S10 EP3: Stanton Kroenert, Innovation and Technology

Stanton is not only a Health Architect but an innovator and early adoptor of technology and AI. 

Has the world of technology finally caught up with Stanton's Architectual technology ideas? Let's find out.

If you would like to learn more about the AHDC, please connect with us on our website www.aushdc.org.au or on LinkedIn at linkedin.com/company/aushdc.

Transcript

[00:00:19] David Cummins: G'day and welcome to the AHDC podcast series, Health Design on the Go. 

[00:00:24] I'm your host, David Cummins and today we are speaking to Stanton Kroenert, a health architect with over 28 years of health design experience. Stanton has designed multiple hospitals across Australia and around the world with meticulous detail with a strong focus on patient care.

[00:00:39] We welcome Stanton today to discuss AI as part of our innovation and technology series to discuss and discover what impact AI will have on the world of architecture. 

[00:00:49] Welcome Stanton. Thank you for your time to be here. 

[00:00:52] Stanton Kroenert: Hi, good to be here again. 

[00:00:53] David Cummins: I must admit, I use AI a fair bit, just in normal report writing and with project management and for daily use. But I didn't even know AI had entered the world of architecture.

[00:01:03] Do you mind sort of explaining what you mean by the world of AI and architecture? 

[00:01:07] Stanton Kroenert: I can probably start by saying I had a go at this about 25 years ago... in that I wrote a computer program that would actually plan out a hospital. And no surprise, it didn't work fantastically, but it's one of those things where I've always sort of had in the back of my mind that eventually computers will do a lot of the stuff that we do as hospital architects.

[00:01:25] But in the last six months, it's definitely changing the way we're doing stuff. 

[00:01:30] David Cummins: I had no idea. Do you mind us digging a little bit deeper? Are you talking about, you just type in "design a cardiac ward" and it designs it for you? 

[00:01:39] Stanton Kroenert: No. So what it's limited to at the moment is really visual stuff.

[00:01:43] So it's really good at coming up with concept design, look and feel of things. So we've, we've got an AI that we've been training for about six months and we can put in basic models and it pumps out different design ideas very quickly, which are really useful because we can take 10 or 15 images very quickly to a client and then discuss what sort of way that they feel about particular images, which then gives us hints on which way we should go in the design process, but it's a very good tool to come up with quick ideas.

[00:02:20] Now, the pure architects out there will probably want to throw stones at me for this. However, it is a very useful tool and that's how we've been using it so far. 

[00:02:30] But it's getting to the point where it's doing some other stuff that in the near term future, we'll be using it a lot more for. 

[00:02:37] A lot of tasks that we do as hospital architects is repetitive and I think a lot of that can be automated and we're starting to do that ourselves at the moment. 

[00:02:46] David Cummins: Yeah, I sort of feel like. Again, you said the pure architects... I remember when it went from drawing, like hand drawing, which I studied at uni to computer and BIM over 20 years ago there was a big change and a bit of resistance, but I feel like also there is certainly architects out there who are very much the early adopters... and they're the ones that, you know, continue to work very, very well in the future. 

[00:03:08] Stanton Kroenert: So, there's a quote that I sort of use with this is AI won't replace architects, architects who use AI will replace architects. 

[00:03:15] And now I'm one of those old school architects. I don't even know how to open a computer drawing. I still draw on yellow trace, which a lot of young architects out there won't even know what yellow trace is anymore. 

[00:03:26] I have no idea how computer models work and all of these kinds of things. However, what I do know is there's no way of avoiding it because the stuff that we're already producing just blows my mind.

[00:03:38] And in fact, our very first foray into it was actually done by a work experience student, a 16 year old (And shout out to Christina Koutsioufitis, who is Harry Koutsioufitis' daughter), one of the GMs at HealthScope. 

[00:03:51] She came and did work experience with us, and we gave her the task for a week to see what she could produce using different AI packages and when I saw what she produced in a week, I was sold. 

[00:04:03] And so since then we've had a lot of time and effort dedicated to it. And we're getting it to do things now that, six months ago, I wouldn't have thought were possible. 

[00:04:12] David Cummins: So just talk me through it. So again, you input and more than happy for you to share the technology name, the technology name, but you input something, a description,

[00:04:19] build a...

