Machine learning in construction has arrived and is helping construction executives and project leaders find real-time insights for managing and reducing risk on a daily basis.

Machine learning is based on the concept that algorithms can parse huge amounts of data, learn from it and then predict something based on what it has learned. Machine learning is usually considered a subset of artificial intelligence, which is the broader concept of a machine being able to carry out tasks for humans (e.g., autonomous/self-driving cars).

Sound complicated? Not for the end user who benefits from the deployment of machine learning-based technology. The use of machine learning should be a seamless part of workflows.

How can machine learning benefit the construction industry?

Projects generate a deluge of data every single day, including text, documents, images, RFIs, BIM models, laser scans, sensor data and more.

What if a smart assistant was on the project to help sift through hundreds of project issues and analyze data to identify the most critical things that require attention today, or to identify which 10 subcontractors out of a 100 are at risk and require attention?

Machine learning can reduce the noise and show organizations the key issues that can have an impact on a project, helping the team identify the most critical things from a quality and safety perspective that need immediate attention. Exposure to data provides transparency to executives. Surfacing risk factors quickly and consistently throughout the project portfolio enables informed and impactful conversations with project leaders on the jobsite.

What kind of competitive advantage does machine learning technology provide to construction firms?

When an executive responsible for ensuring that projects are delivered safely, on time and of high quality can see project risks across their portfolio at a glance, and even drill down to specific projects, the business of construction takes a leap forward. Typical construction reports and dashboards provide information and analytics post-facto, meaning executives are looking at the data only after there was a financial impact. This is the case whether it is quality deficiencies or safety risks and incidents.

Machine learning allows construction firms to predict and mitigate risks before they impact project margins.

For example, problems related to water intrusion are known to be a big cause of downstream costs in the form of insurance claims and warranty issues. Machine learning can help project teams prevent at least a part of these costs by identifying the risks before they become problems, which has a direct impact on profitability.

Are there any challenges associated with incorporating machine learning technology on the jobsite?

Adopting digital processes, such as BIM, and moving to cloud-based software solutions are prerequisites in order for project data to be aggregated and accessible for a machine. For firms that are beyond that initial hurdle, the focus has been on capturing the right data and maintaining consistent data quality.

For example, risks involved with a health care project will be very different from a high-rise residential tower.

Similarly, different trade types have inherently different associated risks.

If companies want the machine algorithms to learn and adapt to specific project types, defining the project and trade information consistently across all jobs is critical. Incentivizing project teams to consistently enter the right information is a huge factor in this, but what’s in it for them?

Machine learning reduces—not increases—the amount of data entry from the team that needs to happen at the project level. As it learns, the machine will infer many details and auto-populate as much as possible, decreasing manual data entry over time.

What role will machine learning play in the future of construction?

Construction firms already are leveraging machine learning, whether it is to identify quality risks at a project level or to scan images and videos for jobsite safety. As more projects start using digital tools, the reach of machine learning applications in construction will grow further.

Machine learning is expected to provide assistive tools to help automate some of the repetitive tasks in construction. If a firm has already taken jobsite photographs, machine learning can identify some of the problems and auto-fill some of the mandatory checklists, as well as assign tasks to the right stakeholders.

Most of the current focus has been on the construction execution phase. If machine learning can insert some of this intelligence into the design and preconstruction stages and start predicting what can possibly go wrong downstream as part of the design process, it would take project delivery risk reduction to another level.