Client Brief to 3D Printed Construction—An Artificial Intelligence Workflow for Architectural Design Process

The pressing need to provide housing for a projected global population growth mandates the construction of 327 million new independent houses over the next three decades. Achieving this formidable task entails either a tenfold augmentation in the population of practicing architects or a tenfold acceleration of design and construction processes. Notably, the latter objective emerges as an attainable target when compared to the former. Nevertheless, the Architecture, Engineering, and Construction (AEC) industry grapples with an array of challenges emanating from the prevailing Architectural Design Process (ADP). These predicaments encompass substantial investments in software skill acquisition and maintenance, the absence of real-time feedback bridging the conceptual and construction phases, the dearth of real-time physics simulations, the challenges of translating architects' visions into on-site constructed projects, the absence of automation for repetitive tasks, the complexities associated with converting drawing typologies, to name a few. This research adopts a qualitative research paradigm, characterized by an exploratory investigation grounded in personal experiences, empirical observations, dialogues, and experimental data. The study introduces a Smart System poised to redefine the ADP by integrating Artificial Intelligence (AI) comprehensively. This innovative system aspires to be both foundational and open, fostering continuous development and improvement. To facilitate this vision, the realm of Building Information Modeling (BIM) and the Industry Foundation Classes (IFC) file format are extended, capitalizing on the Python programming language. The process of prototyping and experimentation unfolds within the open-source 3D software Blender, exemplifying the viability of an open system. It is crucial to underscore that the research places a distinct emphasis on the refinement of extant AI models, prioritizing this avenue over the creation of new models. This preference is underscored by the formidable time and computational resources requisite for training novel AI models, constituting a central constraint within the scope of this inquiry.