Trusting TikTok: The Influence of Human and Machine Agents on the Credibility of Beauty Product Information on TikTok

In recent years, TikTok has emerged as a leading platform for product discovery, particularly within the beauty industry. For Gen Z, the app functions not only as a source of entertainment but also as a trusted space for learning about products through influencers, algorithmic recommendations, and user interactions. However, the credibility of the beauty- related information presented on TikTok remains under-researched, especially when considering the interplay between human agents (e.g., influencers, engagement signals) and machine agents (e.g., algorithmic recommendations, search functions). This study addresses this gap by asking: How do human agents and machine agents influence Gen Z users' perception of the credibility of beauty product information on TikTok? To explore this question, a quantitative research design was employed. Data was collected through an online survey targeting Gen Z users (aged 18-27). The survey included validated scales for constructs such as perceived influencer expertise, reliability, physical attractiveness, meta-voicing, algorithm, and the search affordance. The dependent variable, credibility perception, was measured through multiple items reflecting trustworthiness and expertise. The data were analyzed using Ordinary Least Squares (OLS) regressions to test a set of predefined hypotheses. The findings reveal a nuanced picture. Among human agents, influencer reliability significantly predicted credibility perceptions, while expertise and attractiveness did not consistently show significant effects. Additionally, higher engagement signals (meta-voicing) were associated with increased credibility, suggesting that social proof still plays a crucial role in how Gen Z evaluates content. Regarding machine agents, both personalized algorithmic recommendations and search affordances significantly predicted higher perceived credibility, indicating that Gen Z users place trust in the platform's technological infrastructure as much as in individual content creators. A hierarchical regression comparing human and machine agents showed that the inclusion of human factors improved the overall model. However, only influencer reliability remained a significant individual predictor, while machine-based features, particularly algorithmic recommendations, consistently showed strong effects. These results suggest that Gen Z's credibility judgments are not driven solely by either human or machine agents, but rather reflect a hybrid model, where both play meaningful, though varied, roles.