Episode Four: Architecture and Design in the Era of AI
Artificial intelligence is reshaping society, influencing creative processes, and presenting new challenges and opportunities. The crew digs into the impact of AI, from practical applications to ethical considerations, diving deep into how the technology is affecting architecture and design for professionals and enthusiasts alike.
Nick: Welcome to the Arcfluence Podcast, where we chat about how design impacts the way we live, work, and invest. I'm your host, Nick Karakain, along with my amazing co-hosts, Paul Fatkins and Nino Samardzik. Today we have a guest, Daniel Porter, who I've known for 20 years. He was in my wedding. He was my co best man along with Paul. He lived with us at a certain point in time when we were all in college. I feel like you've all been my roommates at one point in the last 40 years. Nino, have we lived together? Nope.
Nino: We never lived together, thank God.
Nick: Well, hey, Danny, why don't you tell us a little bit about yourself so we can dig into this topic?
Danny: Yeah, I'm Danny Porter. I have 15 years now in experience in digital communications, and I've worked in a mixture of nonprofits, political campaigns, now higher education, and so I've crossed a lot of the field. But I do a lot of work with CRMs, front-end development, marketing, and, most recently, in graduate school. I spend a lot of time in Unity Engine and Figma doing user experience design and programming, and so this is such an interesting topic to me, and I'm super curious to hear how AI, which is such a groundbreaking thing, has affected your career in architecture and design. But I think I would say first and foremost, it's important for people to know that machine learning and AI have been around for years already in medicine, research, and data analysis. So, it isn't new. The recent shift has been in its commercialization, like in language and art, its presence is now palpable. It's everywhere in our daily lives. I remember, Nick, when ChatGPT, the first version of it dropped, we were all, you had me, Nick, and Paul, we were at Galaxy Con in Columbus, it was a science and futurist convention, and this Turing test passing AI just drops, and it kind of blew everything up. And I remember sitting at a panel like, "Nick, Nick, look at this. This is crazy. This is going to change everything." And he's like, "Yeah, watch the panel." And I had been staying apprised of chatbots, I think my whole life out of curiosity, but I'd never encountered something that was this level.
Nick: Well, I got to tell you, I sometimes struggle with organic intelligence. So the thought of artificial intelligence is a little bit above my pay grade. So that's why I associate myself with fine folks like yourself. So you can lead the charge. Danny, for you in your fields, where are you seeing it in your industry?
Paul: I mean, we can certainly get into it. I wanted to take a step back. Let's just start with even machine learning, right? You mentioned that term there, Danny, you kind of tossed it out there. Maybe people don't know what machine learning is. It's like you hear it and think like a computer goes in a room, closes the door with a bunch of books and some coffee, and teaches itself how to do stuff.
Nino: So is machine learning the same as deep learning, or is there a difference between these two things?
Paul: It's not exactly right. Like, again, that's what you think, like the machine is teaching itself how to do this stuff, but really, there's data that has to be put into it, right? For it to even have something to reference. So whether it's some script, like Danny's using it with his stuff, or some image that someone's using to generate some design for their backyard patio area, right? There still has to be some input for that machine learning to occur. So it's not really teaching itself anything. It's really harvesting data that's out there and people are inputting data into it. So there has to be some input but still has to put this data in some kind of orderly manner in order to create the context, that's part of the intelligence. You're getting into classification, right? And then again, like you get into where some of these hallucinations occur, which is like a mistake because maybe there's an issue with the data that comes in, how it's classified, and then how it's utilized in the output, right? And it's like, whether it's a design issue where it's like someone's using AI as a tool to design something or some scripting thing like Danny might be using, again, you have some input that might be classified wrong. And then some output that's wrong. Cause Danny, in your experience, I mean, if you're using AI, is it right 100% of the time you never have any issues with it, or do you still have to go in and check your work if you ask it to do something?
