The Uses of AI in Construction and Evolving Industry Standards
Category | Briefing Papers
By Matthew G. Nelson
Matt is a member of the firm’s Construction Law Department. He can be reached at 612.359.8611 or mnelson@fwhtlaw.com.
Introduction
Change is constant, and it affects the standards of care expected from contractors and design professionals in the construction industry. Artificial Intelligence or “AI” is the emerging technology that could change existing standards of care. An example from nearly a century ago illustrates how changing technology can modify expected standards of care. On March 10, 1928, two barges loaded with coal “were lost off the Jersey Coast…in an easterly gale” while in tow behind two tug boats.[i] The tugs were not equipped with radio receiving sets, which would have picked up broadcasts of local forecasts, and alerted the tug captains to the oncoming storm in time to put in at a nearby breakwater. The loss led to litigation over whether the radio-less tugs were seaworthy for the task. The tug owner’s defense was that it was not standard custom in the 1920s to have radio receivers on board, so the vessels were seaworthy. While the court recognized there was not at the time “a general custom among coastwise carriers so to equip their tugs”[ii] with radio receiving sets, it nevertheless determined that the tugs should have had radio receivers and, because they did not, they were unseaworthy and liable for the loss:
There are, no doubt, cases where courts seem to make the general practice of the calling the standard of proper diligence . . . In most cases reasonable prudence is in fact common prudence; but strictly it is never its measure; a whole calling may have unduly lagged in the adoption of new and available devices. It never may set its own tests, however persuasive its usages. Courts must in the end say what is required[.][iii]
While it was likely a jolt for the coastal shipping industry at the time, the court’s decision rests on the perennial truth that change is the only constant. Members of trades and professions adapt their practices to incorporate new technologies, and standards of care evolve to reflect these adaptations in practice. The construction industry is currently facing the same challenge, with standards of care likely to evolve as AI is increasingly utilized by contractors, construction managers, or design professionals. This Briefing Paper discusses some of the changes AI is poised to bring to the construction industry, and how this could influence industry standards.
What is AI?
Last year, Turner Construction Company’s chief innovation officer predicted that “[a]rtificial intelligence will transform our industry in the next 10 years more than any other technology in the past 100 years….It’s not a question of if. It’s a question of when.”[iv] AI is the currently emerging technology that could precipitate change in construction management and practices in and for the foreseeable future.
AI refers to computer technologies that mimic human intelligence. Face recognition programs and self-driving automobiles utilize AI-enabled programs, and so does Amazon to recommend future purchases based on past purchasing habits or patterns. AI systems apply algorithms to data to arrive at conclusions based on patterns identified in existing data. The robustness and reliability of the outputs depend the quantity, and quality, of the data. AI systems not only mimic human intelligence, they do so by “evaluating” vast amounts of data at rates that far outpace ordinary human capabilities.
The AI du jour is ChatGPT, which is an open, large language model, generative AI program. The term “generative” means that the AI system will synthesize and create new or novel content or “judgments.” This is contrasted with analytical AI systems, which classify and group data in order to make conclusions or predictions (such as weather forecasts for the modern shipping industry), or to illuminate data patterns and “insights” which users can then utilize to chart a course. While analytical AI systems have been around longer, the recent emergence of generative AI systems such as ChatGPT is responsible for the increased attention these systems are now receiving.
The term “open” in the AI context means that the data set is unrestricted, such that any information users “feed” or upload to the open system is not only used to “train” and refine the large language model, but also becomes publicly accessible. You wouldn’t want to feed another’s protected intellectual property—such as an owner’s or architect’s plan drawings—into an “open” AI system (unless of course the owner of the intellectual property consents). There are also “closed” or enterprise systems, that companies or other entities utilize to develop and deploy confidential business intelligence. These “closed” systems are proprietary and exist behind an enterprise “firewall,” so to speak.
Ultimately, AI systems are good at sorting through vast amounts of information, and doing so much, much more quickly than humans. These features present intriguing opportunities for the uniquely complex field of modern construction:
Construction is technologically complex: Construction comprises a host of applied sciences, such as architecture; the engineering disciplines of civil, soils, structural, electrical, mechanical, and others; the materials sciences that govern the extraction, formulation and manufacture of building materials; and, intricate principles of construction and construction management that address the practical building process. Construction’s technological complexity is amplified by its invariable unique conditions—most projects are unique, built to a unique design, on a unique site, by a unique aggregation of companies, operating without economies of scale in an uncontrolled environment, where productivity is affected by weather, geology, local labor skills and availability, local building codes, and site accessibility. Major challenges on every project are the accurate communication of technically complex information to the myriad parties involved in the construction process, and the management of its uncertainties.[v]
For example, an architect could commission an enterprise AI system, upload all of its proprietary business information and intellectual property into that system, and grant varying levels of “clearance” to employees tasked with generating new designs or plans, or with addressing requests for clarification or information and then disseminating those responses. Similarly, a contractor could utilize an enterprise AI system to refine its takeoffs and cost estimating; prepare highly tailored competitive bids; develop, track, and amend schedules; generate value engineering opportunities for owners; or precisely chart site access and sequencing models to streamline coordination efforts. In fact, some contractors are already utilizing AI systems in their work.
