What killed the Dinosaurs? The AI-age!

Is AI over hyped when it comes to the benefits that could be derived when using it correctly?

I will be recording a session in 8 hours for a Siemens Global Conference talking about building automation and how to go Beyond Buildings. We were discussing the need to use technology to solve problems, and that we need to use tools of the time to solve historic problems. Not only in the sense of the building automation side of things, but also for collaboration purposes. Which is not that dramatic since countries like Australia are locked in for the next 12 months, and we could still use some of their amazing brain power in other places. Especially when my counterpart for the Siemens panel discussion is no other than the legendary Bryce Anderson.

So why continue to collaborate with Word docs and PDFs, when people having could describe their challenges by capturing their reality and invite experts to help them? Wherever they are. We are utilizing this approach in an innovation project I am leading right now for three schools in Sweden, and it helps to get stakeholders on the same page. The 3D-models act as a foundation for future IoT-visualizations. Making reality come to life in a continuous way, forever.

Something that I also heard this week from a self-proclaimed AI-expert was that he was continuously surprised that people in traditional industries (he was referring to the automotive industry) were still researching what food horses should eat to make them run faster. Not understanding that the tools of tomorrow are here today, just lacking a bit of adoption. And companies should be more focused on what Artificial Intelligence needs to feed on.

AI is overhyped as much as Digital Twins, IoT, and everything else to be honest as a stand-alone saviour in otherwise traditional industries. I mean, come on. What are the jobs to be done? how are they being solved now? and why haven't anything happened since... forever?

Is AI overhyped when it comes to the benefits that could be derived when using it correctly? No, quite the opposite, of course. The domino effects of doing the right things, and not just doing things right will have the effect of not only making processes more efficient. But also eliminate the need for many due to a completely different process altogether.

The questions should not necessarily be what is wrong with any industry, or if industries are behind. Or focusing too much of the problems that do exist. Because the challenges vary, a lot. And the solutions vary, a lot. It is not about technology, it's more about the people, processes, culture, tools as well as the decision makers that are in each region.

Digitalization efforts are taking time. It is not for a lack of trying. It is for a lack of understanding. Where most Digitalization experts outside the Real Estate industry do not have a clue about buildings. And they think it is as easy as taking data out and then do something with it, not understanding the need for context.

What I do agree with is that we need to use modern tools, or we will be forever destined to be stuck in the past. We are getting there with the help of a pandemic, and it must be about profit as well as a planetary purpose in my mind. One big question right now seems to be if the API-economy of buildings will help in making sense of the data? Or just make it easier to get out, but potentially make it more challenging in the long run?

I wrote about the dangers of the API economy 2,5 years ago, and it still holds true for the most part in the sense that an API-economy might be the stepping stone to more problems in the future. Maybe today more than ever. Other industries are experiencing the problems with data lakes turned into data swamps and some of the largest event-streaming initiatives have also missed out on the meta-data tagging part. Failing to understand contextual relationships at scale.

However, the phenomenal part is that companies and industries seem to utilize taxonomies and ontologies more and more which should make it easier for everyone to digitally onboard buildings faster. There are still challenges with industry specific taxonomies and mapping data too early, too hard to something that cannot scale across domains. But I believe we are on the right track.

And I am of course proud of the Swedish wonder, REC - Real Estate Core, where representatives from my Alma Mater have been instrumental in bringing it to the world (Jönköping University). For the ones that have missed it, Microsoft have partnered up with;

RealEstateCore (a Swedish consortium of real estate owners, software houses, and research institutions) has delivered an open-source DTDL-based ontology (or set of models) for the real estate industry, which provides common ground for modelling smart buildings while leveraging industry standards (like BRICK Schema, W3C Building Topology Ontology) to prevent reinvention.

This is a great start for owners to utilize the data for industry-specific applications within the realm of real estate. During one of the latest published episodes of the Beyond Buildings Podcast we discussed aspects of Ontologies and Taxonomies with the legends, Terry Herr, Joel Bender and the co-creator of Real Estate Core, Erik Wallin.

This image was something I scribbled down after the discussion which was late 2020's, and it has been updated since then. But it shows somewhat that the initiatives back then focused on the real estate realm, and quite a lot on the building automation side of things. Which is great, but if you think about AI, it is also a lot about the context. And not always the context you would immediately need.

There are phenomenal companies out there solving silo specific needs and that is exactly how it should be. They solve foundational challenges for the energy side of things, building automation, the proptech-angle, and in the realm of fire, evac, audio/video, elevators, lighting, FM-services, and everything else that has anything to do with buildings. There are always dangers of the thousand cuts problems, but it seems that we are getting to a future the world desperately needs. But how do we capture meaning from all these existing systems and turn them into fodder for AI?

How can we make all of this come together?

One of the most terrifying experiences when I was a kid was when I sat in our basement alone, watching the movie Event Horizon almost 25 years ago. There is a segment in the movie when they think a guy says, "liberate me", which means set me free. It is later corrected to libera te tutemet (ex inferis) "save yourself (from hell)".

Which is what an AI-first mindset could help companies do, having learnt from other industries that are stuck in API hell and what their challenges lie in leveraging AI at scale. Data lakes could easily turn into data swamps if not understood correctly.

What does AI want for breakfast, lunch and dinner?

