gene{t}rative

gene{t}rative investigate the impact of neural networks on architecture, questioning their potential to enrich the dialogue between designers and machines while rethinking creative processes and architectural notation systems. Drawing on theater—which, like artificial intelligence, relies on collective memory—we focus on natural language descriptions of stage sets. Through a text to image to space design sequence, we experiment with a new approach to scenography, assessing its relevance not only for the play itself but also for theater and architecture as a whole.

gene{t}rative is both a research project and a design exploration, driven by our shared interest in theater and the rise of artificial intelligence. Both theater and neural networks rely on a collective, plural memory to function. The use of AI is not just an interaction between a creator and a machine, it is a form of collective intelligence raising questions about the role of each individual in creating a project. Theater, like AI, activates various types of memory. We focused on AI models capable of generating images from text descriptions, exploring their potential in theatrical scenography. Like theater, these models create an infinite range of sets from an unchanged text. Our goal is to experiment with new approaches to visual and spatial design, project sequences, and systems of architectural notation, focusing on the transition from text to image to space. We embarked on a scenography project using Jean Genet’s The Balcony. This play resonates with our exploration, as Genet emphasizes space over dialogue. Our aim is to reinterpret the play, creating a specialized dataset to revisit part of our collective history and imagination. The key feature of our work is its ability to generate a variety of sets that remain faithful to the original text. These different visual interpretations offer unique atmospheres, despite sharing the same foundational process. This abundance of sets brings the narrative to life through space, externalizing emotions and spatializing the text. The set becomes an active participant in the story. We move away from traditional set changes to embrace constant transitions between environments and dimensions. This new temporal approach dissolves boundaries between locations, creating a dynamic blend of time and space.