Some Actual (Potential) Uses of AI in Construction
- Estimating and Bidding
Any time a contract leans on the words “reasonable” or “reasonably,” it is importing a legal standard of conduct or care from the industry, similar to the incorporation of a “seaworthiness” standard in the T.J. Hooper case. For instance, the ConsensusDOCS define a contractor’s “Work” to include “construction and services necessary or incidental to fulfill Constructor’s obligations for the Project in accordance with and reasonably inferable from the Contract Documents” and require the contractor to “provide all labor, materials, equipment, and services necessary to complete the Work…includ[ing] any Work reasonably inferable from the Contract Documents.” [vi] The AIA General Conditions similarly define the scope of Work by what is “reasonably inferable” from the Contract Documents . [vii]
Contractors typically face time and resource constraints when developing competitive bids. Given the vast amount of information included in bid packages for complex projects, contractors risk becoming responsible for the costs and time required to address undetected scope gaps or under-designed components of projects, if they are considered to be “reasonably inferable” from the project information. Some companies are developing AI-based preconstruction programs to help contractors mitigate and address this risk.[viii] These programs analyze design documents and drawings to flag potential errors or gaps, identifying and escalating issues to the bidding team to address, by requesting clarifications or otherwise. This has the potential to result in more complete project plans and more precise bids. The increased transparency could also shift standards for what is or should be “reasonably inferable” from project drawings.
- Documenting Work and Identifying Impacts
Contractors are well aware of the use of drone-mounted cameras to document or inspect work, particularly on vertical construction sites, bridge spans, or renewables projects. Now, companies in the AI space are adapting this approach to other applications.[ix] Some of these systems utilize positioning systems and LiDAR cameras that can be mounted to hard hats, much like a GoPro unit, in order to film, photograph, and track progress while work is underway or during dedicated site walks. Others utilize the camera-equipped mobile devices that superintendents and QA/QC personnel are already toting around on building sites. AI systems can overlay as-built site documentation and personnel positioning data onto schematics and plan drawings, with changed or to-be completed work, and workforce information, highlighted and tracked to specific dates and locations. All of this information can be routed to home office computers or other mobile devices via product-specific applications[x]:
The technology can also be utilized to submit RFIs instantaneously from the field, by capturing an image of the condition in question and pinning the inquiry to the image, overlaying it directly onto the project plans, and submitting it to the project team.
The potential benefits for developing schedules, look-aheads, and schedule revisions are apparent. Equally apparent are the potential uses for coordinating work while documenting progress or deviations, generating and working through punch lists, and assembling close-out information. While some companies have developed personnel- or drone-mounted products in this space, others have developed robots with 360 degree cameras to walk and document facilities and other sites such as construction projects.[xi]
Additionally, contractors could utilize personnel positioning data overlaid onto schematics and project drawings to establish productivity baselines that can be useful enterprise knowledge when bidding future work. Contractors could also utilize these same tools to generate project-specific measured-mile baselines, and then to powerfully visualize disruptions and lost productivity arising from crowding, stacked trades, or out-of-sequence work resulting from owner or architect decisions or directives.
Attorneys and claim consultants are already liaising with contractors to deploy these AI technologies, to obtain and synthesize the data needed to better evaluate project opportunities, or to better substantiate change requests or claims. At one time or another, most contractors have been involved in a troubled project, where the as-planned schedule was disrupted, but then never updated, even as issues proliferated and delays compounded. Imagine being able to provide a forensic scheduling consultant with time-stamped site-correlated raw data with which to build a precise as-built schedule to demonstrate the impacts, and not merely with spotty and easily disputable daily reports and possibly slanted percent-complete estimates.
- Site Safety
Safety is a paramount concern on construction sites. Both the ConsensusDOCS 200 and AIA A201 make contractors responsible for implementing and enforcing safety plans and procedures.[xii] This includes the deployment of general and site-specific safety programs, with appropriate monitoring, meetings, and outreach, and penalties for noncompliance. Complex projects make for complex construction sites, so it can require an incredible amount of time and resources to thoroughly vet and establish safety protocols and, because life and health are involved, there is always room to do better.
This is why the developers of some construction-specific AI systems highlight their products’ apparent ability to identify and flag means and methods-related risks, and correlate them to project drawings and planned or actual schedules. Similar to uploading plans and “training” the AI tool to identify scope gaps, such tools can also be calibrated to develop worksite hazard analyses, identifying when and where contractors will need to implement worksite hazard mitigation efforts, such as fall protection or hot work procedures, for example. These insights can bring increased precision and granularity to site-specific safety plans, and ensure that safety meetings and outreach are specifically tailored and timely to the work as it is in progress.[xiii] In this area perhaps more than any other, AI has the potential to advance the standards and practices that inform and comprise reasonable safety programs.
