“As we have been discussing so much about digital design and architecture, I should say that nowadays, AI technology is becoming so popular and widely used in the design industry. It is so phenomenal, and its rapid growth far exceeds all the recorded advancements in the world of digital technology. It is so imminent that even non-professionals can create something that appears professional to many, not only in architecture and interior design, but in almost every field: from song composition and music arrangement to creating artworks, graphic design, and creative writing, among many others. Now, we couldn’t hide the fact that it seems so threatening for designers like us, huh? What do you think? Let’s talk more about it on the last episode of this series. Thanks for following.”
As I have always said, “I have nothing against the digital method...” and now we have finally arrived at the most awaited discussion on the role of Artificial Intelligence (AI) in the field of Architecture. In this article, we will explore the term “machine learning,” a subfield of AI, based on a conference presentation written by Mr. Giuseppe Gallo, a PhD candidate in Architecture at the University of Palermo, Italy, and submitted to Academia.edu. In my interpretation, it is through the subfield of machine learning that AI is being applied to the architectural design process.
OBJECTIVE
Although AI is being highlighted in this article, I would still maintain my full support for the importance of manual sketching in the design process. After all, that is actually my main goal, or should I say, an advocacy that I have supported ever since I started discussing this subject matter. Moreover, some parts of this article shall serve as a commentary on the writings of Gallo in order to help us understand more deeply how AI affects the architectural design process. So, some excerpts may be expected occasionally to meet our objective.
According to Gallo, “The proliferation of data together with the increase of computing power in the last decade has triggered a new interest in artificial intelligence methods. Machine learning and in particular deep learning techniques, inspired by the topological structure of neurons network in brains, are omnipresent in the IT discourse, and generated new enthusiasms and fears in our society.” And so, this article is somehow aimed at alleviating that growing fear, especially to the professionals who have spent a lifetime of devotion to the field using just their acquired knowledge and manual sketching method.
DEFINING ARTIFICIAL INTELLIGENCE
According to Wikipedia, “Artificial intelligence (AI) refers to the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and decision-making. It is a field of research in computer science that develops and studies methods and software that enable machines to perceive their environment and use learning and intelligence to take actions that maximize their chances of achieving defined goals. Such machines may be called AIs.”
For me, I still find definitions on Wikipedia sound a little technical in nature, and you still need further reading or more research efforts to get the definition you need, regardless of the specific purpose. I have read the definition of AI in the Academia.edu article I am referring to, but somehow found it a little profound, receiving a combination of opinions from different architects being interviewed at the conference. Most of them are pointing to different subfields of the subject AI, and so, it prompted me to consult with ChatGPT about the different subfields of AI, and here is exactly what I got (exactly what the selected architects interviewed by Gallo are trying to say):
Artificial Intelligence (AI) is a broad and diverse field with many subfields. Each subfield focuses on different aspects of intelligence and problem-solving. Here are some of the major subfields of AI:
1. Machine Learning (ML)
• Definition: A subset of AI that enables systems to learn and improve from experience without being explicitly programmed.
• Types:
○ Supervised Learning: Learning from labeled data to make predictions.
○ Unsupervised Learning: Learning from unlabeled data to identify patterns or structures.
○ Reinforcement Learning: Learning through trial and error, receiving rewards or penalties based on actions taken.
○ Semi-supervised Learning: Combines labeled and unlabeled data for learning.
○ Deep Learning: A type of machine learning based on neural networks, especially deep neural networks.
2. Natural Language Processing (NLP)
• Definition: Focuses on the interaction between computers and human language.
• Applications: Machine translation, sentiment analysis, chatbots, text summarization, speech recognition, and more.
• Tasks: Syntax and semantics analysis, text generation, language translation, etc.
3. Computer Vision
• Definition: AI field that enables machines to interpret and understand the visual world.
• Tasks: Image recognition, object detection, facial recognition, video analysis, and image generation.
• Applications: Autonomous vehicles, medical image analysis, security systems, etc.
4. Robotics
• Definition: Focuses on creating intelligent machines that can perform tasks autonomously or semi-autonomously.