Danny: A hundred percent. You touch on a great point. So, ChatGPT as it exists now is a lot like a very complex version of, you know, on iMessage when you see the next predictive word? When you're just typing a text, it's basically that on a much, much larger scale. And so when you submit a prompt to ChatGPT, it's breaking up all of those words and text into little tokens. And then it's saying, okay, based on that prompt, what are going to be the next tokens? Nick, it might help you with the stock market. If you already know a lot about the stock market and you can prompt it with ways to help you think about a problem or reconceptualize a problem. But as Paul mentioned, it's going to make a lot of mistakes. So if you go in, you're like, "Hey, I need you to build me a great responsive website." Even if you know what you're talking about, you're not going to get a very good output. It's not going to be even usable. But if you're going in and saying, "Hey, I need to address this problem or retackle this issue and you already know a lot about JavaScript." Well then, you're having more of a collaboration. You are the expert knowing exactly what you need to do. And it might be a little more valuable to you, but you're going to have to check its work. What it's trying to do is to present back to you based on your prompt, the most probable and believable answer to your query. Paul, I think one of the misconceptions, to circle back to the design world, is that people think now with AI out there in the world, you just push a button, get your result, and you're done. It's like you don't have to do anything else.
Nino: I think that's because they use the word intelligence, which is a very sophisticated term. It doesn't only include dealing with data but also emotions, among a million other things. Maybe AI is on its way to becoming intelligent, but using the word intelligence is somewhat misleading. Intelligence is such a complicated concept; we even cannot grasp how our brain fully works. So, it's more the appearance of intelligence from what I'm hearing. Nothing against those smart people in Silicon Valley and elsewhere, but sometimes their perceived world from the basement, when they spend a lot of time barely socializing, creates something new. I'm just kidding. But creating something new doesn't necessarily mean having a full grasp of 'intelligence' in its entirety.
Paul: And it's interesting because you're saying, part of your design process when you start out is to look for inspiration, right? And essentially, yeah, I think that's what people are doing with these kinds of limited tool sets. They're enhanced automation, like you said, where it's able to gather all this inspiration together and then output 15 different iterations of something so that maybe it sparks something in you, the designer, to then say, "Okay, this is the direction I want to go in the path I want to follow." It doesn't necessarily give you the magic bullet where it's just done, but it points you in the right direction. But at the same time, I don't want to sound like an old fellow who's scared of the new technology that's coming because this technology will rule the world. I mean, it's already present in areas like the military. But imagine the world without GPS, right? But at the same time, if you put a little more guidance into how we can utilize this in design, it can be extremely beneficial, especially with collecting data. Because as we know, every design is based on location, and every location is unique. We can gather tons of information for that specific site, from weather, prevailing winds, sounds, then we can gather all the code-related stuff regarding that specific area. Instead of spending a lot of time on research and data gathering, which we have to do as good designers, artificial intelligence can help us in that path.
Nick: So, we have determined that AI definitely helps us magnify our work or save us time, right? And gives us a little extra juju for those who aren't maybe at the top tier of their industry. But I know Danny and I have had tons of discussions on how we're now bringing this into our workplace in terms of working remotely and how we actually interact with our computer and how we're staying organized and our creative process, whatever it is we do on the daily. Danny, I know you've done a bunch of different things to magnify your time. Would you mind getting into that a little bit?
Danny: Paul and Nino both make great points. It's not really an artificial intelligence. And I think one of the mistakes we can make when having a conversation with any large language model or even using a programming tool that’s AI is to not anthropomorphize it. It is a statistical tool and it's basing it off of your prompt, and it's not going to be perfect every time. But it can be a good way to expand your thinking space and do a little brainstorming, or if you're a little stuck, like, you just need to put some thoughts out in the ether and start chewing on them a little bit. Now, in my space, I like to automate a lot of things that I can.
