The value of what we do as
architects is still the thinking. Where do you think architectural
visualizers will be in the future? Like what's going to happen with them? At the moment, they are tools. Maybe they'll cross the line
from a tool to a collaborator. Does that mean that the physical ways of
creation are also coming back as a way of pushing back on this, this new technology? I mean, yeah, if I was a rendering
firm, I would be sweating a little bit. I mean, it's true. AI hasn't really taken over. That's a good question. Now so much has been discussed about
AI and architecture and how we can potentially take over the industry. But it's one thing hearing a ton of
speculation and potential use cases. And it's another thing, actually
talking to someone who's at the forefront of this new technology. And this special collab between
Steven from Show It Better and I, we bring you Olly Thomas, the design
technology specialist at BIG in London. This conversation with Olly goes into
what the new generation of architects will likely face in the upcoming years and
how to prepare for this tech revolution. Oli also dives into what he sees as the
best application for AI in architecture at the moment and how that might evolve. And you will even hear how AI is
being implemented at one of the most innovative architecture firms. This conversation really changed
the view I had about the present. moment of AI in architecture
practice, and I'm sure it will have an impact on you as well. All right. So, but I wanted to start from the
beginning, Olly, because you mentioned something about switching from the BIM
specialist position to a more AI centered position within BIG, Bjarke Ingels Group. How did that come to life? Like what happened? Did you train to that position? Is this something that naturally
progressed into something that you're interested? I mean, I joined BIG
initially as a BIM specialist. They hired me, but they knew I had this. This other skill set of
computation for a while. Like I stayed at that
position for a while. It got to a point where we were kind
of like, I need to rename myself. And we actually see this
in the industry a fair bit. Like you're now getting people
with BIM and computation and now AI and all this other kind of stuff. So my title switched from a BIM specialist
to a design technology specialist. Now I'm a design technology
manager of the London office, which means I cover BIM computation. AI, AR, VR, and all of this stuff. It's our umbrella term for. technology that we use to design things. And more recently, I've been kind of
interested in picking up some of the AI stuff and trying to implement it into
the office, which is, uh, a challenge in itself, just keeping up with it and
getting it into the hands of designers, like as, as quick as we can really,
but no, there's no, like, there is no training to get into anything to
do with AI, I guess, apart from like. Computer science or something like
that if you really get into it so it's just been a case of like seeing this
thing being really interested in it and just trying to Keep up with what's
going on and test it and try it out. Really that that's it. Yeah. Because when we went to the office,
uh, at big in New York, we talked with Daniel and Daniel told us that they were
kind of like informally implementing mid journey into their design process, just
to have like a ton of iterations and just testing out the tools, how that was like. Three months ago, but then we see that
in three months, things change a lot. So how has things changed now? How do you guys actually use AI within
the design process if you do at all? Of course, the first thing was to get up
to speed with these things like midget. I mean, we started to latch on,
I mean, well over a year ago now. So we're talking about. Mid journey version three, when
it was almost like, uh, it kind of looked like watercolor painted style. Do you remember that? Um, and then obviously mid journey
version four came, which is a big step up. And it was around that time where I
was like, we just need to get this out. Of course, people already
testing it and experimenting it. So we use what we call a big school,
which is where we like someone from the office presents like, Hey. Uh, this is this AI thing. I kind of give a background of what
AI is, a bit of the history of AI and a much broader perspective of like
what is potentially around the corner. And then more specifically, like,
uh, this AI, this mid journey thing. And then we, at that time we were
kind of comparing DALI, mid journey, stable diffusion, and the first goal
was to just like, get to grips with it. What happens if we type in Björk Ingalls
group and does it, does it understand our Design or not, um, and get it into
the hands of the designers to just try it that some of them used it on the product. Some of them kind of just
played around with it. People created funny memes. People created like fun
architecture projects on. That's always the goal is to just it's
almost like, uh, catching a rugby ball and or like I guess for those of you
in America, like American football and passing it on to the designers and
