A hybrid human-computer interaction workflow combining computational analysis, human interpretation, and generative artificial intelligence can make changes in the representation of violence visible, according to an exploratory study published in Frontiers in Computer Science. Conducted by researchers across institutions in Mexico, including the Metropolitan Autonomous University, Universidad de las Americas, and Tecnologico de Estudios Superiores de Ecatepec, the investigation examined media narratives of violence through structured human-machine collaboration. The authors developed a five-stage exploratory workflow grounded in mixed-initiative design, in which computational tools handle structured comparison and generative variation while human contributors supply critical judgment.
The first stage of the project involved an exploratory computational evaluation of a non-probability corpus containing 105 news items addressing violence in Mexico. Within this textual collection, the researchers applied computational techniques to identify recurring terms, prominent actors, geographic locations, and underlying semantic relationships. The team then organized and visualized these computational findings to prepare structured materials for subsequent participatory evaluation.
The second and third phases brought these computational visualizations into an interactive participatory workshop conducted with 15 participants.
In the fourth stage of the workflow, the authors analyzed changes across textual and visual iterations.
The fifth stage integrated the project findings through expert-supported synthesis, producing both an infographic and a video that summarized the workshop results. The results suggest that recursive human-AI interaction can make changes in the representation of violence visible by combining semantic analysis, participant interpretation, and visual synthesis.
The authors noted significant methodological boundaries, stating clearly that the study employs an exploratory design that does not attempt to estimate population parameters or establish causal relationships. The reported findings and observed narrative shifts remain specific to the non-probability corpus of 105 news articles and the 15 workshop participants. Despite these constraints, the study authors concluded that the collaborative, mixed-initiative framework effectively demonstrates how computational methods can support generative exploration and comparison while keeping human interpretation central to the analysis of complex media narratives.
Subject of Research: Human-computer interaction for analyzing media narratives of violence in Mexico
Article Title: Human-computer interaction for recursive prompt and image generation on media narratives of violence in Mexico
Article References: Mora-Gutiérrez, R. A., Alatriste-Martinez, Y., Manzanarez-Betancourt, O. A., Meza-Zarate, L. A., Montes-Orozco, E., & Torres Cockrell, G. S. (2026). Human-computer interaction for recursive prompt and image generation on media narratives of violence in Mexico. Frontiers in Computer Science, 8, Article 1877314. https://doi.org/10.3389/fcomp.2026.1877314
Image Credits: AI Generated
DOI: 10.3389/fcomp.2026.1877314
Keywords: Human-computer interaction, Generative artificial intelligence, Media narratives, Violence in Mexico, Mixed-initiative design, Narrative analysis, Semiotics, Recursive prompt refinement, AI image generation, News corpus analysis, Participatory workshop, Semantic relationships, Visual synthesis, Infographic design, Video synthesis, Textual iteration, Computational text analysis
News Source: Denise Maddox. (November 18, 2026). Recursive Human-AI Interaction Can Make Changes in Violence Representation Visible. Scienmag.



