AI Art has taken the world by storm and has created division between many people. Some artists argue that AI Art is “the fast-food of the art world”, while others praise everything that AI has to offer.
There’s been a lot of buzz around this topic over the years, but conversations have exploded recently with the public release of newer neural networks like Stable Diffusion and Dall-E. Now more than ever, it’s essential to question AI and how it might be used in our practice.
How will AI in architecture impact our profession?
In this post, we’ll take a look at this discussion but through the lens of architecture and design. I’ve written about AI on this website a number of times, and it’s something that I’m deeply interested in.
With so many recent stories that question the use of AI for creating graphic and written content, I feel that it’s crucial to weigh in on the dialogue to get a deeper understanding of how architects can use these so-called “AI architecture generators.”
1. Laying the Framework: Art, Architecture and Technology
2. AI In Architecture: The Good
3. AI in Architecture: The Bad
4. AI In Architecture: The Unknown
5.💡 Jon’s Take: The Future of AI in Architecture
As with any design exercise, it’s important to start with the fundamentals.
Laying the Framework: Art, Architecture and Technology
As with any creative endeavor, it’s important to lay a foundation for discover — or, a rule book by which we can interpret something.
The relationship between producing art and producing architecture are somewhat adverse. Artists tend to work from real to abstract, while architects work from abstract to reality. Yet, one cannot exist without the other.
Technology drives creativity, and creating art and architecture are profoundly human undertakings.
To frame this conversation, we’ll need to look at the semantics of the words “art” and “architecture” and question their meaning.
What is Art?
It’s impossible to truly define “art”, since it’s a subjective term. What one may think of as “art”, someone else may think of as worthless. We perceive art as we perceive ourselves — so therefore, art can only be defined by those who interpret it.
In a study conducted by the National Endowment for the Arts in 2012, a series of diagrams helps to illustrate how art works and its role in society. I’ve studied these diagrams in depth, and reinterpreted the one below to simplify the main idea:
(Original study found here.)
However, most people can agree that creating art is a means of self-expression and serves as an outlet to help people communicate their innermost thoughts.
For some, creating art is a cathartic release and allows us to channel and process our emotions. Be it in the form of writing, drawing, painting, producing music, or by the simple phrase of “creating something from nothing,” art completes its cycle when it is shared and interpreted by others.
With art, we can acquire new perspectives and deepen our understanding of the world and how the world appears through the eyes of other people.
What is Architecture?
At its core, architecture is the art and science of human accommodation. The ultimate goal of architecture is to create something that exists in concrete reality — within the third dimension.
Art is hard to define, as is architecture.
However, by nature, architecture is a coalescence of science and art — and defining the architectural “design process” is perhaps more linear.
In the most straightforward way to explain architecture: we start with an idea, gather information (research), parse our research against constraints (building codes), create a set of instructions to provide builders, and oversee the construction of our ideas. When it’s all said and done, three-dimensional space is created.
Stuart Graff, CEO of the Frank Lloyd Wright Foundation, defines architecture in a more poetic sense:
“For Wright, architecture isn’t buildings. Buildings are a form of architecture. Architecture is this sense of continuity and connection with everything around us. You might think of architecture as an ecosystem in Wright’s world… how we relate to the things around us, and how they relate to us.”
Stuart Graff, CEO of the Frank Lloyd Wright Foundation
Source: https://www.secondstudiopod.com/
Humans are three-dimensional beings, and if authentic architecture should support the fundamental requirement of “human accommodation,” the end result must exist in the third dimension.
What are AI Architecture Generators?
AI Architecture generators are software programs or online tools that use artificial intelligence (AI) to create or generate artwork that can be utilized in architecture. Some of the most popular AI art generators allow users to enter a text-based prompt and can be used to create architectural graphics of the interior or exterior of a building.
To preface this even further, check out my other post:
Some AI art generators use machine learning algorithms to analyze existing art and generate new works based on those patterns. Others use deep learning techniques to generate original images from scratch.
AI art generators even allow you to input your own images or ideas, and the AI system will create a new work based on those inputs.