teaching them how to throw it and use it right where we are now, I think we're
really still experimenting with these things, like constantly seeing how the
different teams are using it in the design process, keeping up to date with all the
new stuff that's coming up, like I see you guys were experimenting with D5 rendering
and they have this new AI component. So we quickly need to test that
and see, is this something we can get into the hands of designers? So I think we're still very much
in the process of testing, keeping up to date and really seeing how
it fits into the design process. There is a little bit more of tinking,
tinkering, where we're kind of like. We did experiment with connecting our
design tools to things like ChatGPT or, uh, StableDiffusion when we can connect
Rhino and Grasshopper with these things. Um, and these are always like tests,
you know, that we're not software developers, so we're kind of like
hacking these things together. Can we send a screenshot from
Rhino to Stable automatically? Um, at the moment, of course,
we can't do that with... With mid journey, but it's very
much things like that, like really seeing how we can integrate
it into the design process. Of course, I think we're at the point now
where, and a lot of firms are doing this is like now training your own model on the
recent, the things you have in the office. Of course, you want this to be somewhat
contained, like it doesn't get out, but, uh, but that's, of course,
you know, with, with stable people, train models in different ways. You can download them for free
if they make them public, right? So, uh, you know, we're seeing lots of
firms training their own models with the materials they have and seeing how
that integrates into the design process. So it's constantly a trial testing,
integrating and evaluating how Thank you. Useful or how disrupted it is to
a design process at the moment. So that means that it's mostly
using the first stages of a design process in the preliminary stages,
more about iterating ideas. But do you see a place where AI
can, can take time consuming tasks or like those more laborious tasks? As in detailing, uh, maybe scheduling
and all of that, or is this something you could be experimenting with? So I, I think I'm trying, I've, I've
been trying to like categorize all of these AI stuff, these AI tools, right? And let's take mid journey, for example. I think Mid Journey is most
powerful as like idea iteration. It's great to like, you come up with
an initial idea and you can maybe get close to that and most importantly,
visualize and communicate that idea. Um, at the same time, the value of what
we do as architects and particularly at BIG is still the thinking. The, the researching behind these things
and finding out all about the, the site and the, the things around it and coming
up with this unique solution that is based on thinking that's where architects,
uh, create the most valuable and the biggest value and truth be told, this,
this mid journey image creation side of things, It's not necessarily affecting
that because these images are thoughtless in like there is no thinking behind it. It's an image. Yes, the end result of our design
process is an image, which is a render or diagram or something
like that, but really the value. Is in the thinking behind it
and that hasn't changed like mid journey has not disrupted that,
but it has been an interesting, like sidekick for idea generation. Like it, it could be, you know, you
could type in and explore like, uh, a particular thing you're thinking
about the project and it may spark an idea or something like that. So I do see a lot of architects
talking about mid journey is like a super Pinterest. It's kind of like, uh, yeah. An idea in a more curated version. Exactly. Yeah. And, and of course the, the
more control you have over it, the more powerful that becomes. So there's one category,
which is like idea iteration. And I think, um, maybe mid, mid jetty kind
of falls into that category, for example. And then the next way I kind of see
it being used is, um, Almost like, uh, AI automation, and we're seeing bits
of this, I guess, um, the simplest and dumbest way is like, you can now go to
chat GPT and, uh, you write in a bunch of text and it formats it for you. Obviously, that's a very simple way. But it's like, it's saving you the
time of rewriting things or rewriting things, or maybe go to chat, GPT
to look up, um, the built building regulations or something like that. And it actually gives
you the correct results. Of course you have to cross check
it or something like that, but yeah. Um, we're also seeing these like apps
where, um, they're kind of doing plan generation and all this kind of stuff,
which is, it's not quite really using AI, but it's giving us a glimpse into
the future of automating some of the stuff that we spend a lot of time doing. Uh, I always talk about like a
really dumb example is like, Mr. Clippy, you. I don't know if you guys
might be too young for Mr. Clippy, but I'll say that. No, no, no. Talk to me about Mr. Clippy.
