Aplicação da inteligência artificial generativa na verificação de elementos característicos do discurso sensacionalista em textos jornalísticos

Abstract

This study examines the use of Generative Artificial Intelligence (GAI) to identify sensationalist elements in television news reports about “dangerous internet challenges.” The aim is to assess the potential and limitations of Google’s NotebookLM as a tool for ethical and linguistic analysis in journalism. Five news reports were transcribed and analyzed using prompts based on criteria established in the literature. The analysis focused on alarmist adjectives, emotional phrasing, descriptive exaggeration, and sensationalist tone. A Likert scale was employed to classify the degree of sensationalism in each report. Results indicate that GAI can detect linguistic patterns linked to sensationalism, provided it is guided by well-structured prompts. Reports with neutral language were rated as less sensationalist, while those with emotional narratives received higher scores. Key limitations include the AI’s inability to process visuals or audio, the lack of prompt history, and its reliance on human input. The study concludes that, although GAI does not replace ethical judgment, it can support editorial reviews and contribute to curbing unethical reporting practices.