This series explores what happens when a technical system tries to name what it cannot recognize. The object detector observes, translates, and classifies—but often fails. Where there is an offering, it says vase. Where there is a tamal, it says sandwich. These are not mistakes to laugh at, but cracks in a system unable to grasp the symbolic codes of local life.

The project engages a poetics of the untranslatable: objects that resist classification, gestures that fall outside global tagging systems. Each misreading reveals a cultural tension—a space where meaning exceeds the machine’s vocabulary.

As Homi Bhabha reminds us, translation is not the smooth transfer of meaning, but the site of conflict, ambivalence, and rearticulation. There is no simple equivalence. The label displaces more than it clarifies.

In this disjunction between image and machine interpretation, something else appears: a remainder, a resistance, a domestic spirituality that the AI cannot name. Its so-called “universal” vision, as Donna Haraway would argue, is always partial, situated, and deeply unfamiliar to the realities these images contain.

Rather than correcting the glitch, the work dwells in it. The glitch is not visual—it is semantic. It reveals what the system cannot absorb.

This is an archive of what remains unread—and of what never meant to be read by systems like these. A chair in a field, a soda bottle as altar, a plastic flower full of devotion. Nothing monumental. Everything heavy with meaning.

As Gayatri Spivak warns, translation can also erase. And sometimes, choosing not to translate becomes an act of care.