I don't remember. I don't remember. I do, I do. What happened to Mr. Clippy, man? He's gone. But like, I imagine Mr. Clippy could come back in Revit and be
like, Hey, and automatically Adjust your stairs so that they're building, they
meet building regulations in the part of the world you're, you're working in. Right. So we then have like the AI automation
side of things, which, um, I guess chat, dbt would be the simplest example. You see a lot of people building
plugins that are now AI generated, but really they're just connected
to the chat, dbt API that I don't know how much are really using. And then finally, I think you've got maybe
what I call like the AI visualization side, which is almost also flirting with
like the automation side, where It's a bit more stable diffusion, I guess now, uh,
mid journey where you can select things. You got RAs, uh, look, look X ai, the
d, the new D five render rendering, uh, AI where you can be very much more
surgical on, uh, even the Photoshop ai, like you can be very surgical on, uh,
enhancing your visualization process. Let's just take the Photoshop
AI one that has now. I no longer need to spend hours looking
for a person sitting down in the correct position at the correct sun. Yes, well, but I was
going to ask you this. Have you, have you had any
luck with adding people? Because every time people, they, uh,
Photoshop generative fill has a, has a hard time in getting Yeah, it seems
to be anything with people in, in Any of the AI stuff seems to be, there's
something slightly off on the side of the head or something like that,
but, but that will only get better. Um, but of course, sometimes it can be
really simple things like landscaping or trees and, uh, and all this kind of stuff. But again, that's, that's like a bit
of a dumb example, but I think we're, we're potentially the most powerful
at the moment is that middle ground where we're starting to see, like,
you feed it a image from Rhino from a massing or something that is, that is
Based on your thinking, which is where you're adding value as an architect. And then it's, it's quickly allowing
you to ideate through different facades, like different moods of the image. Uh, I think, I think it was Steven. I was watching your video on where
you, you had like, you'd create an image with D five and then you'd gone
through different fields of that. So that's like the third category, which
is maybe a little bit of both of like idea iteration and also automation where. It's basically streamlining
that process now. Yeah, I find that part like super,
um, hard to understand right now because as you said, in the design
process, we still have like a lot of, uh, real estate there to grasp,
like AI hasn't really taken over. But in terms of visualization. Uh, when we see things like D five,
maybe, I don't know, in six months, more Enscape and more, uh, render engines,
like Enscape or Lumion start to also use AI and with mid journey and how
like, uh, particular and specific it's getting, where do you think architectural
visualizers will be in the future? Like. That's a good, that's a good question. Yeah. Um, I actually had, uh, Jan Bunger from,
Jan Bung from Squint Opera on, on our podcast and he wrote a paper called the,
the death of the visualizer or the death of renderers, uh, a little while ago. And, um, I mean, yeah, if I
was a rendering firm, I would be sweating a little bit. I mean, yeah, it's, it's a great
question because first of all. We have real time rendering and that
already as an architect brought a bit more control back to us, you think,
you know, before that you'd have to do the whole V Ray thing and set it up. And now you can create
quite, quite good images. Yeah, maybe not like. You know, super, super production level,
but Enscape Twinmotion, you can create quite incredible images very quickly. And that meant that as an architect,
you get back in the hands to control of, of visualizing. Now with AI, I think, you know,
you're already getting render level quality stuff from, from Midjourney. The thing is, Midjourney is super Good
quality and you can get those results, but you can't necessarily get the control
you want the ones where you have control. You don't quite have that level of quality
yet, but that's only gonna come but Yeah, I think that it's honestly, I don't know
but if I was a viz person, I I think I'd be a little bit more Worried than I am
now, but of course again the value that The, the vis, the good visualization
people add, like me, for example, is the story and the thinking behind the image. Yeah. Um, so I actually think, like, for
example, a mer or, or squint opera or something like that, I see that we'll just
use these, these softwares and tools, uh, to their advantage to, to continue that. So, uh, I do think it comes down
to like that thinking, what is the thinking behind that image? Yeah, basic renders, they're
probably going to lose some jobs, but those, those people that are really
creating amazing things, maybe not. Yeah, those that have a high level of
control of the image and the narrative behind it, maybe the specifics of, of
a scene, either adding, telling a story behind it, because as of right now, and I
don't know if in a couple of years, that will change, but you cannot control that. I sometimes feel that all of
the AI generated images, they tend to have a similar style. You look at them and they all feel
within the same category, which. As of right now, it's not really a good
thing, but I think once it evolves, it will give us a more of a range
of possibilities in terms of result. That's super interesting because I've,
I've also noticed the same thing. I've, um, I think you need to get to it. Good level of, I call
