— 11 min read
Harnessing the Power of AI in Construction
Last Updated Aug 9, 2024
Artificial intelligence (AI) can change how construction projects are done, freeing up time and space for the most important parts of a job by streamlining several steps in the build. Unlocking AI’s potential in construction requires knowing how to use it, how to identify its limitations, and how to introduce it to the people doing the work it's meant to enhance.
This article explores AI in construction, including deployment strategies, best use cases, common challenges and best practices for using it effectively.
Table of contents
Defining AI in Construction
Most of the things currently referred to as “AI” in construction are technically artificial narrow intelligence (ANI). ANI relies on structured systems designed and trained to perform a specific task or set of tasks. Everyday examples include map apps that suggest routes based on real-time traffic, email inboxes that filter out spam or ChatGPT and other large language models (LLMs) that have been designed to understand and generate human language.
Generative AI, however, unlocks new levels of potential for the construction industry. Generative AI explicitly depends on large amounts of data in order to identify patterns, make predictions and provide insights. Its ability to take unstructured information and address complex issues and tasks makes it an ideal tool for construction projects. Companies should be intentional about how they deploy AI to collect and manage data in order to get optimal results.
Building a tech stack of tools to help efficiently use AI for data acquisition streamlines workflows that are already common in effective project management, such as project schedules, material inventories or safety inspections.
First Steps to Implementing Construction AI
The success of innovation in construction isn’t defined by how popular it is, how advanced the tech is or how much it promises to revolutionize the jobsite. An effective innovation in construction can be any tool, method or habit that solves a problem, optimizes the construction process or improves final outcomes. AI is no different. Harnessing AI’s power starts with understanding a problem that needs to be solved or identifying an end goal and assessing whether AI is the best solution to get there.
Training the Technology
AI usually promises increased efficiency, but only when it is applied as a tool for solving specific problems and achieving ideal outcomes. With generative AI, the requirements for building effective AI systems have changed significantly. This new form of AI allows for the creation of multiple “AI experts” who can handle unstructured information and data in a way similar to human operation.
Unlike traditional machines, people have the ability to fill in gaps using their various specialties, and generative AI can now mimic this capability. By leveraging generative AI, the boundaries of technology is pushed to achieve things that were never possible before.
Our approach is to model AI similarly to how we model humanity. We don't hire one person who knows everything in a company; we hire specialists with deep knowledge about different aspects of the business. For the most part, you don't hire a ton of generalists. AI should not be treated any differently. The bigger the models, the more expensive they are to train and operate, including power consumption and other factors. It's not feasible to have one AI that knows everything about everything—that's not realistic.
At Moss Construction, we’ve developed specialized AI systems with deep knowledge in different areas of the business. We’ve developed 16 AIs, each trained as a specialist. It's like putting someone through school: we choose a university program for a specialty, buy all the textbooks, and use that material to build these models. Our AIs are trained on validated, accurate textbook reference materials, not just on information from Wikipedia or the internet. This ensures the accuracy of the responses generated by our AIs.
George Franco
5 Key Applications of AI in Construction
The uses for AI are vast and expanding all the time. AI can be deployed across various areas of an organization and programmed to collaborate with other AIs to address challenges that arise on projects, including procurement, risk mitigation and coordinating work.
We’ve built a platform called Moss AI that is deployed within the company. It's not about having one model that knows everything; it's about creating a digital twin of our operations, just like we have accounting, legal, and other departments. Each AI is an expert in its respective field, and the platform facilitates collaboration between them. Just like humans call meetings with different experts to solve problems, you can schedule a meeting with various AIs. These specialized, detailed, and accurate AIs collaborate to help address the challenges you face, making your job easier.
George Franco
1. Material Procurement
Material management and value engineering has historically required someone to spend hours and hours on locating required items, finding the best price and determining when they can be delivered. AI can help automate much of the takeoff process, including analyzing digital plans and quickly generating lists of material quantities and cost estimates. AI can also help with comparing numerous options for a component and recommending vendors that are more likely to be able to provide it in an ideal timeline.
2. Optimizing Team Performance
Most contractors work hard to decide how to build and schedule teams that are ideal for a given project. AI is able to analyze data and provide recommendations around assembling and scheduling the ideal team for any given project or phase or a project. This process might include assessing workers’ certifications, analyzing past performance, and comparing efficiency with different partners.
