Exploring Gendered Language and Socio-Cognitive Impact in AI-Generated Texts: A Critical Discourse Approach

Abstract

The present study investigates gendered language and social cognition within AI-generated text through AI tools (ChatGPT, Gemini). Van Dijk's social cognition model is utilized to analyze the perpetuation of conventional gendered language, ideologies, and power structures via AI generative systems. The prompts' designs and the corresponding responses from ChatGPT and Gemini are analyzed across macro- and microstructural levels, cognitive representations, and power dynamics. The present study's findings indicate that AI-generated texts depict men as dominant and authoritative figures, and women as emotional laborers and supportive characters. ChatGPT employs more inclusive terminology than Gemini. The incorporation of AI-generated texts into various institutions contributes to the normalization of conventional gender roles. Further development of more inclusive tools is required.