it prompt whispering. Yeah. Some of the good people like the, you
know, Tim Foose, Will Gardner and all this kind of stuff, they have developed
a style, which is quite interesting. Um, but I have myself, like I've, I've
been tinkering for my own, like, you know, you sit there on a Sunday and
you're like, Oh, I've got an idea. And then you, you go down this
rabbit hole of mid journey, but then I've been online and I've seen. Someone's created an idea that,
that looks really similar to an idea that I've generated. And it's almost like you're
asking the same designer to come up with these things. And so you do start to see this, um, this
like re repetition of ideas and style. And I wonder if that's like
part of getting really good at. At the prompting and, and, and the
control of it, but yeah, there is that maybe the corner of it is like
you are seeing it's got its own style that is kind of repeating his
places on that line of prompting. Do you think like, because right
now it is a skill that you develop and there are people getting hired
based on their prompts, prompting skills, but don't you think AI will
really get good at prompting as well. So then it's kind of a, uh, Cycle
that happens where we're prompting gets better, the images get better. And then that prompting skill doesn't
really become useful or I mean, I agree with that or, or no, we already
see that with like, um, mid journey three, mid journey four, for example, I
think you really had to be a bit of a. a prompt whisperer, um, and then V5
came and talking to, uh, to all of these, these kinds of people, they were
like, it was almost like, uh, it wiped the slate clean because instead of
being these quite complicated, like a description sentence, and then all these
like, Additional, uh, tags, let's call them like cyberpunk or ultra realistic. Do you remember the fire at the beginning? You used to see like, uh, V Ray render,
uh, super high, hyper realistic and all this kind of stuff, but then V5 came and
it was much more plain English, like it was not so the prompting got way simpler. And the results were still good. So I do think we're already seeing
that actually, um, we also did like a fun experiment where we, we made
like a little brief, um, And it was actually for like a, a theme park. One of this, uh, one of the guys that
was helping with this big school made it. And we asked chat TPT to like
come up with the concept. It was almost like
giving it a fake project. And we gave it a brief to come
up with like a, a theme park. And we then said, give me
prompts to visualize that. And then we piped it into mid
journey and it created these like quite interesting images. And then it, we kind of
went down this rabbit hole. We were like, Oh, well, we'll tell us
what a theme park ride would be like in this, in this project and stuff like that. And I've also done the same with,
with kind of my own experiments where we're testing, like, what is the
future of housing going to be okay. Create an image of what this would look
like and then piping it into mid journey. So, um, it's, it's been quite interesting,
like this, like you say, all of those feedback loop of AI and that's
where it gets like super exciting. But also a little bit
scary at the same time. That's what I was going to say. I mean like Us three, we seem
like very optimistic about the future with AI because we see
like a lot of ways to use it. But what do you see like a recent
graduated architect who had the plans like a year before to go into an architectural
office, start like from the bottom drawing plans, then creating renders,
then going on to design processes. How will they like their jobs change? I imagine that now, if you were an
architect and you have a little bit of AI prompting experience, it will go
much better for you than if you just know how to use Revit really well. I'm actually more excited for them than
when I was in their position because they will not have to go through, well,
hopefully they won't have to go to through too much of the drawing plans
line by line in CAD or, or like, uh, You know, endless nights photoshopping
or sitting there waiting for 3ds max to render the Worst render you've ever
seen, but still takes like three hours. So, so already from my generation, like
even something that like Enscape, uh, has made the new generation's life so much
better because it's so much less painful. And again, it comes back to the point
of they can spend more time with the thinking and adding value than being a,
a internal junior architect where your. Endlessly modeling or photoshopping or
rendering, you can actually have time to, to input more into the design process. So, uh, I mean, it is a more, it
is an optimistic view, but I, I'm less concerned that, you know, AI is
coming to take out jobs and take over as architects, I actually think it
will allow us to focus in on what. We're actually good at and what we add the
value, which is the thinking process and less being like wizards at CAD or BMO or
whatever, you know, whatever it may be. So I'm more hopeful for them. But again, I think you have to,
you can't ignore this AI stuff. It's very important to learn
how to be an architect. You must learn the thinking process, but
I definitely wouldn't ignore or dismiss. This kind of AI movement at all. Yeah. Whenever we open question boxes or
talk about AI and ask feedback in the comments, that topic always comes up. Uh, with that, with the worst scenario
being the real scenario where architects will lose their jobs, and maybe we're
not talking about in two years, three years, but maybe in five, 10 years. So you don't see it. I don't see it that way. At least I hope not. I see only as, as tools that will help us. Have more time to dedicate to what
