Current Facility Data for Future Design

Since the late 1960s, computer aided design has been a powerful tool in architecture engineering and construction, helping deliver faster design creation, modification, and optimization. But it’s always only been a way to make more precise designs and interact with them virtually. New developments in artificial intelligence and machine learning, though, put software into a more central role. 

Now, with predictive modeling you can use current operational and maintenance data to make decisions about future capital investments and facility design. Instead of using CAD to bring your ideas to virtual life, you feed the software data and ask it to both generate and test designs. You’re pulling the software a full step up in the process. 

All of this is possible because we’re finally close to those classic sci-fi thinking computers. 

What are the differences between artificial intelligence (AI), machine learning (ML), and predictive analytics (PA) for facility design? 

Before looking specifically at how these new technologies can be leveraged in facility design and construction, it’s important to understand what they are and a bit about how they work.  

The basic definition of AI is it’s the intelligence demonstrated by machines. The goal of AI is to reproduce our capacity for problem-solving and decision-making. Because they are good at one thing, self-driving cars and chess computers are examples of artificial narrow intelligence (ANI). Artificial general intelligence (AGI) has the same capabilities as a human while superintelligence (ASI) has more.    

ML is a form of artificial intelligence. Here, you feed large amounts of data into the software so it can learn by drawing inferences from the patterns it finds. Applications have become popular in the medical field, where ML detects early signs of cancer in medical scans.  

PA incorporates AI, ML, data mining, predictive modeling, and statistics to make predictions. Applications include both business and weather forecasting, voice-to-text translation, and financial investing.  

You can also use PA to help upgrade existing facilities and design new ones. 

How can you use predictive analytics and predictive modeling in the architectural design process? 

You can leverage PA for renovations, upgrades, and new construction projects. At the start, you need data, and what you collect and how much of it can depend on the type of facility. So, when using PA to design a medical complex, for example, you might include all your available data for: 

• Physical space 
• Materials management 
• Transport systems 
• Medical equipment 
• IT and clinical communications
• Staffing scheduling
• Patient scheduling
• Arrival policies 

You would also want data on patient encounters, including data on every unique patient’s journey through their department, and if possible, data on the departments the patient moves through both before and after. 

You then feed the data into the software so it can create models to test various designs and solutions. Because everything is done virtually and quickly, you’re saving money on both construction and staff.   

What are the benefits of AI-driven facility design? 

Imagine you wanted to find the fastest route home from the office. One way is to try every possible (but also reasonable) combination of streets and speeds. Over time, you eventually find the sweet spot. Another way, and here you’re saving a ton on gas, you sit down with a good street map and plot out all the reasonable routes, adding any ideas you might have on the influence of traffic. Some routes might be shorter but take more time if they’re also popular with other drivers.  

So, either you’re doing a lot of driving or a lot of manual math. 

If you used predictive analytics, though, you’d get a much more accurate answer, and you’d get it a lot faster. You feed into the software all the maps and historical traffic data. Every time you drive home from work, you add that data to the set. The software looks for patterns that it can then use to find the best possible route. 

And because the AI can work through a lot of data quickly, you’re free to feed it more. For example, data about population densities along different routes. The number of houses and schools can reveal traffic patterns into the future, when those pre-teens turn 16 and get their licenses. And the average weather conditions throughout the year might reveal the probability of traffic accidents. The more the software knows, the better the models it can build for you. 

The famous quotation “In God we trust, all others must bring data” fits perfectly with this process. Because everything is driven by data, there’s no room for emotion or prejudice in the final decision. You don’t have to take the route you’ve always taken back home just because it’s the one you’ve always used. 

What are the potential problems with AI designing facilities? 

There’s another famous quote that also fits perfectly, “Garbage in, garbage out.”  

Remember, the process starts with collecting data directly related to the key performance indicators (KPIs) you want the design to optimize. If you want to renovate a wing of a facility to maximize space management, you need lots of good, clean data on how you’re managing that space now. If you don’t have systems in place to get it, you need to set them up. Otherwise, the quality of the predictive analysis for your future design is as poor as your current datasets. 

There are also known problems with AI that can lead to inequity of outcomes, including statistical discrimination against ethnic and racial groups. Basically, when it’s trained using large enough datasets, the AI becomes racist, and there are already fears the AI at some hospitals is worsening medical racism. This is only tangentially related to the process of using AI and ML to design future facilities, but it is important to remember that AI and ML are not perfect solutions. They come with their own set of problems that need to be hammered out, just like any new technology.  

How can AI in the operational phase also affect a facility’s design? 

You can use predictive modeling to help you design a new facility. But if you knew you could also use it for maintenance and operations, that would affect the design, too. 

Airports are a great example. So much of the current design focuses on giving passengers things to do while they wait for flights. And with all the inefficiencies built into every step of the process, including parking, check-in, security, and boarding, many airports pride themselves not on being efficient but on being a nice place to wait. There are even online rankings for the best airports for long layovers.  

But an AI working in real time could conceivably control seamless air travel, with perfectly timed home pickups using self-driving cars and individually curated boarding. Some now predict airport design in the future is a lot more sparce. Once waiting disappears, so would all the airport lounges and restaurants, souvenir shops, newsstands, and coffeeshops. 

The departure gates would suddenly look a lot like arrivals, with long, mostly empty utilitarian hallways. The goal is to get you out of the building in as straight a line as possible.    

Summary 

Although CAD has long held an important position in the facility design process, it has always been a way for designers to bring their ideas to life. Now, with new developments in artificial intelligence, machine learnings, and predictive analytics, you can feed software large data sets and ask it to create and test its own designs. For example, if you wanted to refurbish, expand, or construct a medical facility, you would collect current data about everything from space planning and patient movement, feed it to the software, and ask it to look for patterns that help it then maximize a design for a specific set of key performance indicators (KPIs).

The challenge is that you need good data to start with, and if you don’t have a system in place to collect it, you need to set one up first. The implementation of AI systems in future facilities could also affect their design. For example, currently airports have a lot of design features to make waiting more tolerable. But with real-time AI streamlining air travel, airports in the future might no longer need the classic collection of newsstands, lounges and restaurants, and coffeeshops.