[00:04:21] Stanton Kroenert: So what we do is we get a computer model... so we use a Revit model. We import it into the program ( as I said, we've been training one, so it's been adapted to how we use it) and then what we do is give it a list of parameters of what materials potentially, we give a particular views we want to see it from, we also can input backgrounds so that we can put it into a site context and that will pump out designs. 

[00:04:46] Now we can twist the parameters to actually say, okay, we want it more glass, less concrete, more concrete, less glass, more wood, less lighting, more lighting... all these different parameters that we've built up over time and then it will produce different variations on that theme.

[00:05:02] David Cummins: Does it get you to what, an SD drawing or a concept drawing or a FISO drawing? 

[00:05:06] Stanton Kroenert: It's just a concept. 

[00:05:08] It's enough to take an image to a client and say, you know, which of these 15, 20 images do you gel most with now? Will that be the final design? No, but if anything, it will quickly eliminate ideas they don't like... which is half the battle. 

[00:05:25] So it's just a really good tool for getting very early ideas. And then we still go through the normal design process after we've got an initial idea. But once that initial idea is set, that's half the battle one. 

[00:05:37] David Cummins: Would I be correct in saying, in the world of clinical health planning with the functional briefs, the business cases, the models of care ... I'm just making it up right now, but is there an opportunity to literally write the functional briefs and the models of care which could input into the architectural feasibility from a drawing? 

[00:05:53] Surely that's not too far away. 

[00:05:55] Stanton Kroenert: Which is exactly what the program I wrote 25 years ago tried to do. So yes, that is definitely coming. 

[00:06:02] The problem is there's so many variables when it comes to planning. And remember we're also planning in 3D. Because we're usually doing multistory building, so it's not as simple as doing an adjacencies matrix drawing, having an area of a department and then letting it work it out itself, it just doesn't work that way, there's just too many variables. 

[00:06:23] But I think as AI is getting better at plain language, it's going to come for sure. Because once it completely understands just plain language inputs, then yes, it will be able to start doing that planning stuff. 

[00:06:35] But at the moment, I'm probably still quicker than what a computer can do in terms of planning out an entire hospital from scratch.

[00:06:43] David Cummins: A fair bit of that would be the input from the users and the consumers. So I wonder if there's even opportunity for AI to collaboratively partake in a user group where it's literally, you know, listening, transcribing and designing on the spotlight.

[00:06:58] Surely that's not too far away either. 

[00:07:00] Stanton Kroenert: I still think that's a fair way away because as I said, it still doesn't understand complex concepts, which a lot of health stuff is. 

[00:07:07] So if you're designing an operating theater, trying to explain how an operating theater works to an AI, yes, once you do it once it should get it, but there's so many inputs that it's not there yet.

[00:07:18] But give it five, 10 years, it will be there for sure. 

[00:07:21] David Cummins: Is there opportunity, because obviously we do a lot of interviews around the world and Australia, and I predominantly interview people who have done their PhDs and research in the world of architecture and health design. I understand this is best practice from a practical world, 

[00:07:35] but surely there's a way of the latest research for an ICU, for example, to make sure that it's incorporated, or at least considered when it comes to these drawings?

[00:07:45] Stanton Kroenert: It is. However, I still think there's actually one big step that's actually even more important. 

[00:07:52] So I think once AI can simulate a patient, I think that will actually be way more powerful because what it will allow you to do, is to do research on hospital design decisions and then see how every decision affects a patient.

[00:08:07] Now, one of the things I'm working on at the moment is a research project looking at the idea of not having windows to patient bedrooms and instead replacing them with LED screens because LED screens now have got to the point where they're the same resolution as what we see. We see in about 8K and that's what TVs are now producing.

[00:08:26] So the premise is will a fake window be the same as a normal window with the advantage of perhaps having a better view than what a real window might have. Now, there's a whole bunch of ethics concerns of actually doing that to patient bedrooms. 

[00:08:41] However, if you could simulate a patient in AI and actually then see what the impact on a virtual patient is by all of these decisions, I think we'll get to a point pretty quickly where we know with total certainty, the impact on all of these decisions on patients and we won't have to test something over a year to see what the impact on, say, average length of stay is, we'll know instantaneously because the computer will tell us. 

[00:09:05] So I think that's actually going to be the quantum leap that I think is going to come. Now, that might be five, 10 years away, but once that happens, that changes the whole game. 

