Label “Buatan AI” pada Iklan TikTok untuk Produk Hedonik: Peran Pengetahuan Persuasi (Konseptual vs. Sikap) terhadap Sikap terhadap Iklan

Simamora, Rospita
Fadli, Fadli
Nasution, Muhammad Dharma Tuah Putra

ARZUSIN: Jurnal Manajemen dan Pendidikan Dasar
Vol 5 No 6 (2025): DESEMBER

The adoption of AI auto-labeling by TikTok since 2024 has established transparency in AI use as a new standard in digital advertising; however, its impact on consumer responses, particularly for hedonic products that are sensitive to warmth and authenticity, remains insufficiently understood. Drawing on the Persuasion Knowledge Model (PKM), this study distinguishes ad-specific persuasion knowledge (PK) into a conceptual dimension (recognition of content as advertising) and an attitudinal dimension (skepticism toward persuasive motives), and examines how the “AI-generated” label influences both dimensions and, in turn, attitude toward the ad (Aad). Using a between-subjects experimental design with a sample of 387 Indonesian TikTok users aged 18–35 years, this research manipulated the presence of an AI label on TikTok ads for a hedonic product (premium snacks). Data were analyzed using path analysis with PLS-SEM. The results show that the AI label significantly increases conceptual PK (β = .487, p < .001) and attitudinal PK (β = .312, p < .001), while conceptual PK in turn strengthens attitudinal PK (β = .394, p < .001). Attitudinal PK subsequently reduces Aad significantly (β = –.618, p < .001). Serial mediation analysis confirms the pathway AI label → conceptual PK → attitudinal PK → Aad (β = –.118, 95% CI [–.172, –.066]) with full mediation (total effect β = –.391). These findings provide three main contributions: (1) formalizing the separation of cognitive (conceptual PK) and affective (attitudinal PK) effects of AI-use disclosure; (2) demonstrating a trade-off between AI transparency and the advertising effectiveness of hedonic products in short-form video contexts; and (3) offering practical implications for TikTok advertisers in the form of content humanization strategies (human-in-the-loop narratives, warmth cues) and label format design (placement at the beginning of the video, semi-transparent display) to preserve affective effectiveness while remaining ethical and transparent. The study concludes that AI transparency, although normative and increasingly unavoidable, entails persuasive costs that need to be mitigated through warm and authentic creative design.