3. Accurate Financial Forecasting and Monitoring
AI can analyze spending trends and historical data to provide more accurate forecasts. Digital tools can help visualize current spending, compare it to projections and recommend improvements based on past projects. This type of real-time financial monitoring can help identify trends or financial blindspots.
4. Risk Management
AI is able to analyze historical projects and data about claims to dramatically improve the processes of risk assessment and applying for sureties. Data can help inform decisions about accepting or bidding on new projects, and analytics can help identify potential risks and take preventative measures, which can lead to a reduced number of claims and possibly lower insurance rates.
5. Identifying Potential Safety Hazards
One of the most sophisticated applications of safety AI involves multimodal AI, which is basically image interpretation. AI can assess data from cameras around the site to identify potential hazards, such as fall hazards, obstructions, proper usage of personal protective equipment (PPE) or proper material storage. Multi-modal AI can perform a full OSHA-level inspection on images, identify everything that's wrong on site and explain how to fix it.
Safety AIs are trained in construction management, safety science, and various reference materials. In some cases, they can even be trained on local building codes.
So this specific use case came from our project in Hawaii, where we're building a solar farm. There was a trench on site, and one of the stops had put a piece of plywood as a temporary bridge over this trench. Someone on the job site felt that the trench might be a potential hazard.
They took a picture of it, but it was a bit vague on the OSHA standards whether the trench was allowed or not. The AI consulted with our engineering AI. Behind the scenes, we have a collaborative system where if one AI isn't confident in its ability to answer a question, it can consult other AIs.
The engineering AI analyzed the image, vectorized it, and performed all the necessary engineering calculations. It measured the dimensions from the image and showed all the formulas and dynamics of using that material. It then calculated the average weight of a construction worker and determined that the plywood could sustain the weight. However, it also found an OSHA code that required a safety factor of four for such scenarios. Based on that, it concluded that the piece of plywood should not be used in that condition.
The person on-site sent me the screenshots of these analyses. Then, the AI called Moss Legal, another AI we have, to write a notice to the subcontractor, informing them that they shouldn't be using the plywood in that way. Moss Legal generated the whole legal notice to send to the subcontractor. In this way, different AIs collaborated to ensure safety on the jobsite.
George Franco
How to Overcome Common Construction AI Challenges
Address resistance with open communication.
People can be resistant when they’re being forced to implement tools they don’t understand, don’t see the use in or fear could put them out of work. Taking time to listen to people’s concerns, address them directly and explain why a new innovation is being adopted are often the best ways to get people invested. Further, the people with these concerns are also those with the most expertise. Making space to talk openly can enlist that expertise in finding ways to maximize AI’s impact.
Cultivate shared values and collaborative opportunities.
Team culture built on values such as open communication, safety and the pursuit of excellent outcomes for clients moves the focus from people’s fears about AI towards its potential as a tool. Similarly, having a culture that values experimentation and doesn’t penalize failure can help people feel more open to trying new innovations.
To help foster collaboration around AI, contractors can create internal focus groups that allow team members to give feedback and suggestions on what information to feed AI.
We have an entire safety department that was introduced to this technology and began deploying it. We started receiving recommendations from people in the field, like, 'I want to make sure these standards go into Moss AI, and that AI gets trained on these standards and references used on the job site.' To address this, the safety department developed an AI committee led by their vice president, with a few other members participating. Anything that gets fed into the AI is first filtered through this committee before it reaches me.
As the company becomes more sophisticated and more departments start using AI, we'll have subcommittees from each department making these decisions. By the time I receive the information, it's basically been approved. If something goes into AI, you can't easily make the AI forget it because we are fine-tuning and training these models. There are methods to revert, but it's not ideal. We've taken a very structured approach regarding which AIs get built. We have what I call the AI Center of Excellence. This is a corporate-level group that makes the bigger-picture decisions.
George Franco
Be consistent with data.
A significant challenge in the construction industry is the amount of fragmented data collected from projects. Because operations are generally decentralized, both employees and AI have to be trained on how to gather the right data and streamline the process across projects.
Data should be collected, stored and analyzed systematically and regularly. Effective data systems often means automating the intake and analysis of data, such as automatically parsing through requests for information (RFIs) to look for trends, as opposed to a person needing to command and manage that process. This alleviates the workload on people but also allows AI to analyze for things that humans might not think to look for.
