The Dialectics of Text and Image: From Mechanical Reproductions to AI Image Productions

This paper presents the outcomes of a funded research initiative aimed at developing pedagogical methods for harnessing AI image production in architectural education. In addition to establishing a theoretical framework, the project involved two workshops in which students experimented with refining textual descriptions of design schemes and reverse-engineering AI-generated images. The goal was twofold: first, to mediate the "translation" between textual and visual representations by grounding their reciprocity in context, program, and precedent studies; second, to capitalize on AI's unintended deviations rather than suppress them.