AI and Arabic Argumentation: A Genre-Based Comparison of ChatGPT and Gemini Outputs

Abstract

This study investigates how two advanced generative language models—ChatGPT (OpenAI) and Gemini (Google DeepMind)—construct Arabic argumentative texts through the view of the Genre-Based Approach (GBA) within the framework of Systemic Functional Linguistics (SFL). Using a descriptive qualitative design, the research compares the schematic, lexicogrammatical, and rhetorical features of texts generated under identical prompt conditions: “Write an Arabic argumentative text about the role of artificial intelligence in Arabic language education.” Findings reveal that ChatGPT produces a humanistic argumentative genre, characterized by a thesis–argument–conclusion structure, experiential reasoning, and affective engagement. In contrast, Gemini constructs a technocratic argumentative genre that follows a claim–counterclaim–resolution schema, marked by nominal density, logical sequencing, and analytical detachment. At the lexicogrammatical level, ChatGPT favors mental and affective process verbs that emphasize interpersonal meaning, while Gemini relies on relational and material processes that foreground ideational precision and objectivity. These results demonstrate that AI systems, despite being non-human entities, encode distinct genre ideologies and epistemological orientations within their linguistic outputs. The study contributes to the expanding discourse on AI-generated language by highlighting how GBA can be adapted to evaluate non-human authorship, offering implications for genre-based pedagogy, corpus design, and critical AI literacy in Arabic language education.