Empowering Clients -- Transformation of Design Processes Due to Generative AI
Johannes Schneider, Kilic Sinem, Daniel Stockhammer
TL;DR
Generative AI reshapes architectural design by enabling rapid client-driven ideation and visualization, while shifting architects toward feasibility evaluation and curation. The study combines hands-on use of a text-to-image tool with expert interviews to reveal both the empowering potential for non-experts and the ongoing limitations in accuracy, regulatory compliance, and contextual understanding. It discusses broader implications for authorship, professional identity, and cultural heritage, emphasizing the need for alignment, explainability, and careful design of future AI systems. Overall, AI acts as an augmentative partner that accelerates certain phases of design but does not yet replace human expertise or the socio-cultural responsibilities of architects.
Abstract
Generative AI (GenAI) is transforming creative fields shaping our culture and our heritage. We focus on widespread interactions between clients and (creative) specialists highlighting a change in interaction patterns leading to a shift from the use of expert creativity towards AI-supported client creativity. More specifically, we explore the case of architecture as designing houses is complex involving extensive customer interaction. We investigate the effects of GenAI on the architectural design process and discuss the role of the architect. Our study involved six architects using a general-purpose text-to-image tool for generating designs and providing feedback followed by expert interviews. We find that AI can disrupt the ideation phase by enabling clients to engage in the design process through rapid visualization of their ideas. In turn, so our thesis, the architect's role shifts towards assessing feasibility of such designs. AI's feedback, though valuable, can hamper creativity and innovation by suggesting altering novel, innovative approaches towards more standardized designs. We find that there is uncertainty among architects about the interpretative sovereignty of architecture and identity when AI increasingly takes over authorship. Our findings can also support the design of future AI systems by pinpointing weaknesses and highlighting a novel design process calling for tighter client integration. In our discussion, we also generalize our findings on a broader societal level elaborating on the change of a number of characteristics such as power, capability and responsibility in the triangle of AI, experts, and non-experts. We also discuss risks such as cultural uniformity when it comes to using AI to design artifacts central to our cultural heritage.