This study examines the intersection of AI and Arab life narratives, investigating how chatbot-driven generative systems reinvent and distort established storytelling traditions. Because AI models are trained on hidden and opaque datasets, they raise critical concerns about Western bias, censorship, and misrepresentation. Arab storytelling is shaped by lived experience, collective history, and relational ways of knowing, positioning perspective, reality, and intention as central to its narrative logic. Drawing on the work of Sabrey in de-colonial Arab philosophy and Crawford in critical AI studies, this study examines how generative AI models represent Arab culture and narrative forms, suggesting the extent of imbalance, filtering, and Western-centric assumptions in LLMs. Using a hybrid close–distant reading approach, the study compares AI-generated stories to each other and to a similar narrative shared by an Arab storyteller, examining shifts in cultural framing, accuracy, and content. Adopting a comparative, practice-based methodology, the study began with a custom version of ChatGPT 5.2 Pro in English, drawing on a controlled set of academic texts about the region, and was instructed to generate life narratives in Arabic and English about Arabs using only this material. Through iterative tests, it became evident that the model drew on external knowledge. The same prompt was subsequently tested across other platforms (GPT Pro 5.2, GPT-5 Mini, DeepSeek, Jais 30B, Gemini, and Thaura.ai) to examine variations in narrative output.

Authors:
Shahd Nigim, New York University, United Arab Emirates

About the Presenter(s)
Shahd Nigim is a research fellow at NYU Abu Dhabi. Her interests focus on AI bias, cultural representation, and digital storytelling. Her current project examines how artificial intelligence reshapes Arab narratives, authorship, and meaning.

Connect on Linkedin
https://ae.linkedin.com/in/shahd-nigim-bb9064254

See this presentation on the full schedule – Saturday Schedule

Bookmark