Ultimately, AI art generators are designed to mimic an artist’s style, using algorithms to generate original artwork based on parameters input by the user.
As a result, the images produced by these tools will vary in quality and originality.
There are many areas in which AI could be integrated into the architectural design process — from creating visual concepts to writing code for parametric design. So the phrase “AI Architecture” is somewhat nuanced.
Creating architecture has historically been a time-consuming, iterative process. So to suggest that we could plug in a query to AI and have it “design” something for us seems a bit far-fetched…
But that’s also what we thought about creating art before the latest “AI Art Generators” hit the mainstream in 2022. And now, people are questioning the future of AI in creative industries.
Many artists feel that their work is being ripped off or that their styles are being “stolen” by AI art generators. These discussions raise an essential point for our architecture and design industry, and now is the time to weigh in on the conversation.
But before we get into the downsides of AI, I’ll cover some of the benefits that using AI in architecture provides us.
AI In Architecture: The Good
For over 50 years, architects and designers have integrated computers into nearly every aspect of the design process. From creating concept sketches to developing construction documents, technology plays a massive role in the quality of work we produce. Utilizing the latest technology to design architecture is not only smart but also necessary.
That’s not to say that traditional design methods are becoming less important in the design process… but it’s becoming hard to ignore the rapid advances in BIM technology, generative design, fabrication, and architecture visualization.
Within the last five years, we’ve become accustomed to using new software like Enscape, Dynamo, and Python for Revit. Or, Grasshopper and Ladybug for Rhino. You name the animal, and a software developer might name the latest and greatest parametric design program after it!
Even Adobe software — a staple for nearly any creative profession — is starting to utilize AI in photo editing software like Lightroom to help streamline the editing process for photographers.
2022 was an impressive year for AI.
With the release of datasets like GPT-3 and Stable Diffusion, AI has finally found its way into the visualization world.
And for the practice of architecture, being able to visualize photo-realistic architectural concepts in a matter of seconds has, frankly, blown people’s minds.
Using AI to Create Architecture Concepts
MidJourney has quickly become one of the world’s most popular AI art generators. Architects and designers have found ways how to use MidJourney for architecture concept renderings, and even some of the largest firms in the world are using the software to help clients visualize buildings.
MidJourney uses the AI model called Stable Diffusion to create its images. As Stable Diffusion models continue to develop and “train” from billions of images used in its neural network, the quality of images being produced is clearly being refined.
The images below are generated with four MidJourney versions released in 2022.
I used the following prompt to create these images in the latest four versions of MidJourney:
“modern villa in a forest filled with rain and fog, 8K, detailed, raining, realistic, photographed by Mike Kelley”
Looking at the images above, it’s apparent how AI models are rapidly evolving and producing higher-quality images. The “architecture” quality is also becoming more believable and accurate.
AI in Architectural Design Software
Architects and designers are also finding other ways to integrate AI into the design process, such as using ChatGPT to create custom scripts for modeling in generative software.
Even for software like Revit, plugins are being created to utilize text-to-image AI technology to redefine how we create architectural renderings. If you want to see this in action, check out this video on “How to Create AI Renders Directly from Revit” by Ricardo Morales Quirós.
He’s using a plugin called Veras for Revit, created by EvolveLAB.
Rather than plugging a text prompt into AI like MidJourney, this software integrates directly with Revit 3D views. For any 3D view you set up in Revit, you can plug in a prompt to reimagine that view’s entire look and feel. Changing materials, lighting, atmosphere, and more — happens on the fly.
Although this technology is still very new, it’s incredible that text-to-image rendering capabilities are becoming more grounded and less “random.”
And what’s more, the longer these AI models train, the more sophisticated they become. When AI can teach itself, the evolution will be exponential.
💡 Jon’s Take: The Good Side of AI in Architecture
There are many other ways architects and designers utilize AI in the design process. This technology will only evolve and carve new methods into our practice — ideally, for the better.