AI could eat a lot of things but it certainly prefers quality data. Preferably seasoned with some meta-data. This could come from existing systems, but also something that most people tend to forget. People. People are still providing initial as well as continuous contextual intelligence for systems and AI-initiatives, adding their domain expertise in a continuous way. Where systems might be fantastic in isolation, but when it comes to the real world, bringing systems and people together is a challenge of its own. Most often because domain experts have no idea how to communicate with each-other across domains, and the fact that they do not necessarily know where boundaries stop and start. This takes time that companies do not have, nor the planet.

But fear not my friends. Remember that all the tools and technologies in the Building Buzz cycle are available today. If there is a will (this might be the problem...) there's always a way!

Because what better way for explainable AI to be explained through that of Digital Twins?

My perspective of Digital Twins, as well as AI, is that on part should be made for people. And the other side for machines. We need to create a world for humans, and machines. We got graph based-ways of working, taxonomies, ontologies, RDF vs LPG, and everything else that can help in recording reality in the best way possible. What would that look like? And what would the benefits be?

Beyond Real-time collaboration, BIM+ Real-time data.

In the movie Event Horizon which I talked about earlier we get an explanation of the question;

"What is the shortest distance between two points?"

I will not spoil the movie for the ones who have not seen it, but that 2-3-minute segment has been with me for the last 20 years. And in traditional click-bait mannerisms, it is not what you think.

The shortest distance is if you can connect both points in space and time. With the combination of real-time data, IoT, virtual reality, taxonomies, we can define our reality and let anyone in the world innovate with us. Not limited by space and time, where the Digital Twin acts as a time-capsule.

How else could I get one of the leading Building Automation specialists is in Australia to help me out in the best way possible? By utilizing the combination of modern technologies, we finally have a way that will allow people, machines, AI, past, present and future to coexist in the same space and time. What can we understand then? What would this to for the skill shortage gap when we could capture the knowledge of domain experts, and that of industry guidelines into a digital twin? This could also be used for education purposes, and not only capture knowledge, but much more easily transfer knowledge from systems to systems, and via domain experts to domain experts.

In the project I am leading we aim to bring more transparency to the dynamic interaction between real-estate, district heating, and energy producers in a smart city context.

Simply put, we will connect temp, humidity, CO2, sensors for 70+ rooms with actuators to 226 radiators to control the system via AI-based ways of working. We will go for a distributed intelligence approach, utilizing wireless Modbus, map it to tagging standards, like Haystack, Brick, REC, and create one API to the building. As Sabine Lam wrote last month, Digitalization is surely changing the role of what integrators need to do in order to try get towards future ready facilities. We will utilize wireless mesh, wireless Modbus, standardized ways of working and a combination of RESTFul, MQTT, and in discussions with graphql approaches and Kafka-based streams where everything is built to be bi-directional and data tagged according to multiple standards leading to truly software defined buildings.

Owners should be in control of their past, present and their future and allow others to create value from their buildings. Not having to ask for permission where data and control strategies are held hostage by vendors and integrators.

This is the starting point for this project and there are a lot of other benefits involved as well.

What we also want to do is to lay the foundation of Digital Twins by utilizing the drawings, turning them into 3D-models, and then adding real-time data from everything. Lighting that digital twin up like a Christmas tree. And then just animate it with occupancy so that we can SEE how the system is working, and how it is affecting its context, and how the context affects the system. In real-time if that is interesting, but even more so, do the calculations and simulations and run the building from the future. This will enable all stakeholders to finally pull down the curtains and create transparency into "that Voodoo that you do". And understand that it is not only about the buildings themselves. It is about the context.

We need to start creating the future, today.

The ingredients exist. The people. The systems. The processes. But they are not in sync. Everyone and everything are operating in separate realities. The combination of modern tools and technologies can change that. Not only to bring people and systems into a shared reality, but also allow for AI-augmentation based on a contextual data fabric. Digital Twins for instance, can create an organizational wide understanding of the actual content, the products, AND how they relate to usage, supply chain and the overall lifecycle. Steve Jobs explains it well enough when it comes to the obsession with processes, neglecting the content and context. Modern approaches done right could easily help companies make decisions 9 times faster, with access to real-time data. What would that do for any industry?

This below was made by the company SEKAI which have a very future oriented approach when it comes to Digital Twinning. They combine open-source technologies, making them come together through a Digital Twin enablement platform. They have an ML-assisted way of ingesting data from different sources, tying them together through graph-ways of working and ontologies. And then exposing the combined data sources via a Data notebook to simplify data manipulation for data scientists, as well as low-code interfaces for faster value creation. What I like about Digital twins in general is that they aim to work as a complement to existing solutions, bringing them together in a way that could potentially record any reality that exists, has existed, and simulate realities that are about to exist.

vi. CAD + IoT + Ontologies + AR + VR = Reality Recording at scale via SEKAI Digital Twin Platform If you look at the most modern platforms and approaches, this is what is going on behind the scenes. The Digital Twin Consortium members are all about similar approaches, albeit in separate offerings, and in several industries like Healthcare, Manufacturing, Natural Resources, Infrastructure as well as Aerospace and Defence. And Chalmers Digital Twin City Centre, which I recently joined, they are also working with similar tools and technologies.

Again, everything exists. "It's just bringing it together". But some take longer than others and work is not always automated, nor simplified through no-code/low code approaches to democratize modern value creation at scale. Time is the most valuable currency we have where open-source tools and technologies do not always have to worry about up-time and SLAs. That is what enterprise platforms are for. The real world usually demands solutions that are robust, useful, and attractive solutions that get the job done. But it is never an either or. Which is why we need interoperable solutions coming together,