the architects are really capable and expected to be instead of wasting time
doing rendering is doing time consuming tasks and will free us up to have, have
more time to dedicate to those tasks. So that's how I see it. That's, I think how
Steven sees it as well. And you, so that's not the case here. I guess that we'll lose our jobs
in the office will be useless. That's always the topic. It's, it's, yeah, it's funny
how it always comes up. This will be the famous clip
where big guest doesn't no longer leads me because AI is like a
dominant in all this kind of stuff. But, uh, yeah, I mean, of course,
it's a more optimistic view of it is absolutely going to disrupt the industry. Do not make a mistake about it. I do feel we're in. A very unique moment, uh, where
our industry is going to change. Um, I did, uh, have post on our ATN
Instagram about how I, the era we're in. And I, I feel like we, and how
it's going to affect practice. So I, I, I do think we'll see
architecture practice change. And, um, for example, I think. I think we're moving into what I, what
I would call the AI collaborative era. So, which is architect 3. 0, this is stolen from the book life 3. 0, which is, uh, I highly recommend
to read about it's, it's much more about general AI affecting, uh, life. Um, but I kind of stole that analogy
to create, uh, uh, an architecture equivalent where architect 1. 0 was when we're kind of the manual era. That was back in the day, so almost
before my generation where it was like thousands, hundreds of people on a floor
plate drawing one building by hand. And, uh, then we had architect 2. 0, which is kind of the era we're
in the computer aided era where we use computers as somewhat psychics
or, or almost like, uh, computerized slaves to help us do things quicker. That could be, uh, AutoCAD grasshopper,
all of these things, like, like a grasshopper script is just, uh, you
giving a computer a ton of instructions and it does it very much more quicker
than we can do it, but it's not smart. It's just like, uh, it's
executing this thing for us. And so we were in the computer aided
era and then you saw an architecture practice go from like a floor plate
of Of mostly men at that time, drawing by hand to a much more diverse office
where you could take a picture of our office right now and big where everyone
is sitting in front of computers. Um, and let's say anywhere from 10. To five to 15, depending on the size
of project can handle a project. Whereas before that number, maybe 50,
50 people, I do think we will see the new era could be like much smaller,
nimbler practices, uh, where small teams collaborate with AI to tackle. These similar products, and hopefully
we'd see a much more diverse, uh, array of different practices. Um, so I think, yeah, practices might
get smaller, but we might have much more variation of, of practices. Um, I could definitely see
that, uh, in the future. Um. And actually, there's a,
there's a really good example. Um, I just did a podcast with, uh, uh,
the founder of Look X AI, who also has a company called, um, Um, she was called
when you, her, and I highly recommend to check out, um, the podcast she did
with, uh, digital futures, I think it's called, where her small team were using
AI to test the tools that they make that they're kind of like a technology
company, but they started to enter competitions using AI and with like a
three person team, they actually won. The competition and I felt that
was, that's really an indication of what I think the future will have. So it will disrupt us, but I
think, uh, it's not necessarily like AI is taking over jobs. We might see a more interesting array
of practices and things like that. I think, I think I have the book here. Which is in Portuguese,
but yeah, that's it. Max Tegmark's Life 3. 0.
It's really interesting. And I love the analogy between
what he describes at Life 3. 0 in relation to architecture. That was really interesting. Yeah. And in a sort of way you see, um, like
other industries like photography with. Like so many advanced tools within
photography, many are going back into the analog and going back to printing photos
and trying to maybe even taking photos with their phone and trying to print
them out and trying to have a more, um, manual workflow now that they know that
they can spend some more time with that. How do you think that's going
to also go into an architecture kind of phase where we are more
interested in creating maybe models? Maybe hand drawings that are not
necessarily, you know, so time consuming, but more for our design process. Yeah, I think that's getting back to
the core thinking of the architect. So like, um, yeah, if you don't need
to spend the afternoon, um, You know, photoshopping this, this image, it
means you can spend that extra half day sketching the actual idea, like
before, before you get to that point. Right. So I think, yeah, you, you, you could
see us get back to, to sketching. And, and I think you, you're also seeing
architects now getting into like sketching to AI, they, they sketch and then can very
quickly see somewhat rendered version of their sketch, which again is, is that. idea iteration part where you can, you
can go sketch the render now really quick. So yes, I do think you'll see more
people like going back to, to, to drawing that kind of stuff. The, the thinking part, I do also
think we will see a rise in like. Artisanal architects, like, like, do
you see how we're now like, see people coming back and everyone's going back
to like the old way of doing things, like making coffee artisanal, and then
