Talking Buildings: a pleasing partnership between people and buildings

Wico Mulder from TNO is an expert in the field of data and artificial intelligence (AI). Wico explains how people and the building, and indeed buildings, can help each other manage energy flows more effectively. They do so thanks to AI in the existing building.

Is the combination of AI and collaborating with people not contradictory?

"'No man is an island', wrote the English poet John Donne almost 400 years ago. What he meant was that no one is truly self-supporting. Recently a colleague asked me if the same applies to AI. For Talking Buildings the answer is simple, namely that it is correct that AI is not self-supporting, as Matthew Johnson and Alonso H. Vera also observed in their article entitled ‘No AI Is an island’. Humans can simply cannot survive in this modern world without smart algorithms and algorithms cannot exist without humans. Collaboration is precisely what it's all about."

Why does AI hold so much promise as a solution?

"In the Climate Agreement it was agreed that in 2030 the Netherlands would emit half as much CO₂ as it did in 1990. You won't be able to achieve that without AI at an integral level. If you regulate individual buildings you will still have a peak load on the energy grid because, for example, they will all be using energy for heating from 7 to 9 o'clock. This means you have to use a technology which can process information at a different level and which links up with both the building and the energy grid. AI systems are very good at doing this and they have the major benefit of scalability."

What is collaborating with TNO on algorithms so special?

"First and foremost everything we do is intended to increase reliability. If an algorithm takes a decision for the climate control system, it can also explain why. In other words it is explainable AI. Users will only want to adopt the AI if we can guarantee reliability. Secondly we make sure that the partnership between people and the algorithm stays on the right track. What does it do to you if a building takes decisions? Do you just leave it alone, or do you want to intervene every now and then? The third thing is scalability. These days people still have to analyse the data streams in order to link them to the AI model. How can you have AI carry out this step?"

What role do people play in the partnership?

"AI trains itself. In the beginning it may well be the case that the algorithm finds something which the building manager says is impossible. The AI may think that a slightly lower temperature is perfectly fine, or continuing to heat until 26 degrees. In such instances it's important that the manager can make adjustments. That learning effect is at the heart of AI. It's an adaptive system with sensors which measure on the one side and people who can then see whether things have gone well on the other."

What does the ultimate partnership look like?

"Imagine you want to show round a group of foreign visitors at 8 o'clock in the morning. The building then has to be heated up beforehand but you cannot find a switch anywhere. How do you then communicate with AI? One possibility is an app but, on the other hand, you do not want to have to keep checking the app or a screen. Another possibility is for AI to send you a message that it has noticed something out of the ordinary that you should perhaps have a look at. Only then can you really talk of teaming."

Is sharing all that data actually safe?

"In order to prevent problems such as grid congestion it's essential to share data with other buildings. This must be done securely, for that specific purpose, on a data platform on which the informative messages are shielded. TNO is also an expert on data sovereignty, whereby you combine, enrich and analyse meaningful data safely and efficiently. That goes beyond protecting yourself against hackers. It's about the feeling of having control."

What benefits does AI generate, on balance, in existing buildings?

"AIMZ guarantees an energy saving of 10%, although in practice it's usually double that and that's despite energy saving not even being the primary goal. The idea is to get a grip on energy and investments in existing real estate and the energy transition. Is quite a shock when you realise that AI can immediately save so much energy because it directly converts insight into improvement without any human intervention being necessary. Why is the boiler starting to work at such an early time of day? It's not just about observing things, but taking action and AIMZ can immediately make adjustments to prevent it happening again. An energy management system that pays for itself and does not involve any extra work on your part."

How complicated or easy is it to start using AI?

"The AI is screwed onto the existing climate control system without any extra sensors being installed. The innovation requires a minimum of effort by building managers and makes the whole procedure easier. Currently buildings are often already monitored by an installation manager or facilities department. But constantly assessing your buildings in the hope of saving 10 to 20% energy is something you're better off leaving to the AI. AI enables the building manager to work together with his system and this makes it suddenly become effective and scalable."

Piet Jan Bloem is director of AIMZ, a company based in Groningen that is helping to create a CO₂ neutral society by using AI to enable buildings to communicate with each other, their environment and people. Piet Jan talks about the cooperation with TNO.

How has the partnership with TNO benefited you?

"We noticed that we were tapping into a gigantic source of knowledge which resulted in new initiatives. We already had the knowledge to link AI to buildings so that they could improve. In fact, we were initially looking for a technology to improve the sustainability of real estate. However, TNO challenged us to make it scalable, stackable and more flexible so that we could create a more future-oriented business case. That's why we're now working on getting the buildings to learn from each other interactively, so they'll know what to do from day one."

What will the partnership look like in the future?

"We would like the partnership to continue. Although I'm enjoying the challenge of making AIMZ a successful company, it is even more interesting for the Netherlands to be leading the way in implementing these kinds of technologies. When I see what kind of AI knowledge and expertise we have in our country, it's massive. We really are doing good things and by connecting parties and seeking cooperation we can go on to do great things."