• Applications: Manufacturing robots, drones, medical robots, service robots, and more.
• Sub-disciplines: Kinematics, control theory, and robot perception.
5. Expert Systems
• Definition: AI systems designed to emulate the decision-making ability of a human expert in specific domains.
• Components: Knowledge base, inference engine, and user interface.
• Applications: Medical diagnosis, technical troubleshooting, financial planning, etc.
6. Cognitive Computing
• Definition: AI systems designed to simulate human thought processes and decision-making.
• Related Technologies: Natural language processing, machine learning, and reasoning.
• Applications: Personal assistants (like Siri or Alexa), decision support systems, and intelligent automation.
7. Knowledge Representation and Reasoning (KR&R)
• Definition: Focuses on representing information about the world in a form that a computer system can use to solve complex tasks.
• Tasks: Logic-based reasoning, ontologies, semantic networks, and probabilistic reasoning.
• Applications: Question answering, inference systems, and AI reasoning tasks.
8. Artificial General Intelligence (AGI)
• Definition: A hypothetical AI system that can perform any intellectual task that a human can do.
• Research Goal: Achieving human-like cognitive abilities in machines.
• Current Status: Still theoretical and not yet realized.
9. Fuzzy Logic
• Definition: Deals with reasoning that is approximate rather than fixed and exact, mimicking human reasoning.
• Applications: Control systems, decision-making, and systems where precision is not required.
10. Swarm Intelligence
• Definition: Inspired by the collective behavior of decentralized, self-organized systems (e.g., bird flocks, insect colonies).
• Applications: Optimization problems, robotic control, and network design.
11. AI Ethics
• Definition: The study of ethical issues related to AI development and usage.
• Issues: Bias in AI models, data privacy, accountability, AI fairness, and ensuring AI systems are aligned with human values.
12. Neural Networks
• Definition: A computational model inspired by the way biological neural networks in the brain process information.
• Applications: Deep learning, image recognition, natural language processing, and more.
13. Automated Reasoning
• Definition: Focuses on the automation of logical reasoning and proofs.
• Applications: Theorem proving, verification of software/hardware systems, and automated decision-making.
14. Human-Computer Interaction (HCI)
• Definition: The design and use of computer technology that interacts with human users.
• Applications: User interfaces, voice assistants, and smart devices.
These subfields often overlap, and researchers may work in multiple areas simultaneously. Together, they contribute to making AI systems more capable, adaptive, and intelligent.
I think I am now finding myself getting too far from my real objective. But somehow, this time I just have to use my intuition to simplify the definition. Let’s just begin with a machine that can think like a human. Perhaps now, you are thinking about a robot. But this robot, although it can talk, perhaps write, or communicate with us in any way it can, still, the information it could provide will depend on the data received from humans, initially from its creator or inventor, then from the users themselves. The learning process could be both supervised and non-supervised, allowing the AI to process available data by itself or while being trained by the inventor or the users. In the long run, both the AI and the users tend to benefit from each other and develop a reciprocal and infinite learning process. Now, in the light of architecture, I would rather attribute the traditional sketching to human ability and the parametric architecture to AI ability. For me, both are useful tools in the design process and cannot be separated from each other. Well, by integrating some of the technical definitions initially derived from research, I hope I am able to satisfy the objective that we are trying to reach here.
WILL AI BE USEFUL IN ARCHITECTURE?
This was an important question Mr. Gallo asked when he interviewed ten architects of different specialties from February to July 2019. It somehow turned into a survey where the participants tend to vote on certain categories. So, to cut the story short, here’s a table and an excerpt of the result:
“Machine Learning is the technology that obtained the highest score with a total of 53 out of 70 achievable points, followed by digital manufacturing with 51, third “other computational methods” with 47, then Internet of things with 38, BIM and Augmented Reality with 37, last Virtual Reality with 30. It is interesting to note that Machine Learning and "other computational methods" both obtained the first place in the personal rankings of the designers four times, as well as happened twice for BIM and once for Digital Manufacturing. It is therefore clear that based on the experiences and expectations of the interviewed designers, machine learning and its derivations are expected to play a role within the architectural practice, a role that, for many of the interviewees, will be decisive in ten years.”