you get a little tattoos of coffee. We were joking on a, on a podcast. I'm sure we'll see like
artisanal architects. You'll get a little T square tattoo on
your arm and, and you're not allowed to use a computer and stuff like that. It's possibly because
there's always the world. When we as a society go so much into
one side, uh, and it feels like we're over pushing this way of AI computers,
everything so digitally made, there's always that, that pushback, that,
that slingshot back to the more manual method, more to manual creation. That's something that also came up with
the Instagram questions that, that I posted a lot of people saying, asking. So if AI is becoming so relevant. Does that mean that the physical ways
of creation are also coming back as a way of pushing back on this, this
new technology and that's, I think that's, that's the answer, right? Yeah, it could do. Yeah, I think it's always
double edged, isn't it? Because like, um, on one way you can
interpret that, like, uh, for example, in the office at big, it was, uh,
obviously big was, uh, It came from, it was called a, is a baby REM, right? Bianca worked at OMA. OMA was a very famous for those
blue physical models, right? The fur models. And I think that was a completely
part of the design process of big at the beginning, right? It was these blue fur models and actually. It was about the era when I joined
in 2018, suddenly these 3D printers all merged in the workshop and we
moved from these handmade physical models to 3D printed physical models. But it didn't really take away
from the reason why we're creating a physical model is to see it. Feel it and, and put it into a context. You can never beat seeing, uh, even
a physical model just in front of you, even with VR and all this kind
of stuff, you can't beat seeing it and feeling it and, and touching it. That's the valuable thing. Whether that's someone making that by
hand, uh, really painful or it's 3d printed, I could argue that actually
that that's not where the value of having that physical model is. It's, it's about having it
in your hands at the end. So, uh, yeah, again, I think
it's, it's a bit of both like that process will be digital, but, but
also physical at the same time. Yeah. It's crazy that when you're talking about
VR and then we, um, Met up a few years ago, you were talking about how the VR was
also getting into the design process where you're trying to see like the type of
hologram of your building of your model. Where is VR now with, you know,
with the recent launch of, you know, what Meta launched and where it's
going, where do you think in terms of architecture and design, where
do you think it's being implemented? So the VR and AR side, I think,
um, I think is limited by the technology that we have right now. Um, still, I think people struggle
with the headset thing, like. Yeah, it's it's always a little bit
awkward when you're in a meeting and you one person has a headset on they
feel like a bit embarrassed because they don't know what's going on. But it does have value that you can you
can feel like what it might be like to be in that space and things like that. I do think we're in the era like
these big old goggles that we have is kind of like the era when. You know, maybe our, our parents
had those massive phones. Like we're in that era of,
uh, the giant phone era. Um, and actually I was a little bit
disappointed when Apple came out. I thought theirs was going to be slicker,
like a more meaningful, like a flask. Like glasses or something, but of course
that will come and, and that's when that technology will, will really take off. I mean, VR, of course, it's
still part of the process. It's, it's still being used. I wouldn't say it's a, a core part of
it, but it's absolutely, you know, we use it at the right time and when products
need it the same with augmented reality. It's again, like when there's
this particular use cases for it. AR side is much more about being
on site and stuff like that. Um, and of course we
have products on site. We've been using it on our headquarters
because of course it's a unique case where we are the designers, but we're
also the client and we can go on site and overlay a big model and stuff like that. So there's some really interesting
stuff going on in that space. Um, but I think the most interesting
thing in that space and whilst everyone is like FOMOing. Over AI right now is what's
happening in unreal right now, I think is, is super exciting. Um, and I, I, and I wonder maybe
that, that unreal side will unlock. The next era of VR and AR and all
that kind of stuff, um, because there's some incredible stuff going
on and what's happening like for the people that don't have like any idea
what's happening in real or so first of all, like unreal five unleashed
this, this ability to create worlds. You know, and, and handle huge
models that are incredibly detailed. Um, and we're also seeing lots of plugins
and features that are coming with unreal. Now you can, I think you can
download the whole of Google maps and put it into unreal. And so. Again, now you, you in theory, no longer
need to model your context because you could, you could download, um, like a high
resolution, uh, model from Google Earth. Um, I think that there's,
there's things like that. There's also. NVIDIA's Omniverse, which is super
exciting, where it's kind of like a, it's, it's kind of like a game engine
is a really simple way of explaining it, but you see, um, people bringing things
from Unreal or Rhino or whatever it is into this place called the Omniverse and
then running insane simulations using the computational power that NVIDIA have. But yeah, there's, there's loads of stuff
going on with, with, uh, Unreal, like the. The quality and the real time