Danny: I've always been of the mind that if something is being copied and pasted more than once between one spreadsheet and another, that's a breakdown and a waste of time. I agree. So, any tool I can use to churn out more automations, eliminate more time-consuming tasks that shouldn't be time-consuming, I'm going to use. And AI definitely can help with that. I think that's how I approach it in my day-to-day. The perception of it being a quick fix, which I think Polly mentioned, is evident on Facebook today. Every single ad about AI suggests it will do your marketing for you or replace other tasks. But it's only helpful if you have a lot of grounding knowledge in the topic because it responds entirely to your prompt. Granted, it has the whole wealth of human knowledge behind it, making it earth-shattering. With any field, whether it's architecture or education, viewing AI as another tool for humans will put us in a better place. When you start thinking it's going to do everything for you, you'll find it won't.
Nino: And this is a good point because we, as a society, need to conquer all these universes beyond planet Earth, and for that, we need help. Don't get me excited, but honestly, if it can avoid tedious tasks, we have more time for our creative juices to open new horizons. That's a valuable use of this new super technology. I remember when I bought my first iPad, I just used it for viewing content. It was hard to work on it, like converting PDFs. Then the second generation came out with the pencil, and I started sketching. Now, I don't even have the latest version, and I cannot imagine my life without it. It helps me a lot, and now it's indispensable for my sketching work. So, the more we use and accept this new technology, the more benefits we'll see from it.
Paul: I liked what you said earlier about needing to work with it despite some limitations. As time goes on, it will evolve and become more refined. Think about Tesla and their self-driving software - where they started and where they are today shows the progress. Even now, when you do a Google search, they have the AI-generated responses. It's amusing, but it points us in the right direction, starting us on the path we should be walking down.
Nick: We need a way to save all of Nino's sketches since you do sketch the most. They're all in my head. We should find a way to implement those into some type of system.
Nino: I mean, everything is in the cloud, but my fear is that one glitch could erase everything. My AI is good until it's unplugged from the power. You know, we're so impatient as a society today. We just read titles and don't take the time to digest content. This is quite common since I work as a project manager and see it in young staff coming right out of college. They're smart and hardworking, but understanding context and expressing emotions is a gap. It's about being a good listener because there's a generational gap. Bridging it involves patience and listening.
Danny: What this will probably mean is that a lot of people in education will need to rethink how they approach assignments. Students are going to use it. If they understand how it works and where it excels and doesn't, that's crucial. Maybe part of an assignment is having a conversation about the era and history and submitting that conversation, showing engagement with the topic. We need to be really creative, for better or for worse, because it's here to stay.
Nick: This got me thinking about using AI for helping with email communications and conceptual design. What if we empower our reporters or potential clients to start the conceptual design process? There's enough information out there now that they can use AI to generate space with the elements they love. So, is there a way to put together a type of mood board that our potential clients can create on their own and then submit to us? It would help us understand them better and deliver a product in line with their visions and expectations.
Paul: Yeah. I mean, I don't see why not. Essentially, it depends on what your input is. So, you refine your input. Maybe it's like, "Oh, I want more of a warmer palette." So you have some initial image that's created by AI, and then you refine it by making it warmer.
Nino: Yeah. Remember, especially when they have more costume-based customers, they want to build something from the ground up. They will actually send us or email us some inspirational photos. Do you remember when we were working on that property in Hockey Hills? They were actually inspired by one of these properties they visited, not too far away. So, yeah, I see that and I agree with Daniel, definitely, turning back to something that is coming isn't a good approach. But let me compare, go old-fashioned because now we are in a transitional phase. And the better we learn about it, the better we can even help that artificial intelligence help us do another task. If you just turn it back, it's going to develop anyway. But it's not going to be the tool that's going to help us more because it's developed by someone who doesn't really fully understand your role. You know what I'm saying? Right. And all this is good to be on top of the wave.
