Linguopragmatic Features of AI-Generated Text in the Media Discourse of Social Networks (on the Example of Texts Devoted to the Governor Election in the Nizhny Novgorod Region, Russian Federation, 2023)

Abstract

Aim. To study the distinctive features of neural AI-generated texts on example of political posts on Telegram channels, and to make their classification to both identify the markers of the generated text and prove the presence of such artificially created texts in media discourse.Methodology. Descriptive methods (collection of linguistic material, classification of found linguistic units, description of their features in relation to media discourse), and the methods of content analysis (qualitative and quantitative analysis of markers of the generated texts) were used in the research, which implied AI-generated text analysis.Results. According to the results of the study, 423 linguistic features of AI-generated text have been detected in Ilya Pomerantsev’s 38 posts. The quantitative and percental content of each type of marker are presented.Research implications. The results of this study are valuable since they offer insight to the principles of neural networks, the process of text generation made by AI and its natural language processing. The experiment clearly demonstrates the linguistic level of generative neural networks. The practical significance of this study is the proposed classification of the linguistic peculiarities (markers), which can help specialists analyze the text and then identify the elements of AI-generation.

Ähnliche Arbeiten