seems to be getting extremely good. Of course you got like all the
mega, the Quixel mega scans and all this kind of stuff. It's um, it seems to be a
really, really exciting space. And I, I think everyone's looking
at this AI stuff, but actually this Unreal stuff is, is as exciting. Um, and eventually
those two worlds will... We'll combine that's where
the simulation will come. That's where we realized that we're all
just a game, like a really high quality. Yeah. Yeah. That's where the life 3. 0 book really starts to kind of make
you think about the possibilities. Wow. We're just speculating, right? Because AI evolved so rapidly every three
to four months, it doubles in process powering in I mean, it's capabilities. So we don't, we don't know, like in five
years, it's, it's very hard to predict. Uh, yeah, I, you know, this
is my, these are my thoughts. Like no one can predict
really what's going to happen. I mean, maybe, maybe these people that are
making these things, um, that being said, I, I did, I've had a few conversations
with people that are building these, some of these tools and, you know,
we're in this image generation era. Of course, 3D is, is
coming around the corner. Um, you type in things and you
will not just get an image. You might get a 3D thing, but still
that 3D model you get, I think would, would be very basic to get to the level
of detail that we need as architects. You're not talking about
like a rough mesh model. You're talking like a very accurate. Accurate model if you're talking
about building or even a BIM model. And so I do think we're a long way from
like, uh, you know, text to real BIM. There are already some examples
with Hyper has done some, some kind of text to BIM things. It's very. rudimentary, but it's, but it's a
light into what the future could hold. But I do think we have a little bit
of a way to go before you're like typing in and a BIM model appears. Like on a, on a lighter note,
do you think, how do you use AI on your day to day? Because for example, I, I feel like I have
an assistant right here with me all day. And I ask her questions, like
sometimes I, like I say, thank you. Uh, you know, I, I like in my writing
emails and trying to solve a big task, like, Hey, like, how do I automate,
I don't know, these 10 emails. And they give me answers. I find it very, very useful and it's
not necessarily architecture related. It's more like business management. How do you personally like use AI? It's probably the same. Like, uh, we all, we all now
have this chat GPT sidekick, uh, that we're all using. And even if it's something as
simple as rewording things, um, I'm not the best writer. So it's, it's super interesting to
use for me, um, to reformat things. And like you say, like speed up the
mundane, the emails and stuff like that. Um, I think, uh, obviously
for me, I'm usually like, um, Trying to help in the office. I'm trying to help the designers use it. So I may, I may be like just teaching
them some of the add ons, like, Oh, how do you change the aspect ratio of an image
or like these little tips and tricks? Like they have an image they want
to use and work on top of it. So I'm like, okay, well, if you set this
setting really high, it's just going to reinvent your image if you set it low. Um, so I'm, I'm spending more time
kind of Giving people tips and tricks and, and, or getting them up to speed
with the basics of, of mid journey. And then I say on a day to basis,
I come home and then I tinker, okay, I have an, I have an idea. I think I most enjoy using it personally. Like, like many architects, you
get ideas coming into your head and you're like, man, I wish I had like
a, a month to model that in Rhino and script it and create some images and. Enscape or yeah, or unreal or
whatever you're using and, but now I can sit down and be like,
Oh, Hey, what, what if you made a skyscraper of containers in Hong Kong? What would that look like? And, and, and you're like, I can get
quite close to idea really quickly. Um, but it's that last bit that you
kind of, That's when you would need to come in and like, I want to refine
it and get to the detail of it. Um, so yeah, I think those are
the three ways I've been using it. It's like one production, uh, it's
almost like a production sidekick. Two is, is helping and teaching to
get it into the hands of designers and three personally, I've just
been enjoying playing with it and, and like visualizing ideas really. So funny. Like saying thank you to
chat, just in case, right. Just in case they in the
future, like maybe they can. They come back. He always said, thank you. Oliver was nice. So let's not do it to him. I always say, thank you. Yeah. How about, how about you guys? How has AI affected what you guys do as,
you know, content creators essentially? Right. Like, um, has it affected the
way that, has it reduced your. Um, editing time or, or have you
got, I've seen these AI apps for like creating reels and stuff like that. I've never tried it, but have you
got any good, uh, got any good advice for other creators out there? Like, has it affected
what you guys do yet? I think the tool that I mostly
use is, is the, the Photoshop one generative feel with, uh, like making
thumbnails, uh, or fixing something from an image that will take it. At least 30 to an hour, 30
minutes to an hour to do it. I tried to be to your force. And I think Steven is much,
much stronger on that sense. How about you? Yeah, well, I've tried to use, um, a
lot of tools that, you know, the first, like the basic ones are the ones that
generate like automatic captions for you. And, uh, for your reels and
like reframe the reels, I think in, in, uh, in Premiere. I don't know if that's AI or not, but