Nick: Absolutely. I mean,
Paul: Well, and that's why it's like, you know, you don't see the process in the factory of how things are made. So you don't really truly understand what you're getting. That's why prompts are so important. Some advice to give to somebody who might try to utilize this to give us inspiration, if they're trying to use one of these image generators or whatever, something to come up with a description of what they're looking for based on their area, the location of the property, et cetera. Be careful with your prompts and prompt them again and again. And again, you might have to do it four or five times until you get something that's kind of what you want or what's in line with what you're thinking, right? It's not just going to spit out your answer and you're done.
Danny: Um, they just did a lunch and learn series where computer science professors did a whole series of zoom chats just talking about how AI works, how it's structured, what's going to happen. And then the communication school, emerging technology, and they're constantly thinking through and doing work on from the non-technical side but more what are the implications for society? And to what Nick and Paul, what you guys have been talking about with having people enter design ideas. I think if I were talking to someone that were going to use this for design ideas, they should know two things upfront. One. It's going to skew positive. So if you tell it, "Hey, I have this idea and this is what I want to do. And I want to put a giant foam camel in my living room," Chat GPT is going to come back and say, "That's a great idea." But you should be aware of that because it's responding to your prompts. So if you come in really full of excitement about your theme, like my Star Trek themed Airbnb, it's going to tell you what a great idea that is and how you can do it. And it's going to build on your prompt so just knowing that upfront that this is, this is a tool for maybe you to expand your own ideas and not to go like, "Is this the best idea in the world?" That's how it's built. It's very nice. Yeah.
Nick: So essentially, You still need to work with your design professional, to put you back to reality. Not only that, just construction documents in general with code changes and zoning, that's something that's still kind of out of reach. Constant conversation with people on the field. But at the same time, so it gives you the client's perspective. Maybe he wants to have, that's what interests me.
Paul: I love that point. Like experiencing the space, right? You put something into a computer, the computer doesn't go on site, walk the site, feel what the space is, feel what the ceiling height is, look at the natural light coming into the space, you know? Even what you're inputting is very limited to these tools because they can't go out in the field and do these things, it's a computer somewhere that you're just prompting and it's giving you a reply. And like Danny said, it's a positive and reaffirming reply. I think for the architecture industry, you have to create these digital twins of the space. Yes and input a mountain of data of what that space is into the computer to get something useful out of it.
Nino: And when we are talking about the giant camel, I just remember last summer I was in Punta Cana. We went to this fancy restaurant. It was like a giant giraffe holding the chandelier in her mouth. So it was like the whole space was dark black, and there, in the middle, is a giant giraffe holding the chandelier. I'll send you pictures. That was their wow factor. But, as Danny said, we didn't know how these things are going to look in five years or even in a year. It can be more breakthroughs. And we have to be aware; we have to educate ourselves on that. Honestly, I think we, as designers, have to be on top, because this is the new wave coming. We have to be aware of it and we have to have a much deeper understanding of the technology and what is coming. We still don't know what's going to roll, but we have to be more prepared than less prepared.
Nick: Send me a picture. Seems like if it's a short-term rental or a hotel, it might make sense. It was the restaurant go. It was like a high-end restaurant. They've already monetized it. Yeah. So, I'm seeing a lot of education courses now on how to leverage it. Paul, I'm going to put you on the spot. What are our next steps and what can we do to be knowledgeable about this wave that we're in right now and how can we implement those attributes?
Paul: It just goes back to Nino's point, right? We just don't ignore it. You try to leverage and utilize it and understand it. But again, even five years from now, like Danny said, we don't know what that future is going to be or how good it could be. Maybe it can spit out a set of drawings that's good enough to get a permit. But in my mind, I still think there's going to be difficulties for it to get to that point because it can't go on site and experience what that space is. To take that out of the equation when you're trying to design and tell the story of a beautiful space without having to go and experience it, how are you going to take a ton of photos and scan it? And that's, that's the emotional portion of it that we can help with. And as Danny said, it's going to be as accurate as the data it receives. So, if the data it's receiving is not comprehensive, then God knows what you'll get.