Conclusion
The foregoing provides just a few of the many construction-specific potential uses for AI programs.[xiv] There are many other potential applications. For instance, AI insights could prove valuable for preconstruction teams in generating highly detailed constructability analyses and predictive analyses of procurement streams, milestones, and costs. AI-enabled precision tracking could be of particular benefit when performing cost-plus work. And the use of AI could also be marketed when trying to stand apart from the field when pursuing best-value awards. In fact, some industry observers expect that the GSA or other government authorities will be the first to require use of AI capabilities on pilot public projects.
Another caveat: AI comes with its own risks. AI outputs must always be verified and correlated to real-world conditions by the personnel responsible for developing and leveraging AI-generated insights. The buck stops with a person, not an algorithm. We have studied the risks posed by AI, and strategies for allocating and managing those risks, but those topics are beyond the scope of this Paper.
With that, it is worth briefly contemplating the literary quality of the T.J. Hooper case. It is a story of captains navigating heavy seas and emerging technologies, while storm clouds and evolving standards gather on the horizon. While that may sound daunting, perhaps the most consequential aspect of the story is the fact that there were then—as there have always been—early adopters curious to determine whether and how new technologies can be utilized to improve efficiency, results, and safety.[xv]
This is how standards evolve, and why trades and professions are dynamic, rather than static. Those who thoughtfully engage with emerging techniques or technologies are those who set the tone and guide the development and evolution of industry standards of care. As the currently emerging technology, some predict that AI will transform the construction industry more than any other technology has in the last 100 years.
Time will tell if this forecast is correct. In the meanwhile, contact the attorneys at Fabyanske, Westra, Hart & Thomson if you would like to explore the ways in which AI could enhance your bidding, preconstruction, contracting and project management, safety, or claims management practices. In addition, when disputes arise, the use of AI on construction sites will present novel issues in the areas of information exchange and discovery, and the utilization of consultants. The attorneys at Fabyanske, Westra, Hart & Thomson can help you navigate this emerging landscape, to best present your positions in mediation, arbitration, or litigation.
Announcements
Fabyanske, Westra, Hart & Thomson, P.A. has been ranked as a Band 1 Construction Law Firm in Minnesota by the well-recognized Chambers professional rating service.
Dean Thomson (Band 1 Construction Law)
Kyle Hart (Band 1 Construction Law)
Mark Becker (Band 1 Construction Law)
Jesse Orman (Band 3 Construction Law)
Rory Duggan (Band 3 Real Estate Law)
Julia Douglass (Up and Coming)
Here’s what Chambers has to say about FWHT: “The team is esteemed for its dedicated construction group, which brings deep industry insight to disputes, contract negotiations, planning issues and arbitration matters. The group represents diverse participants from the public and private sectors, including design professionals, insurers and subcontractors. The firm frequently acts on large-scale regional and national projects. It has attorneys who are well known in the sector, having been appointed to leadership roles at a variety of industry bodies.” “This firm’s results are always exceptional, so there is great value in their services.” “Their team of attorneys know construction and construction law inside and out.” For more information click here.
Fabyanske, Westra, Hart & Thomson, P.A. is pleased to announce that Mpls. St. Paul Magazine has named Julia Douglass as as one of the “Top Women Attorneys in Minnesota” in its April 2024 issue. For more information click here.
Mark Becker is recognized by Minnesota Lawyer as a Power 30 attorney in the field of Real Estate and Construction Law. Congratulations to Mark at Fabyanske, Westra, Hart and Thomson for this prestigious recognition. For more information click here.
[i] The T.J. Hooper, 60 F.2d 737, 737 (2d Cir. 1932).
[ii] Id. at 739.
[iii] Id. at 740.
[iv] Quoted in Lawrence, Robyn Griggs, “AI is coming for construction, experts say,” Construction Dive, January 4, 2023 (available at: https://www.constructiondive.com/news/ai-is-coming-for-construction-experts-say/639570/).
[v] 1 Bruner & O’Connor, Construction Law, § 1:2.
[vi] ConsensusDOCS 200 (2017), §§ 2.4.28 and 3.3.1 (emphasis added).
[vii] See AIA A201-2017, §§ 1.2.1 and 4.2.12.
[viii] See, e.g., June 17, 2024 Engineering News Record, at p. 17.
[ix] See https://www.openspace.ai/products/capture/.
[x] Screen captures from https://www.openspace.ai/products/capture/ and https://www.openspace.ai/products/bimplus/.
[xi] See https://bostondynamics.com/industry/construction/.
[xii] ConsensusDOCS 200 (2017), §§ 3.4.3 and 3.11; AIA A201-2017, §§ 3.3.1, 4.2.2, 10.1.
[xiii] See https://www.openspace.ai/resources/use-cases/safety/.
[xiv] This Briefing Paper is certainly not an exhaustive account of currently available AI products or their capabilities. For that, the interested reader should consult representatives from the companies that are developing and deploying AI tools.
[xv] In fact, while it was not yet an industry standard to do so, several captains navigating in the area had indeed equipped their vessels with radio receiver sets, had received broadcasts of the forecast, and put in at the Delaware Breakwater to wait out the storm. The T.J. Hooper, 60 F.2d at 739.