. Data were analyzed in SPSS through item validity and reliability tests, classical assumption tests, multiple linear regression, t-tests, an F-test, and the coefficient of determination. AI-generated visual content had a negative but statistically nonsignificant partial effect on purchase intention (B = -0.157, t = -1.705, p = 0.091), whereas AI-generated voice had a positive and significant effect (B = 1.019, t = 10.583, p < 0.001). Jointly, the two predictors significantly explained purchase intention (F(2, 97) = 88.946, p < 0.001; R² = 0.647). The findings demonstrate that visual and auditory AI cues should not be treated as a single homogeneous stimulus. In this sample, clear, natural, and informative synthetic narration was more influential than visual polish alone. The study contributes to AI-enabled social-commerce research by separating two content modalities and identifying their unequal roles in purchase intention.