TikTok offers interactive features, such as stitch, duet, and reply-to, that enable creators to respond to others and co-create science content, but it remains unclear whether these features translate into participatory science communication in practice. We developed an AI-powered framework to analyze 2,796 TikTok science short videos with 31,782 image frames and 2,796 audio files, examining how interactive features intersect with communicator identity, communication objectives, and audio-visual presentation strategies. We found that interactive videos were a small share of the overall science videos, yet they showed both promise and risks of participatory science communication: they were actively adopted by scientists and influencers and often used to inform the public and debunk misinformation, but they also facilitated the circulation of pseudoscience. Using latent class analysis, we identified four recurring styles of interactive videos: professional-led explanatory science, ordinary user-led personal commentary on science, influencer-led entertainment or experimental science, and ordinary user-led visual science. This study enriches our understanding of participatory science communication on TikTok and offers an AI-powered analysis framework for studying multimodal science messages on short-video platforms.
Li et al. (Sun,) studied this question.