If you’re curious to learn more about how architects can use AI for design, be sure to check out some of my other articles:
Best AI Art Generators
For Architects and DesignersAI Art Generators are becoming valuable tools for architects and designers, but which one is the best to use? Here are some of the best AI Art Generators for Architects and Designers, ranked and reviewed based on price, accessibility, features, and more!
The integration of AI into the architectural design process is going to be immensely helpful for streamlining the design process. Being able to rapidly produce design concepts or material renderings means that we’ll be able to shift our focus back to more critical tasks in the design process.
We’ll be able to use AI to take care of mundane production tasks and realign our focus on the critical thinking aspect of architectural design.
But of course, there are two sides to every coin — and some people are starting to feel that the creative industry as a whole is on the precipice of a robot-controlled future.
AI in Architecture: The Bad
Depending on who you ask, AI is actually quite terrifying.
Lucky for us right now, AI doesn’t have the capability of being self-aware. In other words, it doesn’t know that it’s creating something.
However, this rapid advancement of AI makes us question our own “humanness,”.
The concern comes in the efficiency of AI — and the sheer volume of content it can produce in just a matter of minutes… Some would even say that AI in architecture threatens to “replace” humans.
But now, it’s more important than ever to reflect on humans’ relationship with architecture and art.
As AI art generators become more sophisticated, artists have begun voicing their concerns about the legality of where AI is collecting its training data from.
Is AI “Stealing” from Artists?
In this article, the author takes a look at how Lensa is using Stable Diffusion and “ripping off” the styles of human artists.
In the lens of architecture, similar questions are being raised. Another post on Archinect questions the “Real Threat of Artificial Intelligence to the Architecture Profession.”
One of the interesting takes from the comments in this article is the prospect that AI could make it more difficult for entry-level designers in practice. With AI being able to quickly produce some of the more mundane tasks that might be handed over to, say, an intern — it leaves even more questions on the table that deal with learning the fundamentals.
But wouldn’t that just “free” up the time for somebody to learn something else more valuable? For example, rather than spending 8 hours creating one concept rendering, that entry-level designer could focus on a more technical side of design, such as creating construction details — something that AI is not sophisticated enough to know how to do.
Ultimately, there are so many moving parts in architectural design that we certainly use the help of AI to save time. If we can save time producing the mundane, we can reallocate our time to learn something currently impossible for AI to assist with.
When AI evolves and DOES learn how to complete some of the more technical tasks in architecture, we’ll continue to pivot and find new ways to produce better designs.
It always ties back to the fundamentals.
AI is not “Theft”
Some people believe that AI is not “theft”, and AI models have the right to access the same information that we, humans, can access.
In this Reddit Post, the author claims:
“I think the view that AI art is “theft” mainly comes from misunderstanding how AI creates the images or what current artists take for granted when it comes to the creative process, or mere frustration with the ease in which new images are created.”
Looking ahead, here’s the hot-topic question…
Will AI Replace Architects?
Based on the questions above, someone could suggest reasonable concern for the possibility of AI becoming sophisticated enough to “steal” architects’ designs — but to outright “replace” the process of creating architecture seems far-fetched for now.
Artists are concerned about AI “stealing” their styles — and the evidence is hard to argue. We’ll need to start questioning the same for architecture, as AI becomes more advanced and continues to use existing architectural precedents as its inspiration.
However, “Steal” is undoubtedly a strong word to use — especially given that any “research binge” that architects and designers would indulge in for inspiration nearly always results in skimming through books or the internet to find graphic inspiration.
But to make sense of AI in architecture, we’ll need to look deeper into exactly how AI art generators work.
If you’re curious, be sure to read my other post:
One sentiment I often use in architectural design is “if you know where it comes from, you’ll know where to take it.”
As with any creative endeavor, it’s vital to try to gain an understanding of its roots by constantly asking questions — especially when it comes to using AI in architecture and design.
As mentioned earlier, art influences architecture — so if the art world is affected by AI, then inherently, the world of architecture will also be impacted.
💡 Jon’s Take: The Bad Side of AI in Architecture
To be honest, I’ve never felt more conflicted about using technology like AI before. It often feels like “cheating” in many ways, with the sheer volume of architectural iterations that can be produced within minutes.
