Inteligencia Artificial Aplicada a las Ciencias Sociales

Artificial Intelligence Applied to the Social Sciences: An Analysis of Algorithmic Visualities and Subjectivities on TikTok

Resumo

This article analyzes the visual and expressive dynamics of TikTok through artificial intelligence techniques applied to a corpus of 7,398 videos. The aim is to show how visual methods supported by machine learning can contribute to understanding sociocommunicative and cultural phenomena in digital environments. For this purpose, an unsupervised learning model (exploratory SOM and K-means clustering) was trained, which made it possible to identify four main clusters of audiovisual organization. The first reveals dense and saturated compositions of elements that, beyond the technical, allow us to interpret the screen as an algorithmic chroma: a flexible space of assemblage that organizes gestures, effects, and visual trajectories. The second corresponds to the loops and rhythmic repetitions characteristic of the platform, which create circularity and continuity in the experience. The third shows how bodies and digital effects combine in unstable surfaces, where the human and the technical intertwine without clear boundaries. The fourth cluster points to a regime of expressivity that stimulates constant participation and self-exposure. Taken together, these findings demonstrate that TikTok cannot be understood solely as a transmedial space, but rather as an algorithmic interface that organizes rhythms, affects, and digital subjectivities, encouraging reflection on the political and sensory impact of platforms in contemporary life.