Project Summary: 

This research project explores the opportunities of using Image Generative Artificial Intelligence tools (Image GenAI) to support architects and engage clients early in the architectural design process. Intended outcomes of this research project includes the creation of an Architectural Des(AI)gn Model and Toolkit for supporting efficacious architectural design visualisations and designer-client dialogue efficiency with Image GenAI.  

This is because timely feedback loops between architects and clients are foundational to iterative architectural design processes. However, the language used between architects and clients can be ambiguous and often lead to errors later in the design process. Image Generative Artificial Intelligence technologies (Image GenAI) have eased the instant translation of text into sophisticated visual outputs. Architects are now greater supported to visualise large volumes of ideas in a more efficient and timely manner. Clients can now greater engage in design activities, such as inspiration seeking, without a reliance on, or limited by, traditional design skills to do so. Presenting a great opportunity to support designer-client interactions with Image GenAI tools. 

Today, there remains a need for greater understanding, balance, and exploration of designer-client interactions in architecture. Specifically, for a greater balance in design efficiency, creative expression and exploration on the uses and misuses of emerging technologies among different individuals. Hence, more human-centered, socio-cultural, and socio-technical approaches are called to capture new ways humans designers, and clients, think and engage with emerging technologies, like Image GenAI (Lim, 2024; Pouliou et al., 2024; Vartiainen & Tedre, 2024).  

Thus far, preliminary analysis of interaction data and survey responses revealed similarities in Image GenAI use and dissimilarities in survey responses among architect and non-designer individuals. Further analysis of prompting behaviour revealed architect’s preference for the first prompt and images generated compared to non-designer’s preference for the last prompt and images generated. Two distinct methods of design exploration were also observed among architect and non-designer individuals. For the design practice, such findings suggests that while acts of selecting design ideas may be shared, architects should lead in acts of prompting, while non-designers are encouraged to lead in acts of variating. 

References 

Lim, C.-K. (2024). From pencil to pixel: The Evolution of Design Ideation Tools. Proceedings of the International Conference on Computer-Aided Architectural Design Research in Asia, 3, 89-98. https://doi.org/10.52842/conf.caadria.2024.3.089  

Pouliou, P., Palamas, G., & Horvath, A. S. (2024). Decisions We Should Put in the Algorithm: Mapping architect’s attitudes towards computational and AI-powered tools for practice. Conference: 29th Annual Conference for Computer-Aided Architectural Design Research in Asia (CAADRIA), Singapore.  

Vartiainen, H., & Tedre, M. (2024). How Text-to-Image Generative AI is Transforming Mediated Action. IEEE computer graphics and applications, PP, 1-12. https://doi.org/10.1109/MCG.2024.3355808