That raises the point in my primary concern with using AI…
AI produces A LOT of “Worthless” content.
When I’m pumping out images in MidJourney, a lot of sifting is involved. For every 20 MidJourney images that pop up on my Discord screen, only one image is actually useful.
Now, scale that — and you’ve got a lot of useless images.
Of course, it can be argued that that’s just part of the iterative design process. You’ve got to produce a lot of content, put it all on the table, and critically determine the best iteration to move forward with, given the project’s scope.
My concern is not necessarily about the quality of the content that AI produces — because, let’s be honest… the content will get better as AI evolves.
My concern is about the second layer in the iterative process — critical thinking.
With the release of ChatGPT 3, my social feeds have been completely inundated with new articles, tips, tricks, and “How to Use ChatGPT for (anything).”, or… “500 Prompts to Use for ChatGPT”, or… “I asked AI to Design 60 Buildings for Me in 1 Hour”… and the list goes on.
Just because AI can produce a lot of content doesn’t make it meaningful.
What adds that layer of “meaning” is how we distill that content into something valuable — something that somebody can use to improve a specific process.
By now, we’ve seen the power of AI and the striking images it can produce. We’ve seen the power of ChatGPT and how it can crank out paragraphs faster than we can think of sentences. But the process shouldn’t end with the product created by AI.
AI should serve as a tool to inspire the design process, not become the design process.
Don’t let AI “Be” the Creative Process, but rather a tool to support creativity.
The moment we become too reliant on AI to think critically for us will be an exponential downfall for every creative industry. When I see headlines like “ChatGPT Wrote 20 Articles in 15 Minutes”, I’m left to question those articles’ quality.
I can see a future where AI produces nearly every article published online — and the only human interaction in making the content is writing the prompt, then copying/pasting it into a post online.
I’ve seen this happening all over the place in architecture and design schools with MidJourney images — and even beyond the boundaries of the digital universe.
Now we’re seeing AI everywhere with written content.
The responsibility is in our hands, not AI. The more content we consume, the more we need to question it. And with the implementation of AI in content production, there will be a massive influx of “thin” content in the future.
AI In Architecture: The Unknown
A lot needs to be discovered about the future of AI for art and architecture. It’s difficult to predict the trajectory of technological advancement in AI — but given the information that I shared earlier, it’s evident that AI is rapidly evolving and becoming better.
So many other industries outside of architecture and design have used AI for years, like healthcare and transportation. But this is one of the first times in history that AI has directly impacted how we conceive architecture and design ideas.
Here are 3 “unknowns” that we should pay attention to for the future of AI and architecture:
- Ethical Concerns. There are questions about the potential for AI to be used to perpetuate social biases or to infringe on privacy. It is not clear how these issues will be addressed in the future, or what the consequences of ignoring them might be.
- Job Market: There are concerns that AI could lead to widespread unemployment, as machines take over tasks that are currently done by humans. On the other hand, there’s also the possibility that AI could create new job opportunities, as businesses seek to integrate these technologies into their operations.
- Societal Impacts: It’s not clear how the widespread use of AI might change the way people interact with each other or the way that they perceive their own role in the world.
It’s really exciting to consider the potential possibilities with AI in architecture, but it’s also important to approach the development and use of these technologies with caution and care.
As far as ethical concerns go, the biggest concern right now is about whether or not AI should be regulated.
Should AI be Regulated?
People have suggested that AI should be regulated in what online sources it pulls data from — websites or social media. In fact, The Harvard Business Review states that “AI Regulation Is Coming” and that people have been growing more concerned with the use of AI over the past decade.
The question of whether AI art should be regulated is a complex one, and there are valid arguments on both sides.
On one hand, some people believe that AI art should be regulated in order to protect the integrity of the art world and to ensure that artists are properly credited for their work.
They argue that AI art has the potential to mislead audiences, who may need to realize that a work was created by a machine rather than a human artist. Additionally, they argue that AI art could be used to deceive collectors and buyers, who may be willing to pay more for a work if they believe a human artist created it.
