Jornalismo em vídeo gerado por inteligência artificial : narrativas e credibilidade

Abstract

This work deals with video news generated by the Artificial Intelligence system. The use of automation programs in newsrooms is growing and brings to the public an increasing volume of these productions. The adoption of these systems leads to changes in the traditional news-making practices since the algorithms replace man in capturing, organizing, and distributing information. There are also reflections on the format of the news. With the objective of identifying the characteristics of these videos and discovering the credibility strategies and the role of the journalist in these constructions, an analysis was made about the narratives of these productions. The theoretical references for this are, among others, Sodré (1977, 2009a, 2009b), Motta (1999, 2008, 2013), Cannito (2010), Mota (2001), Becker (2018), Vizeu (2002), and Mourão (2001) that deal with narrative, audiovisual narrative, and editing. As for journalistic automation, the central concepts are of Diakopoulos (2018, 2019, 2020), Dörr (2016), and Latar (2018), and among Brazilians there are Squirra and Carreira (2018). Using Content Analysis as a methodology, it was initially proposed a classification of the automatically generated audiovisual journalistic narratives into three types: basic, diversified, and mimetic. Then, each type was analyzed in detail from the perspective of images, sounds and texts. The study revealed the absence of journalists in the narratives and the consequent restructuring of the credibility relationship with the technology itself as a new agent. The adoption of objectivity and distancing as strategies of attractiveness and credibility was also observed.