Paul: I think for the architecture industry, you have to create these digital twins of the space. Yes. And input a mountain of data about what that space is into the computer to get something useful out of it.
Nick: Which is basically what we do anyways when we do a 3D
Nino: digital model. But also, there is a shift. Both me and Paul have been doing all this from one level of rendering. The next level would be super high, like realistic renderings. And now we see the shift to going big, less realistic because there were drawbacks because clients were perceiving, "No, I don't really like that shade of that material. Can you do this?" And our side was like, now we have to produce these very high-quality renderings because this is what the client is expecting to receive exactly. Like, why is this tile rectilinear, you know, versus octagonal? It needs to be octagonal, and you build this space, and he's going to put your rendering against that space, and it's like, the light is different, the color is different. Exactly, the mood is not the same. That's a great point. Then we were like, okay, let's tone this down. Let's try to just send the general feeling of the space, the design. Because you can easily rabbit hole, then be right there on a site and being scrutinized for not delivering what you promised. You know what I'm saying? So, every technology has a double-edged sword, but we're still going to use it. I think it's going to change the world, of course. Like the internal combustion engine and the internet changed the world. And these things may have even bigger potential, that make really big statements.
Danny: You know, I want to say to you, architecture, fields of design, user, these fields have always been about pushing the limits, innovation, change. One thing that's one thing, and this is what I would also tell young people is that AI doesn't innovate, it builds off of what already exists and reassembles and rehashes. And so it still takes human thought and human ingenuity to create something new and to contribute to the field. And that's where you need professionals and designers that aren't just plugging in prompts and using what the output is verbatim. So again, I think you're right. It can be part of maybe some of that process, part of the automation of things, but it'll never really ever be a replacement for the creativity of human beings. Oh yeah,
Nino: and when you see it today, we still have especially in the architectural profession, how easily we gave up on the residential world and just let it have cookie cutters placed all around with just different versions of the colors. We already have that. I mean, people are like, okay, I'm fine with it. That's how much my budget can afford. Not everybody can afford the architect. But that's a good point. If you're a developer or builder and you're developing this neighborhood and you have five or six different designs that have subtle differences so that you can customize it, so it all doesn't look the same. I bet AI would then say, here's all this information. Now how can we really make this neighborhood feel more unique? That's why I love living downtown in bigger cities because everything is different. But, people have to go through trial and errors to create something. First, you're going to say, I want a big house just enough for our kids, and I live in the suburbia. There is no creek in the backyard anymore. I have tons of neighbors. But the house is big; it serves the purpose. And people eventually will say, after listening to the same tune over and over again, let's listen to something else. So, don't be scared of these technologies. It will just help us open new creativity levels. It's scary for me too. It's going to just produce the documents and I'm going to be useful if I know what that thing is going to deliver and I'm top of that game. And now it helped me to explore, to unlock another level in my brain that I never even challenged before.
Nick: I got to tell you, I'm excited about all this. I mean, I grew up with a rotary phone and, in 40 plus years, the world's different, completely different. So I want to see what it's like in 40 more years.
Nino: Yeah, he's right. There is something in us as human beings. We should not ever try to like look for an alternative because we can be the most destructive force on this planet but also the most creative one too. So I still have huge faith in humanity, especially in the profession that we are doing, but we just have to every day be students, learn and deliver more. Always ask ourselves. I know when I design something, it's phenomenal in the evening. I wake up in the morning, it looks like crap. So, you know, honestly, you have to be otherwise you're never going to push yourself and just enjoy it. Have fun.
Paul: I like that. All right. So I think we'll wrap up there with what Nino just said. But, if you guys are out there listening to us, be sure to like and follow us. We'll see you on the next one. Take care. This has been the Arcfluence Podcast. Please subscribe on the platform of your choice and leave us a review. We'd love to hear from you. As always, you can get more information and additional content on arcfluence.com. I'm Paul Fatkins. I'm Nick Karakaian. And we'll see you next time.