The Algorithmic Gaze

When you ask an AI image generator to create a picture of a “Middle Eastern city,” what does it produce? Minarets, desert tones, and veiled women — visual tropes that owe more to nineteenth-century Orientalist painting than to any lived reality in Cairo, Beirut, or Tehran. Kate Crawford’s “Excavating AI” traces how the training datasets behind image recognition systems encode the biases of their creators — from racist labels in ImageNet to the overrepresentation of certain bodies and the erasure of others. Kotliar’s “Data Orientalism” extends this critique to show how algorithmic systems construct the “non-Western other” through patterns of visual and textual data, reproducing colonial knowledge structures in computational form. This session is paired with a hands-on Midjourney workshop where we will run our own experiments and a historical exploration of how photography has framed the “Orient” since the nineteenth century.

How do training datasets become instruments of power? What does it mean when machines “see” the Middle East through a lens shaped by Western archives? And can we intervene in the algorithmic gaze, or are we simply reproducing it?