if it just, it just reframes your, your. Like old video ads, captions, and
makes it much more interesting. So I've been using that. I've also been using, um, chat to generate
like 20 title ideas, or maybe when I'm trying to tell a story through a video,
sometimes I give it like my basic view of how I'm trying to tell the story. And then I'm like, Hey,
this is what I have. Could you maybe restructure it to
make it sound more interesting? Or are there any other possible ways,
possible ways of telling this same story? And then. What else? I think I'm maybe, yeah, just
answering emails and creating more like administrative tasks. ChatGPT is helping a lot. You've used that video image
to video AI tool as well. I remember you posted something. Oh, Runway. Yeah. Runway, yeah. Talked about that. I used it recently, and
I thought it was crazy. I mean, I used the free version, so
like the results were very low quality, like 70, 720 pixels, but I know that
you can pay not a lot of money, and you can have with your still render Like,
uh, I think there's like a, and if you, if you go into Twitter, there are many
people who have created entire movies out of runway mid journey in chat GPT. And they're really high quality movies. And they're, they're amazing,
which, you know, if you, if you wanted, you could use it for the
whole architecture animation reel. You just export five different renders
and then runway can animate it. And that's it. I do think that's something
interesting about, uh, and Mariana brought this up in a. And the podcast I did with her,
where she was talking about, uh, algorithms feeding algorithms. And of course, one of the things
we've seen with this AI image generation stuff is this explosion of
these, uh, Instagram AI characters. Um, and of course there is something
really interesting in like, you need to post, let's say you need to post
every day on Instagram to, to continue to grow and like to create a project. Every day is hard work, even if you, even
if you create a product and you get 10 images out of it or something, it's a lot,
it's, it's a lot of content to create. Now you can create a hundred ideas in one
day and you can post every hour if you have to like, and now we're seeing, you
know, this, this algorithm that generates images is feeding this other algorithm,
which is the Instagram algorithm or the YouTube algorithm or whatever's. I guess it's less, less YouTube,
but like, there's something quite interesting in that. And I do wonder how that will
affect the content creation world. I think it goes back to that idea that
I was saying, like, when you go too much into the digital way of thinking that
there are some people that counter back on, on the manual way of doing stuff. So I think that you see more and more
of this, um, AI generated content, a hundred, a hundred ideas in a day. Yeah. So, but I, so I think the human part
of, of all of this process is, is going to prevail in some sense, and it's
going to get more importance at the end of the day, because in a couple of
years, we'll, we'll see more and more we interacting with AI generated content. So where's the, the
real people behind this? And I think those are going to gain
more traction than you would ever. Because of that saturation of AI
algorithm, I think it comes back down to again, the thinking like you guys,
it's about the ideas that you have to create the videos and the content, the
knowledge that you're giving to your viewers, whereas like, uh, and, and
I think for the image side, to me, it gets interesting when I understand a bit
more the thinking behind the image, uh, rather than it's just people posting. Yeah. Endless, endless images. Uh, and I think that those also seem to
be the ones that grow the most as well. So yeah, hopefully it's the
human brain that will prevail. You would be really interesting to have
this, this conversation again, like in a, in a year or two, obviously we're
going to do more stuff before then, but imagine like seeing these predictions
in, in, in a two year time and see like how, how did that evolve and. Yeah, I'll be right about it or
maybe I'll be, uh, I'll be homeless or something cause I've got no job. Yeah, you know. Uh, but yeah, one way or another,
I do think we're in a, we're in a very interesting time, a very
unique time in architecture. Um. And I think there's a lot of opportunity
for architects to maybe create something new, or we might see new types of
architecture where we see the, the, the biological brain and, uh, the
artificial brain or the synthetic brain come together and create things that
we may not, not have thought of before. Um, so yeah, I think it's scary. I keep saying this, this
period, it seems very scary. But also very exciting at the same time. Yeah, it's true. Um, and I think the most important thing
is for people, like people maybe listening to this and maybe Confused or not sure
where to go is, is to, to not ignore it. Like, uh, try it out. Um, I think I was talking to
Tim Fu had some great advice. Like this is just the beginning. It's it, you're not too late to jump in
and get good at mid journey or, or chat. UBT. These are the tools that we have now, who
knows what we'll have in the future, but. It will of course help you to
keep up to date with these things. So I think that's the key thing is,
is experiment, have fun with these tools, whether you like them or not. Uh, at the moment they are tools,
maybe they'll cross the line from a tool to a collaborator very soon. You could argue, maybe they
are collaborators right now. Um, but I think the most important thing
is have fun with it and play with these and test them out and don't ignore it.