[00:09:16] David Cummins: My knowledge of AI is a little bit limited in the sense that I think predominantly it comes out of America, but are there practices in the world doing what you're hoping AI can do now or everyone's pretty much level playing field 

[00:09:32] Stanton Kroenert: No, the only examples I've seen is people still playing around with images, like we are, of early concept designs and there's been a bit of effort to planning.

[00:09:41] But again, with hospital planning, it's just too complex at the moment to still do that. Eventually, we'll get there as I said, once it understands plain language that potentially are in user groups, but you still have to tease out the information and that's not what I was good at.

[00:09:56] It doesn't know what questions to ask whereas that's why I think I'll still have a job for the rest of my career because I know what questions to ask. 

[00:10:03] Now, will the AI be able to interpret the answers to the questions and provide a better answer than I could? Yeah, potentially. But I think the teasing the information out and then deciding what is more important information than other information is still a trick that it won't be able to do anytime soon.

[00:10:20] David Cummins: So you talk about AI, your job being safe for your retirement, which is realistically knowing you at least 30 or 40 years into your retirement. But after that 30 years, how safe are architecture jobs? How safe are those roles in the world of AI? 

[00:10:38] Stanton Kroenert: The way I look at it is there will still be a place for architects.

[00:10:42] And I'm just talking architecture in general, not necessarily hospital architecture in the way that you still like handmade pottery or people like handmade shoes, there's still some sort of status applied to handmade things. 

[00:10:57] And I think architecture will eventually become that in that you'll still have utilitarian kind of design, which can all be done by.

[00:11:06] AI and you know that probably 80 percent of it or 90 percent of of things could just be designed by AI in the future, but you'll have that still small segment of society that will still value this product, which is almost like handmade pottery isn't as good as, you know, Bob-made pottery, but there is something about it that makes it more valuable.

[00:11:27] And I think that's the way architecture will become as well. There'll be a status applied to something that is designed by a human. 

[00:11:36] David Cummins: Yeah, that makes sense.

[00:11:38] I'm just thinking out loud, I know one of the programs is Da Vinci, or you can do some form of defake for a person. But is theree not also an opportunity for, at one point in time, a CEO, or an executive of a hospital, to say.. " I would like a tall green building with heaps of windows and open spaces for community groups and connection to country where literally they're just talking to a computer and it provides multiple facets".

[00:12:03] From then you can say, let me explore one more aspect of this and therefore a feasibility business case can be provided. Is that correct? 

[00:12:12] Stanton Kroenert: Sort of.... 

[00:12:13] so I think what eventually will happen is you'll be able to get a health service plan and put it straight into an AI, and will it pump out a hospital design? Yes. 

[00:12:22] Could you have a CEO of a hospital say, I like something that looks like this? 

[00:12:26] Yeah. But still, having spent all these years training as an architect, there still is some aesthetic value in having an architect talk to an AI in a way that will produce outcomes that are better than what a non-trained person in what looks good can still produce. 

[00:12:44] And I still think there is a place for an architect working with AI, is always going to be better than an architect working without AI and vice versa. 

[00:12:54] And AI working without an architect is not going to be as good as an AI working with an architect. 

[00:12:59] David Cummins: Yeah, you're a hundred percent right.

[00:13:00] I was just thinking of some of the human errors that happen in documentation. When you get a builder and that doesn't fit, but I wonder if it would even have the opportunity to reduce some of those errors?

[00:13:10] Stanton Kroenert: 100 percent. 

[00:13:12] And I think also there's going to be another stream where it's going to take it is also I was going to come up with new materials for sure. 

[00:13:19] And so the way that we build buildings will also be incorporated with these new technologies. I actually comes up with, like, I'm thinking, you know, like car design with all the carbon fiber and everything, why can't we build a hospital out of carbon fiber? 

[00:13:33] It resolves structural issues because it's seamless, it can resolve a whole bunch of cleaning issues, you don't have joins that you have to deal with, which is always one of the challenges in hospitals at the moment. 

[00:13:44] And so using 3D printing tech that AI can generate better ways of building stuff. I think that's going to be another stream of activity as well, apart from the design side.

[00:13:56] I think the new materials, new building techs that's going to change the game as well. 

[00:14:01] David Cummins: I can only assume that this is similar to the world of services like mechanical and water. It's similar to the world of landscaping, design or interior design. Or do you think that there is a point where this would just be limited to architecture?

[00:14:13] Stanton Kroenert: No, it'll, take over everything because there's got to be more efficient ways to deliver services in a building that AI will be very good at calculating. And I think to be quite honest, I think services engineers are probably going to go before architects because again, a lot of what they do is can be automated.

[00:14:30] Once you have a architectural design, which is usually the order of things, we do an architectural design that we then give to the services engineers to make it work. 

[00:14:39] And yes, there's always tweaking. But again, if an AI can actually do the consultant engineer's role in terms of I give them a design and it can work out how to do all the hydraulics and the mechanical and the electrical... again, I could see those services going before architectural services go, because the first step is doing an architectural design.

[00:15:00] David Cummins: It sounds like the only people we'll need at the end of the day is the architects and the builders, just like in the olden days. 

[00:15:06] Stanton Kroenert: No, probably the only people we'll need is the patient. 

[00:15:10] David Cummins: The patient, you know, what about the builders? AI is like, taking them over. 

[00:15:14] Stanton Kroenert: Eh, again, 3D printers will eventually be able to do everything.

[00:15:18] David Cummins: I must admit, I've kept my eye on the world of design and evidence based design, but I haven't actually seen too much evidence based design in this space at the moment. Are there many people doing research on it? 

[00:15:30] Stanton Kroenert: No, that's what I was saying. 

[00:15:32] Once the AI can generate a patient in the virtual world, that then solves all of the evidence based design because then all you have to do is add all of the different things you want to do to the environment around that patient and then it will just be able to calculate what the impact on the patient is.

[00:15:51] And the other thing is, you'll be able to manipulate that AI version of the patient depending on what is wrong with the patient. So you can tailor it to, if you're in a surgical ward, you'll be able to work out, okay, what's the effect of, okay, something dumb like having multi-bed rooms for nosocomial infections in a surgical patient.

[00:16:12] Once you have a surgical patient modeled, then you can pretty clearly work out what the effect of having a multi bedroom would be. Even something like the effect on sleep, having a multi bedroom, how would that affect, you know, each patient's sleep? Well, they can work out the decibel level from one bed to the other if someone's snoring.

[00:16:30] There's a whole bunch of stuff that once they model a patient, I think that's going to solve all of the evidence based design questions we've got. 

[00:16:37] David Cummins: So what should people listening today do today? Should they be embracing it, trying to, like you've said, write their own program, inputting data into a program, should they be avoiding it? Should they be trying to put their head in the sand? 

[00:16:50] What should architect students be doing today in this world. 

[00:16:55] Stanton Kroenert: From a selfish point of view, do nothing because then I'll make sure that I've got a job for the rest of my career. But if you want to compete, you're going to have to embrace AI in some way. 

[00:17:04] And I think the place we started was a good place in terms of just using a visualisation package. It's a pretty soft approach. If you don't adapt with AI as it's adapting, you'll just be a dinosaur.

[00:17:17] And even though I'm a dinosaur, cause I still draw by hand, I'm a dinosaur that sees that the comet's coming. And if I don't find somewhere to hide, then I'm going to go extinct as well. 

[00:17:26] So, that's why we've started going down this path as well. 

[00:17:30] David Cummins: Every time I speak to you, it's always mind boggling.

[00:17:32] I thank you so much for your dedication to not only architecture, but into this innovation world as well. 

[00:17:38] At the end of the day, I think you're a hundred percent right but there'll always be a need for some form of creativity in our designs in reference to pure logic, whether it's an accountant or a mathematician, maybe, maybe not.

[00:17:50] But definitely, I think the world of architecture and health architecture... it's pretty much the balance between logic and emotion where you do have to have both your creative side and your logical side working at once.

[00:18:00] So I'm sure AI might be able to get there one day but certainly today and certainly I can't see it happening in the next 20 plus years but I'm very very glad that we've got people like you to to be at the forefront of health design and I really am interested to retouch this conversation in about 20 years time to see where it's landed.

[00:18:17] So thank you so much for your time and dedication to the industry. 

[00:18:19] Stanton Kroenert: No problem. 

[00:18:20] David Cummins: You have been listening to the Australian Health Design Council podcast series, health Design on the go. 

[00:18:24] To learn more about the AHDC, please connect with us on LinkedIn or our website.

[00:18:28] Thank you for coming.