When artificial intelligence is used effectively, there is no better way to extract valuable information from data and proactively monitor organizational performance. Its applications in construction are vast and continually expanding. Complex computational tasks are now being automated. AI can find, in seconds, what may take a human many hours of trawling. The best AI solutions are also adaptive, able to learn and improve over time—saving teams precious time while increasing accuracy and lowering costs.

Critically, AI solutions are built on data and designed to recognize incredibly complex patterns that no human (or dashboard) could. The best uses of AI exploit this ability to return highly specific and precise predictions that are tailored to an organization’s past performance.

One of the biggest inhibitors of AI uptake is deciding where to start. To find the answer, organizations must look inward. First, determine where the organization is hurting. Pinpoint the cause as precisely as possible. From there, data must be put in order. This will almost certainly involve the configuration of a backend data infrastructure. In AI, and especially in machine learning, the more data that can be collected and centralized, the better. Once an organization truly understands its challenges and has a solid data backbone, AI solutions can be brought online.

Increasing team productivity and efficiency

AI can have big benefits for the onsite construction team as well as the back of house project management and controls team. For those working onsite, greater consistency, improved productivity, enhanced safety and, ultimately, better decision-making are all possible with automated processes and faster, more accurate reporting. A sampling of onsite construction technologies would include:

• Construction robots. Robots that can lay brick or automatically reinforce and pour concrete are currently in development. Though still experimental, the intent is for automation to make the physical work of construction safer, faster, more consistent, and significantly cheaper.
• Drones/site surveying. Tracking project completion is a manual and resource intensive task. Drones and other site surveying technology can automate the inspection process and generate accurate, real-time progress reports. The latency between an action being completed on site and becoming discoverable to project managers is greatly reduced.
• Automated safety inspections. Computer vision techniques can be used to monitor a site and automatically send alerts when a violation is observed. For example, if a worker is detected not wearing a safety harness, the site supervisor will automatically receive a text or email alert. This means safety risks can be identified in real-time (before becoming full-blown safety violations) and sites can be inspected remotely.
• Work sequencing. On a construction site, many different workers must complete many different tasks. The question of exactly how to schedule this work in the fastest and most cost-effective way is hugely complex. Algorithms used in the field of logistics for order fulfillment have been adapted to find the fastest and cheapest work schedules without manual computation.

There are also several advantages that AI can bring to project management and project controls teams:

• Estimating resource requirements, costs and schedules. The project controls team makes key decisions about project execution based on the project data available to them. Machine learning models excel at using data to make accurate and precise estimates. This translates into better resource planning, more accurate cost and schedule estimates as well as improved responsiveness to any issues that may arise.
• More precise project estimation. AI methods can be trained to learn which features and scenarios are most closely associated with good or poor past performance in order to identify red flags in a project at the planning stage. In this way, steps can be taken to proactively mitigate risks at the outset, rather than constantly reacting to predictable problems.
• Quick scenario generation and sensitivity analysis. Project feasibility analysis is a perpetual challenge, AI methods can be used to model many different project scenarios very quickly, allowing for a more thorough project evaluation. This can be used for sensitivity analysis or to test the feasibility of a given plan across a wide array of business scenarios. It can also help ensure only those projects with the greatest chance of success are executed and that project risks are well understood at the outset.

Solutions in all shapes and sizes

Some feel that AI is still an experimental field that involves a huge investment of both time and capital to be successful. While this is true for cutting edge applications like automated construction robots, there are many advanced tech solutions that have been around for a decade or longer that are widely used, well-tested, and surprisingly easy to implement.

A common misconception is that organizations feel as though they must build their own AI solutions in-house or restructure their entire workforce in order to realize benefits. Today, this is decidedly not the case. There are many pre-built and easily customizable tools available that can plug right in—instead of building an application from the ground up, all that is required is a little customization. With just a little training, key employees can learn the skills to use and interpret AI solutions—hiring a squadron of data and advanced tech experts isn’t necessary (although hiring a few won’t hurt).

Gearing up for AI

In a nutshell, many good AI solutions already exist, and much can be attained with the standard tried and true approaches and looking inward is always the first step to embracing AI. Knowing specifically what the organization does (and does not) want to improve should guide the entire AI journey. Start small, stay focused and build from there.

Finally, it is important for key decision-makers to have a solid understanding of AI in construction and essential that a data-informed culture is created. If using basic metrics is a struggle, successfully deploying advanced analyses is likely to be a challenge. With a good data backbone and a little training, however, organizations can lay the foundation for an open attitude toward AI and begin to reap the numerous benefits its solutions offer.