Özet:

Abstract In recent times, the realm of Artificial Intelligence (AI) has made a significant and far-reaching impact across various aspects of human life, with academia being no exception. One noteworthy emergence in this field is OpenAI's ChatGPT, an AI-driven language model that adeptly generates comprehensive text in accordance with user requirements. This research endeavors to delve into the parallels and distinctions existing between AI-generated argumentative essays and those crafted by human hands. This examination revolves around their underlying logical meta-functions. To achieve this objective, the study employs the System of Clause Complexes, encompassing taxis and logico-semantic relationships (as outlined by Halliday & Matthiessen in 2014), as the foundational theoretical framework. Furthermore, the research gathers a set of 10 argumentative essays each, generated by ChatGPT, composed by proficient English-native users, and authored by ESL learners. These collected essays are subsequently subjected to annotation, a task facilitated by the utilization of the UAM tool. The findings that emerge shed light on notable disparities. AI-generated argumentative essays prominently exhibit a higher proportion of clause complexes, paratactic structures, and positive additive relationships when compared to their human-written counterparts. Conversely, certain elements such as variation, alternation, and conditional relationships are conspicuously absent in AI-generated essays, whereas humans adeptly employ these elements in constructing their arguments. This discovery imparts crucial insights to the creators of AI, offering a deeper understanding of both its limitations and advancements. Moreover, it also holds relevance for English language educators and learners, who can draw upon these findings to enhance their understanding of the nuances and distinctions between AI-generated and human-crafted argumentative essays.

Anahtar Kelimeler: