ChatGPT on TikTok: Multimodal Representations of Artificial Intelligence and Engagement Patterns

Generative artificial intelligence (AI) tools such as ChatGPT are increasingly integrated into many realms of society and everyday life, generating both promise and concern. This underscores the need to examine how they are represented on social media, where their meanings, uses, and implications are publicly negotiated. Drawing on a snapshot of top results returned by TikTok's algorithmic ranking system, we examined 4458 TikTok videos with English-language descriptions related to ChatGPT, combining topic modeling with interpretive multimodal content analysis. We identified six overarching themes ranging from informational content, including tutorials and sociotechnological discussions, to entertainment-oriented content such as playful interactions with AI and memes. Multimodal analysis showed that creators blend visual, aural, and textual resources in distinct ways to construct meanings and assessments across themes. Engagement analyses revealed that entertainment-oriented content attracted higher interaction than informational content, although informational content was more prevalent. With a focus on TikTok, this research illustrates how public communication about emerging technologies is shaped through platform-specific multimodal conventions and engagement dynamics. The findings provide insights for stakeholders who seek to promote responsible AI communication on video-sharing platforms where algorithms shape which content is prioritized.