This article is the seventh in a series focusing on the Architecture of the Metaverse. ArchDaily has collaborated with John Marx, AIA, the founding design principal and Chief Artistic Officer of Form4 Architecture, to bring you monthly articles that seek to define the Metaverse, convey the potential of this new realm as well as understand its constraints.
In so many ways the introduction of AI software has created a sense that we live in a world predicted by science fiction novels. We commonly have rested on the assumption that computers will never be capable of designing. That time has now arrived, and with it comes an opportunity to confront the positives and negatives these new technologies offer.
AI-based design software has the potential to dramatically increase design quality produced by the profession of architecture, it will be especially effective in elevating the average design firm's quality due to three trends that have evolved in the way we design over the past 60 years.
Related Article
Can AI Make Architects Better Storytellers?First, we have deepened our reliance on and confidence in language-based conceptualization in design. Second, we have, as a profession, moved with great effort towards collaborative design practice models. Third, there is a greater demand for design that responds to human needs for emotional resonance. Currently available AI software is particularly strong in supporting these trends in design, and we can reasonably expect future software to be an even more powerful accessory.
What remains to be seen is if this is ultimately a good or a bad thing. There will be significant tradeoffs. How we, as a profession, prioritize or even define “design quality”, the role of the human touch, or the importance of vision, will very much be central to an analysis of how we use these tools, and how we want to evolve further as a profession in the service of humanity, and perhaps in the service of all living things and systems. This applies to both the tools and the trends.
1- Language-Based Design
There was a time when the balance between concept and form favored the subjective. Architects saw themselves as artists or as an applied art. Modernism fundamentally challenged this by placing a much-needed emphasis on concept and "the big idea".
Venturi criticized “picturesque” designs in Complexity and Contradiction in Architecture as superficial. We raised the priority of problem-solving. In this sense, we might see the first hundred years of modernism as moving towards an architecture of abstraction and ideas. This rebalancing vastly improved the profession's ability to deliver practical and useful buildings. But along the way, we moved towards a language-based design process, and away from a form-based visual process. We now rally around "Giving form to ideas". This shift has also resulted in a greater emphasis on objective analysis and values while downplaying if not being deeply suspicious of subjective values. The verbal is a natural format for AI-based design.
We have also shifted our design process from a handmade analog one utilizing sketching, physical models, and perspectives towards a nearly digital one embracing 3D modeling (e.g., FormZ/Revit) and rendering (e.g., Lumion/V-Ray). In all of these a human still crafts form. Recently Generative Design software (e.g., grasshopper) allowed for equation-based form-making, which initiated a curation-based process. This has now evolved into Ai (e.g., Midjourney) sketching, where Ai crafts the form, based on natural language prompts. One can see a future coming very soon where Ai crafts form in a 3D model format and the design team’s role is primarily one of curation.
Currently, there is a wide disconnect between the words we conceptualize and the forms we create. An AI-based design process will challenge us to ensure a greater “fit” between these two, as we will be able to immediately see the results when we change or refine the words we use to prompt a design outcome. In an AI-driven process, we will only be as good as the words we use to dream, and the resulting forms we curate. Words and concepts are much easier to collaborate on than the subjective aspects of form.
2- Collaborative Models
For the last many centuries architecture has largely followed a "Heroic" model, of the sole visionary, most often male, who creates inspiring forms without the help of others. In truth it has never been this simple, considering the legions of people required to design and construct buildings and cities. Over the last 60 years, we have philosophically shifted towards collaborative models.
I sense that we are in the middle phase of understanding the potential of collaborative models. There is still much work to do. Ultimately, we may want to explore the efficacy of embracing Vision + Collaboration, through understanding the value and constraints of each approach. In a horizontal collaborative model, everyone wants to participate, and anyone can be a designer. There are limitations to this goal, but AI may be able to provide significant assistance here.
