A Comparative Discourse Analysis of AI Generated and Human-Written Texts: Coherence, Framing, and Persuasion

Abstract

The rapid growth of artificial intelligence (AI) in natural language generation has reshaped how written discourse is produced, assessed, and consumed. Even though AI-generated texts are becoming similar to human-written texts, there are still concerns regarding their discursive richness and communicative efficiency. The current study is the critical evaluation between AI-generated and human-written texts, which is based on three dimensions: coherence, framing, and persuasion. Moreover, this study analyzes how texts are framed to construct meaning, leading to certain interpretations, and their overall impact on audiences. It is in contrast with other studies that mostly evaluate linguistic accuracy on superficial level. A mixed-methods research methodology, which integrated qualitative discourse analysis with systematic and structured evaluation measures, was employed to analyze a corpus of 60 texts in the academic, journalistic, and argumentative genres. The findings illustrate that texts generated through AI possess a significant amount of grammatical and local coherence, yet they fail to effectively construct compelling arguments or cohesively connect various ideas. Humanly composed texts, on the contrary, tend to be logical and coherent as a whole due to the way they show how thoughts are arranged and the way underlying themes progress ahead. Regarding framing, AI-written texts are more prone to delivering information in general terms, whereas human writings can build interpretive frames, relying on context and communication objectives. The AI-generated texts are convincing yet lack adaptive variation and, therefore, unable to capture the attention of the audience. This paper explores that AI-generated discourse is well-knitted through discourse markers at linguistic level, yet it is limited in its capacity to act as a dynamic communicative process and engage the learners. The results of this research study can be used in the application of AI in education, media, and knowledge production because it is evident that one must approach machine-generated text with critical thinking.