Such a piece of concrete information somehow supports my own definition of AI, as I have provided above, and has somehow satisfied the objective of this article. However, if you still need further clarification, let's discuss further, and please feel free to leave a comment.
EXPECTATIONS OF AI TECHNOLOGY IN ARCHITECTURE
It has been five years since Gallo conducted his research work, and the above survey was projected for ten years. Here’s an excerpt:
“Architecture is a complex practice. On the contrary, sectors where Artificial Intelligences are showing an important impact, have a more linear nature than that of our profession. The interviewee goes on saying that, by breaking up the architect's work into separate tasks, describing the process rigorously, it is easier to imagine an AI capable of solving these operations individually. It is therefore important to ask ourselves several questions: Are we able to manage these enormous potentials to generate new concepts and ideas? Can we describe this complexity so that a machine can process it? Maybe in the future.”
Now that we are actually more than halfway through the projected time of the survey, do you think that at that time, they might have overlooked the capability of AI technology in the field of Architecture? The way I experienced it, it came so drastically at a very high speed, that even non-professionals could produce professionally looking products on their own. I guess that’s where it becomes intimidating for the professional community.
AI may sound intimidating due to the unpredictable speed in terms of the development of technology. However, as we continue reading Gallo’s research, he says:
“It is therefore still too early to understand how much these technologies will erode from an architect's professional practice, and certainly nothing in terms of responsibility. In this sense, Arthur Mamou-Mani declares that even by using AI, designers retain the right to control the design process at any time, making choices and questioning answers provided by artificial intelligence.”
This was exactly what I was trying to point out as I concluded in my first episode:
“In the next episodes, we can expect AI to enter the arena. Oh well, let's just welcome it, but I believe we should not let it dominate the show. Instead, let us use our own creativity and use AI as a modern tool only that we have full control of. Use it to enhance our own ingenuity, nothing more, nothing less.”
Now, it seems like the majority of the comments I heard from other famous architects add up to my confidence that I am on the right track when I say I support the advocacy of retaining the manual sketching method in the design process. For me, this is some sort of sustainability that matters in the field of architecture.
FINAL THOUGHTS
In the first episode of this mini-series, I mentioned in my conclusion that "it is the cultural identity and the sense of originality of the architect or artist that I want to emphasize and preserve in this endeavor. The bottom line is that we should stop arguing about which one is best. Let's discuss this with a sense of balance." That was when I discussed the integration of computer technology in the manual sketching method used in the design process. Well, I would say it would be the same thing in the use of AI technology. Instead of being intimidated, let us be confident that AI is a helpful tool in our professional practice. If we could train it, then we could definitely control it. Let’s be friends with them, a new colleague whom we can trust and rely on based on accuracy and consistency. But what about loyalty? Oh well, don’t you dare compare them to a colleague next to your cubicle. Just kidding aside...
Thanks for joining me throughout this mini-series. Hoping we could have more of this...what do you think?
JOEY CASTANEDA, Architect
CITATIONS:
Thanks to Wikipedia for the initial definition of "Artificial Intelligence."
https://en.wikipedia.org/wiki/Artificial_intelligence
Some excerpts and a table derived from a research work submitted to Academia.edu are as follows:
The
role of Artificial Intelligence in architectural design: conversation
with designers and researchers
By Giuseppe
Gallo and fulvio
wirz
Photos courtesy of Pixabay.
FAIR USE DISCLAIMER:
This document may contain copyrighted material whose use may not have been acknowledged by the copyright owner. We believe that this constitutes a "fair use" of the copyrighted material as generally provided for in every copyright law of the country in which it was produced. If you wish to use this copyrighted material for your own purpose that goes beyond "fair use," you must obtain permission from the copyright owner. On the other hand, if you see that your copyright material appears on this website and you disagree with our assessment that it constitutes "fair use," contact us.