One of AI's great strengths is that it can function as "A pencil without an agenda". Even within a highly collaborative setting, “The person who holds the pencil, or the mouse, controls the design”. When a human holds this agency there is an implied or implicit agenda. There is no personality conflict with an Ai pencil, an Ai pencil does not have a dominant or submissive personality. AI functions to provide as many options as requested, and it has proven to be able to do this at rates that far exceed those of a human. In highly collaborative settings, generating multiple options quickly and cost-effectively is critical to getting appropriate feedback from the group. In an analog setting, one might expect 1-3 ideas to be explored over 1-2 weeks, depending on team size.
An AI-based design process can generate 20-30+ options in that same time frame. This allows a wider range of voices and ideas to be considered, and with greater detail, before the design window closes. This will reinforce a culture of curation, which is already a fundamental aspect of a collaborative process. At the same time, this will create a significantly wider gap between “human” vision and the form-making abilities of the machine. It will call into question what is "Original Content", but in a collaborative setting, that concept is inherently set aside. We will need to judge whether building as a machine concept, rather than the quirky eccentricities of human touch, is a good or bad thing.
3- Emotional resonance
- AI can help raise the quality of the average architect’s ability to be an effective designer.
- AI draws on a wider database and can be programmed to value and prioritize specific cultures and their formal references. It can "pander" or be more disciplined, depending on the words that are input.
- AI synthesizes form based on the library of images that are provided to it. It currently does not make qualitative judgments, but rather the quality is affected by the level of curation given by the people who assemble the database.
- AI draws primarily on these references. Imagine a deep rich library of design references that include items you have never seen but are of a very high quality. AI can draw on those references and create what is unexpected to you and perhaps the whole team.
The downside is AI currently does not "innovate" in the same way humans do. Many people have challenged whether Ai can create "original" content. It does however create odd and eccentric combinations that in some ways mimic human design patterns. My experience with AI-driven content is that it can be consistently "lovable" to the general public, it generally creates emotionally resonant designs. Often in ways that an average designer might not think of. Most of the 50s and 60s-based futuristic designs I have seen from programs like Midjourney are quite compelling and are intensely popular on social media. This will allow us a better set of tools with which to design towards the public’s interest.
4- The Human Touch
This discussion about AI provokes many additional questions of profound importance. Perhaps, relative to the design of architecture, one of the most salient is the issue of "Vision". Currently "vision" is often seen as the quirky and uniquely human synthesis of form and ideas that are unusual and unexpected, they create innovation (something not seen before) or the brilliant harmonizing of disparate things compellingly (things combined in a way not done before).
I tend, as an artist, to look at Art as the act of sharing your humanity through an expressive medium. Others might take the view that Art is the experience of content in an expressive medium. Most critically, the viewer is important to both statements. This is a variation of a time-honored philosophical paradox: "If the viewer cannot tell the difference, does it matter how the design was created, or which one is more valid?"
On a deeply personal level, it still matters to me. I still crave the human touch. I am still inspired by the visionary, by the unexpected, and value the poetry of the human condition, in all of its compromised glories. It is in how we navigate and respond to the human condition, where the sweet spot happens, where memorable art and architecture are created. In most things that get built, we don't allow ourselves the indulgence, to express this in architecture. Until we are willing to return to that, perhaps AI can help.
How Ai Will Make Everyone a Better Designer: For Better or Worse is written by architect John Marx, AIA, the founding design principal and Chief Artistic Officer of Form4 Architecture, an award-winning San Francisco-based firm that designs prominent buildings, campuses, and interiors for Bay Area tech companies such as Google and Facebook, laboratories for life-science clients, and workplaces for numerous other companies. In 2000-2007, Marx taught a course on the topic of placemaking in cyberspace at the University of California, Berkley, and in 2020 he designed his first project in the Metaverse for Burning Man: The Museum of No Spectators. The following year, John Marx led a design team charged with creating a $500B portal to the Metaverse.
