Analysing the usage of AI art tools for architecture

Abstract: The recent advancements of commercial text-to-image and image-to-image generation platforms have created a surge in interest in many creative disciplines, including architecture. In this paper we analyze 58 million publicly available queries from the platform Midjourney to find architectural use cases. We utilize various statistical and NLP methods to extract quantitatively how users utilize and collaborate on the platform. For some of the most popular queries we discuss qualitatively the generated results in their applicability to architecture. Our results show that while there are still limitations in image generation models, they are already widely adopted for architectural use cases.