Artificial Intelligence: Bits & Mortar

(PHOTO: © SOMKID / ADOBE STOCK)

The technology sector, or as we often refer to it, “Silicon Valley,” loves to disrupt industries, and specifically legacy industries, so the construction sector has been a prime target, but we’ve yet to see the kinds of change that removed VCRs from our homes and companies like Blockbuster from our shopping malls.

So, what’s the hold-up? Disruption is kind of like a puzzle; solving it requires understanding all the pieces. In the case of construction, that means understanding the development, design and enough of the construction process to better plan and mitigate risks. If technology wants to disrupt our industry, it will need a complete picture, and an understanding of how all the pieces fit together.

Organizations throughout the industry are diving into data collection. Be it in the design, construction, or operational phases of a project, the fragmentation of our industry is making it difficult to both collect data and see how it fits together. There are multiple reasons for this, including the confusion and risk surrounding what information you can and cannot share, whether it is confidential or proprietary, etc. In addition, developers, designers, and construction firms are not eager for their competitors to benefit from their data, so they often hold it close to their chests.

Unfortunately, this siloed approach in the industry has not yielded great results for technological innovation. We have seen attempts to implement foundational changes, such as the introduction of different contract types (public-private partnerships, design-builds, integrated project deliveries, alliance, etc.) coupled with advancements in technology like BIM and reality capture (laser scanning, drones) to mobile devices (iPads, phones, apps) but we have not yet been able to see the entire industry unlock its full disruptive potential. Still, we are betting that that will change soon, thanks in part to the advancements in computer vision as well as artificial intelligence and its subset of machine learning.

Developers and government bodies are trending toward a data-forward approach, accelerating the progress towards better leveraging data for future use, however it is not as simple as having Silicon Valley enter with a new platform or solution, or even throwing venture capital money at a new approach to industrializing construction projects. The key here will be to have skin in the game; having the knowledge of design and construction so that it can be paired with technology to offer efficient solutions effectively.

So, it is not a technology problem we are facing in the industry; it is a paradigm shift towards embedding these new technologies into how we do business.

Construction sites have seen an influx of sensors — data collectors — all through the supply chain. These are installed in drones, cranes, trucks, software and handheld devices. Through the use of these sensors, projects have become data foundries. Our jobsites make mountains of data every day, but they can quickly become data sieves. Machine learning programs are the only thing capable of crunching the numbers, but even artificial intelligence has to know what to look for. Feed an AI system 10,000 dots and it will find the shortest line between them in seconds, but construction has never been that simple.

Artificial intelligence and machine learning are missing context. Different sites have different circumstances, different clients and different contractors, and no two are the same. The reality is that different builders build differently. If artificial intelligence is going to make sense of all that, even with the advances in machine learning, it will need some help.

It might be helpful to look at machine learning as it’s been deployed to optimize climate controls in facility management. Smart buildings can control all of their parts, so it is possible to harness machine learning to monitor and adjust the climate as needed throughout the day.

Machine learning and artificial intelligence can learn the particulars of our hypothetical building and its climate, and the optimized climate controls will reduce waste as well as save on operational costs, so it sounds like tech has saved the day. The problem is that the building’s life-cycle maintenance cost schedule is now shot since most mechanical rooms are not built to be switched off and on 10 times a day. That’s the power of machine learning minus the operational context.

As another example, let’s examine the use of AI to help monitor the progress of a construction project.

We have a lot of technology available to use. We can create sophisticated 3-D models that serve as the base of our design. We can fly drones that capture hundreds of photos to create a digital replica of the project. And an AI system can compare the data gathered to our project schedule. All of this can provide an objective breakdown of current progress compared to the planned schedule, and it can be broken down even further into scopes of work.

It sounds like another area where technology should be revolutionizing the industry, right? The only caveat is that the current industry and paradigm do not support a data-first approach to projects. Therefore, the vast majority of projects would not be able to use a robust 3-D model as a base.

Disruption will inevitably come to the construction industry; there is simply too much opportunity to ignore the sector. But, from all that we have seen, it will take a marriage of the latest technology with legacy expertise. In the quest to find the best path forward, Silicon Valley will need us as much as we need them.

Rory Smith and Hammad Chaudhry are both with EllisDon where Rory is the manager of DDE marketing and Hammad is national director of digital project delivery services.