Build trust.
It’s hard to trust what we don’t understand. Many people don’t know how AI works or how it can complement their work. Building trust often starts with training people on how systems work, including information on its limitations.
While AI can seem intimidating, it’s important to understand that AI introduces new complexities, but doesn’t necessarily make everything easier, or mean employees lose jobs. By managing the increased complexity AI brings and ensuring people are properly equipped with AI tools, employees and contractors will remain essential to the work being done on jobsites.
Benefitting From AI in Construction
In the long run, the people working in construction that will benefit the most from AI aren’t necessarily the most tech-savvy. They are the people who value continuous improvement, seek ways to improve what they do and are open to new ways of thinking. Effectively implementing AI is similar to effective project management: it requires identifying problems, coming up with solutions and being open to ways to improve.
For all AI is already doing, its future is still largely unwritten. Being part of the process of embracing, creating and giving feedback on AI innovations increases the likelihood that the future will be one where technology is a tool that benefits people and helps them spend more time on the most important and rewarding parts of their job.
Was this article helpful?
Thank you for your submission.
100%
0%
You voted that this article was . Was this a mistake? If so, change your vote here.
Scroll less, learn more about construction.
Subscribe to The Blueprint, Procore’s construction newsletter, to get content from industry experts delivered straight to your inbox.
By clicking this button, you agree to our Privacy Notice and Terms of Service.
Categories:
Tags:
Written by
George Franco
George Franco is the Vice President of Advanced Technologies and Innovation at Moss Construction. George spearheads the shaping and execution of a comprehensive innovation strategy that aligns with Moss’ overall growth objectives, including identifying and implementing emerging trends and technologies into the firm’s core operations. With more than 20 years of experience in technology and 17 years at Moss, George has concentrated his efforts and has played an essential role in Moss’ technological evolution. He began his career in construction operations and scheduling, before driving the implementation of virtual design and construction (VDC) and building information modeling (BIM) companywide. George earned a business degree from the University of Florida and the distinction as a top-of-class graduate from the U.S. Navy’s engineering core. He earned graduate certificates from MIT’s Innovation and Technology program as well as Stanford’s Graduate School of Business. George sits on the Board of Advisors of Florida International University’s School of Construction Management.
View profileJames Hamilton
43 articles
James Hamilton is a writer based in Brooklyn, New York with experience in television, documentaries, journalism, comedy, and podcasts. His work has been featured on VICE TV and on The Moth. James was a writer and narrator for the show, VICE News Tonight, where he won an Emmy Award and was nominated for a Peabody Award.
View profileKris Lengieza
Kris Lengieza is the Global Technology Evangelist at Procore Technologies. Kris brings a wealth of experience and passion to the intersection of construction and technology. Previously serving as the VP of Global Partnerships & Alliances, Kris oversaw a diverse ecosystem spanning channel, ISV, public, and association partnerships. His recognition as one of the Top 40 Construction Professionals Under 40 by ENR and BD&C underscores his impact in the industry. Kris’ journey began with 15 years working in the construction field, where he embraced technology as an early adopter and strived to seamlessly integrate data across all construction solutions. As a futurist and construction tech evangelist, Kris now collaborates extensively with industry innovators, tech organizations, and construction companies. Together, they explore transformative technologies that promise to revolutionize our work processes. Kris has played a pivotal role in Procore’s product strategy, delivering industry and technology insights to improve how Procore’s solutions serve the industry.
View profileExplore more helpful resources
Construction Permitting: Essential Steps for Compliance and Approval
Before breaking ground, every construction project must pass through an important checkpoint: securing the necessary permits. Construction permitting isn’t just a formality — it’s the process that guarantees the project...
Virtual Design and Construction: Technology for Building the Future
Technology is transforming industries from the medical field to the world of entertainment. And construction — although a historically slow industry to adopt change — is no exception. Today, there’s...
Unlocking the Power of Construction Data Analytics: Insights for Better Decisions
In the construction industry today, data is being collected constantly — on phones, on computers and tablets, by cameras — and even on some of the devices used to perform...
Key Types of Estimates in Construction
The challenge of determining the cost of building a construction project can seem like an impossible task. Considering the unique nature of each project and the fact that material and...