IN-PERSON Lisbon Livable Cities. Section A

Abstract

The proposed presentation introduces a novel methodological approach to architectural representation through the use of generative AI, aiming to challenge dominant perceptions of Gulf cities, taking Abu Dhabi in the United Arab Emirates as a case study. While generative AI platforms often reproduce stereotypical imagery of the Arab Gulf as a landscape dominated by megaprojects and iconic “starchitect”-designed buildings, this research proposes a recalibration of that narrative. By retraining an AI model with a curated dataset of over 7,000 original photographs—capturing quotidian structures, vernacular typologies, and overlooked urban conditions—the project extracts what can be described as the architectural DNA of Abu Dhabi. This method is not simply about generating images; it is about shifting the architectural canon. Drawing on Mario Carpo’s theories of digital reproduction and canon formation, the project redefines what is considered representative or worthy of preservation. The resulting AI-generated outputs form a visual landscape that is both novel and strangely familiar—contextual, climate-responsive, and deeply rooted in the aesthetics and lived experiences of the city. The process merges design speculation with machine learning, presenting a new tool for urban representation that resists homogenization and fosters cultural specificity. Through this case study, the presentation not only highlights a groundbreaking use of AI in architecture but also repositions image-making as a critical act of authorship in the post-digital era. This research demonstrates how generative AI, when trained with intention and locality, can become an inclusive instrument for storytelling, heritage documentation, and envisioning future architectural identities.