On the other hand, others argue that applying regulations to art or any creative endeavor would be a significant step in the wrong direction.
Art is about expression — and in a lot of ways, there “are no rules” when it comes to creating art.
They point out that other art forms, such as digital art, have yet to be regulated and that AI art should be treated the same way. They also argue that regulating AI art could stifle creativity and innovation and prevent artists from exploring new techniques and styles.
Ultimately, the decision of whether or not to regulate AI art will depend on the specific context in which it is being created and exhibited.
It may be necessary to establish clear guidelines and standards for the use of AI in the art world, in order to ensure that artists are properly credited and that audiences are not deceived.
💡 Jon’s Take: The Future of AI in Architecture
AI is not the end but the means.
We should use AI to help inspire our design process or to help us break through creative blocks. But we can’t simply exchange something so innately human — critical thinking — with the “shiny object syndrome” that AI is giving us regarding productivity.
It’s great that we now have tools to help speed up our iterative design process, but that should only open the doors to being able to distill that content into something truly groundbreaking…
Regardless of where you personally stand on the conversation, one thing is apparent.
AI is here to stay.
It will continue to find uses in other creative industries to help increase productivity and inspire new ideas.
If you want a bit of solace when thinking about the future of AI, remember that AI isn’t self-aware. It doesn’t know of its existence or what it is creating. As humans, we have the luxury of self-awareness — and self-awareness is arguably impossible to replicate using technology.
AI is part of the creative process, not THE creative process.
Regarding AI art generators specifically, this is an excellent time to reflect on what we value when it comes to art and what we value in designing architecture. No matter how AI evolves, there will always be one key ingredient that makes designing architecture a uniquely human ability.
The design process!
Curious how architects can integrate AI into the design process? Check out this post:
If we want to “play the game” with AI in architecture, we need to get a better understanding of how AI works. We keep an open mind to where technology can take us in the future.
Those who don’t want to “play the game” with AI design certainly don’t have to.
But be aware that this technology is rapidly evolving, and now is the time to get a basis of understanding how it all works.
Here’s what I’ll say for anybody who is concerned about the future of AI in architecture:
Like with any tool, becoming a “master of AI” requires understanding exactly how we can use it in the design process. Rather than being intimidated by the technology, embrace it. Learn about it. Discover the roots of how it was created, who created it, and how it operates.
The way by which architecture is created is directly connected to technology.
Think about the evolution of design technology and how it has shaped how we create architecture. CADD software has made it exponentially faster to produce design concepts and drawings, but this technology has only existed for the past half-century.
Yet, the seeds of architecture were planted thousands of years ago — and many of the original architectural precedents of our ancestors still inspire modern architecture today.
Technology is ultimately a catalyst that will help us reimagine the ways in which we create architecture.
Architecture shouldn’t simply appeal to our visual sense. It should strive to evoke deep feelings and thoughts in those who are experiencing it.
Computer screens can’t replace physical architecture, and to call any image produced by MidJourney or other AI graphic software “architecture” would be misleading — and frankly, untrue.
Architecture isn’t solely about producing pretty pictures or fancy renderings.
We need to constantly question our process.
How was this architecture created? Why was this architecture created? What does this piece of architecture mean to different people based on their unique walks of life?
What we choose to value in the architectural design process is important. Architecture should tell stories and connect with people.
AI is great for tossing ideas around and expanding inspiration, but at the end of the day, having a unique creative process will set people apart and maintain the “humanness” behind art and design.
The key is to learn how it works, why it works, and how to integrate it into your creative process. Embrace AI as a step forward rather than a force to replace us.
Don’t be discouraged by AI in architecture — use it as a tool to reimagine the future.
Jon Henning
Hi, I'm Jon. I write about emerging technology in architecture, engineering and design, and I want to help you push boundaries with the latest tech trends in the AEC industry.
Want to step up your AI Architecture Visualizations? Grab a copy of my FREE eGuide, which contains over 100 examples of AI Architecture prompts to use with MidJourney 5.
