Faizan Zaidi, Director of Technology, shares his insights on AI technology within the architecture industry for AIA New York State.
In the ever-evolving landscape of technological advancements, the intersection of artificial intelligence and architecture has begun to show signs of radically reshaping the foundations of the profession. As AI tools permeate deeper into our work lives, it's an increasing possibility that Generative AI with its remarkable potential finds its place as an essential tool for architects and designers.
When assessing the current status of Generative AI tools in the architecture, engineering and construction industry, however, it's essential to contextualize their position within the broader framework of the Hype Cycle for emerging technologies. In 1995, Jackie Fenn developed a framework to chart the rise and adoption of technological trends over time.
Hype Cycle embodies the idea that every emerging technology experiences a pattern of boom-and-bust before ultimately discovering practical applications that foster gradual growth. An illustrative example of this phenomenon is the late 1990s dot-com boom, characterized by a frenzy of internet adoption and subsequent market collapse, which led to the demise of many internet companies. However, from the infamous collapse emerged enduring giants like Amazon, Adobe, and eBay, which have fundamentally transformed various facets of our lives. In a similar vein, Generative AI in the architecture industry can currently be placed in the technology Trigger’ phase, where actual utility is difficult to distinguish from the overwhelming enthusiasm. While a plethora of AI-driven tools is making inroads into the industry, it can be challenging to discern long-term adoption trends amidst the current landscape.
Current state of generative AI.
Looking strictly from the lens of technical prowess, even in its nascent stages, Generative AI tools are showing capabilities that were previously not available to architects and designers. Tools that have gained traction in the industry have been for the most part associated with the conceptual and planning phases of a project. Majority of these tools use a Large Learning Model (LLM) which is trained from troves of existing data and visuals. With these tools, a text prompt of a vague idea can transform sketches and concepts into formal design visuals, opening a wide landscape of design inspirations. This is particularly valuable during conceptual and preliminary design phases. Imagine an architect using a simple prompt, like "urban sanctuary," and witnessing an AI system swiftly converting it into a series of visual concepts, acting as a wellspring of creativity and innovation.
Generative AI’s capabilities have also shown promising results in the initial planning and feasibility phase of a project. Currently, the task of generating test fits, massing and site feasibility studies is often marred by time and budget constraints which limit the range of options explored. With Generative AI, architects can accelerate the creation of test fits and exponentially increase the variety of options generated. This is a game-changer, especially when considering different sites and their unique challenges. Suddenly, architects are equipped with a tool that not only streamlines the decision-making process but also unveils a multitude of design paths previously uncharted.
Another use case that has not seen tools developed around it yet but is the logical next step to transform 3D model coordination into an autonomous system. Incorporating machine learning to coordinate models between the architect and various trades involved. By leveraging past Revit models and the lessons learned from them, firms can train their proprietary AI systems to make informed decisions during the coordination of future projects. This not only enhances efficiency but also serves as an early warning system, catching costly errors before they escalate into significant setbacks.
Rise & Impact of the Design Assist Tool
Reviewing the potential use cases of Generative AI, a distinct theme can be observed -The emergence of AI as a Design Assist tool, working in tandem with architects and designers to exponentially increase their ability to iterate and produce. Generative AI's role as a design assistant holds immense potential, essentially cutting down the time required to translate an idea/inspiration into a deliverable. AI becomes a collaborator, providing various iterations of original concepts envisioned by architecture firms, ultimately working with the design team to augment their creativity and efficiency.
The impact of Generative AI will eventually, at a leadership level, force fundamental discussions around firm structure and organization. We are stepping into a technological era shift where with the leverage of AI the resources required to complete a task will need to be reevaluated. As AI tools weave themselves into the fabric of daily design workflows, architects will be compelled to revisit contracts and proposals. The efficiencies gained in various project phases will inevitably reshape fee structures, reflecting the newfound effectiveness brought about by AI integration. Without a doubt, the next facet of change management will be centered around the implementation of Generative AI. In the long run, AI tools exert a substantial influence on staff utilization and hiring practices, the accelerated pace at which certain project tasks are completed prompts a reevaluation of team structure. Architects will eventually have the freedom to allocate more time to perfecting project designs, confident that AI is adeptly handling repetitive documentation tasks.
Impending Challenges
However, this progress is not without its challenges. Over time, firms must grapple with the decision to train proprietary AI systems using their past projects' data or purchasing off-the-shelf AI tools. The former option aligns with the firm's design language but demands substantial investment in time and resources, whereas the latter discounts uniqueness and proprietary design language. Firms will also be required to continually audit their process to ensure they are delivering innovative and optimized solutions that meet potentially conflicting objectives; including aesthetics, functionality, and sustainability – something AI may initially struggle with. Human oversight will be required at every turn to fully understand and communicate the design intent and confirm it aligns with the client’s vision.
Rapid evolution of Generative AI tools has also stirred legal complexities and ignited a host of copyright disputes. Many early developers trained their AI models by scouring the internet for data, operating under the assumption that all online information is fair game for use. This practice has recently led to a surge in copyright infringement claims, with AI models being trained using proprietary data without the original author's consent. In the context of the architecture industry, rendering and other showcased works, readily available on firms' websites, have become vulnerable to unauthorized usage for AI modeling. As Generative AI continues to mature and finds its place in mainstream industry practices, architecture firms will face the imperative task of revisiting their legal contracts and copyright protection clauses to safeguard their data and 3D models, which have now assumed a newfound significance as the essential fuel required to train AI models.
Analyzing the HypeCycle chart, It will be wise for most ost firms to let the Generative AI trend mature and reach the ‘Slope of Enlightenment’ prior to making substantial business decisions, driven by this new technology. Regardless of the trend, the embrace of Generative AIushers in unprecedented opportunities for creativity and efficiency, revolutionizing how architects approach their craft. As AI tools continue to evolve and become more integral to the field, architects must navigate the balance between harnessing the potential of AI and preserving the unique human touch that defines architecture.