In Construction, AI Can Be Your Superpower

The construction industry is the industry of superheroes. In a typical superhero movie, evil external forces invade a city, a country – or even the whole universe – without warning. The authorities cannot find an effective, systematic way to defeat them using their everyday arsenal of weapons. So, they turn to superheroes who have the extraordinary powers and knowledge needed to defeat the invading forces. Every mission is unique and different, requiring heroes with specific skills, powers, and strategies.

Just like the mission that is central to most superhero movies, every construction project takes place in a unique location, with specific requirements and goals. So, AEC companies assemble and deploy their highly skilled "heroes" to complete their design and construction mission.
But unlike a superhero movie, in which the heroes rarely age or lose their powers, the superheroes of the AEC industry are beginning to retire. That personnel attrition, plus the growing global climate emergency, increased project complexities, lower profit margins, and lack of productivity, brings lots of pressure to our industry. This is why we need to think differently and solve our problems systematically.

Technology (and more specifically, AI) is one of the weapons the construction industry can draw on to help solve these problems. AI technology, if built the right way, can codify the valuable, hard-earned knowledge of our construction superheroes for generations to come, ensuring that it will never be lost.

AI In Action

As an example, teams at Haskell have been looking into codifying the knowledge of their welding inspectors and augment their capabilities across their projects. Haskell builds a lot of manufacturing facilities and distribution warehouses for the food and beverage industry. In these facilities, they must make miles of stainless-steel pipes that carry the factory's main products, such as wine, milk, and yogurt. To build these long pipes, a certified welder needs to weld many shorter pipes together.

To ensure that these welds are of the highest quality, they have a team of two certified weld inspectors. These very highly paid inspectors go to the sites and visually inspect each weld. The challenge is that the two certified weld inspectors cannot inspect thousands of welds in every hundred projects at the same time. So, the ratio of the quantity of welding that needs to be inspected to the number of inspectors is overwhelmingly high.
The solution?

Gary, a certified inspector at Haskell with more than 40 years of experience, talked to the innovation team at Haskell about putting a virtual AI inspector in the hands of all welders. What if welders take a photo of the welding they just did, and the virtual inspector tells them that it was a good or a bad weld.

To train such an AI system, the innovation team at Haskell faced a significant challenge. They did not have enough examples of good and bad welding. But, this challenge did not stop them.

They went to the construction site and took many welding photos from different angles and under lighting conditions. Then with the help of Gary and his immense inspection knowledge, they classified photos into good and bad welds. This is how they created their initial dataset.

Working with a technology partner, they created an AI system that can tell the welders if their welding is good or not. Now, all the welders have a "virtual welding inspector" in their pockets, empowering Gary and the other certified inspectors to receive many photos of wildings taken every single day. Now, every single weld is being inspected and documented for the future.

With this solution, both Gary and the AI solution improve over time: Gary is enabled to inspect many more welds at the convenience of his home or office. For instance, if he used to inspect 50 welds a day, now he can virtually inspect 1000 welds a day. Hence, practicing more and getting even better! How about the AI system? When the AI system is in doubt and provides a solution with less confidence, Gary can correct it with his immense knowledge and put that data back into the AI system for future training and improvements. It is as if Gary is teaching the virtual AI inspector!

Your Knowledgeable Superheroes

Your knowledgeable superheroes, like Gary, will retire one day, but their knowledge could stay in your company if you build an AI system that mimics their capabilities. The benefit of such AI systems is that, unlike your superheroes, they are not bound to any physical location. Therefore, they could be scaled across your organization in the hands of all your employees and stakeholders (like welders at Haskell).

Who are your superheroes, and how will you augment their capabilities and preserve their knowledge in your company for the generations to come?

Excerpted from "Augment It: How Architecture, Engineering and Construction Leaders Leverage Data and Artificial Intelligence to Build a Sustainable Future."