Algorithmic Orientalism: Visualizing Bangladeshi Feminism Through Generative AI

Resat Amin

Track: Main Papers Track (6 to 9 pages)

Keywords: Generative AI, Digital Orientalism, Bangladeshi Feminism, Algorithmic Bias, Epistemic Justice, Responsible AI, Postcolonial Theory

Abstract: As generative artificial intelligence (GAI) systems become central to global knowledge production, they increasingly shape how political movements are visualized and understood. This project examines the representational logic of leading text-to-image models ChatGPT, Gemini, and Grok in their construction of feminist activism in Bangladesh. Through a qualitative visual content analysis of 100 AI-generated images, this study identifies a systemic pattern of digital Orientalism and representational imperialism. Findings reveal that these models consistently fail to render Bangladeshi-specific political identities, instead defaulting to ``Indianized'' facial features, Westernized protest aesthetics, and distorted, illegible Bengali script. I argue that these outputs are not neutral technical glitches but are manifestations of algorithmic bias embedded in uneven global data ecologies. These distortions enact a form of epistemic erasure, where the distinct political history of Bangladesh is subsumed under dominant regional and Western tropes. Furthermore, the study identifies a ``modal alignment gap,'' where AI systems textually acknowledge their biases while continuing to reproduce them visually. By reframing these technical failures as an intra–Global South imperialism, this research challenges the adequacy of current Responsible AI frameworks. It concludes that without a commitment to linguistic justice and representational sovereignty, GAI will continue to function as a regime of imperial representation that silences subaltern voices in the digital age.

Submission Number: 50