Taxis and Logico-semantic Relations in AI-Generated Vs.Human-Written Argumentative Essays: A Comparative Study

Abstract

Recently, Artificial intelligence (AI) has marked a drastic, potential and tremendous impact in different fields of human life and academia is one of them. Open Al's ChatGPT is one of the emerging and remarkable AI Language model tools that respond to the form of comprehensive text according to the requirements of the users. This research aims to find out the similarity and differences between AIGenerated argumentative essays and Humanly Composed argumentative essays on the bases of their logical meta-functions. For this purpose, the System of Clause Complexes consisting of taxis and logico-semantic relations (Halliday & Matthiessen, 2014) has been employed as the theoretical framework. Furthermore, 10 ChatGPT-generated, 10 EN users-written and 10 ESL learners-written argumentative essays are collected and annotated by using the UAM tool. The finding reveals major differences; AI-generated AEs carry a higher percentage of clause complexes, and paratactic and positive additives as compared to human-written AEs. Moreover, variation, alternation, and conditional relations are non-existent in AI-generated AEs, however, humans construct these relations during argumentation. The finding provides insight to AI-creator about the limitation and advancements regarding it as well as to English language teachers and learners.