ChatGPT and the Shaping of Literary, Cultural, and Media Discourses

Abstract

Scholars, critics, and media institutions have long shaped cultural and literary canons by selecting, interpreting, and circulating knowledge. In the digital era, tools such as ChatGPT increasingly participate in this process, raising important questions about their role in structuring discourse within literature, culture, and media studies. The significance of this research lies in examining how such tools may influence the transmission of knowledge and the visibility of certain authors, theories, and perspectives in the human sciences. This study aims to explore how ChatGPT shapes cultural, literary, and media discourse. More specifically, it seeks to identify whether the model reinforces dominant canonical references or integrates more diverse perspectives, and to determine how language (English versus French) shapes the production and framing of these discourses. The research adopts a qualitative comparative approach. A corpus of ten questions was submitted to ChatGPT (GPT-4 paid version) in both English and French. The questions address three main areas: postcolonial literature, feminist theory, and media and cultural criticism. The analysis is conducted manually using a qualitative grid focusing on lexical and semantic choices, suggested references (authors, works, and theories), and the interpretative angles privileged in each response. The findings reveal that ChatGPT’s responses vary significantly depending on the language used. The English responses tend to provide more theoretically grounded and conceptually precise analyses, while the French responses are generally more descriptive and simplified. In terms of references, the English outputs privilege Anglophone canonical figures, whereas the French outputs introduce more linguistically diverse references but with less theoretical depth. Overall, the results suggest the presence of an implicit hierarchy of knowledge influenced by dominant academic and cultural canons. This study is limited by its small sample size and its focus on only two languages. Nevertheless, it highlights the role of ChatGPT as a mediator of knowledge in the humanities, capable of shaping literary and cultural narratives. It also suggests the need for a critical use of generative technologies in education and research, and opens perspectives for further studies on linguistic and cultural biases in AI-generated discourse.