Skip to main content

Article

Marsh UK 2024 Construction AI & Cyber Event

Discover insights on artificial intelligence and cyber security from Marsh’s UK Construction Practice.

Benefits and risks from the construction industry embracing AI

As new technologies continue to transform the construction industry, it is critical to remain mindful of the associated risks that are introduced, according to Kelly Butler, Cyber Practice Leader, Marsh UK.

Embracing this technology can enable organisations to “reshape the construction industry and elevate operational efficiencies”, Butler said at the ‘Construction Artificial Intelligence (AI) & Cyber Event’, hosted by Marsh UK. The session was support by Steve Hurley, Head of Sales for Health Care, Life Sciences & Business Services, Google, who gave valuable insights into the benefits of smart construction.

Building on Marsh UK’s ‘Construction Summit — The Future of Risk’, this event looked to continue conversations on the future of the construction industry — and the particular role generative AI can play in its development.

Generative AI and smart construction

Generative AI can develop emergent capabilities and leverage a wide knowledge base to answer open-ended questions, generate informative content, and solve problems beyond the algorithm’s initial purpose. This can provide a myriad of benefits to the construction industry, including the following:

  • Physical asset reporting: Through physical asset reporting, generative AI has the capacity to accurately identify and report on safety issues at construction sites. The performance of both staff and equipment can be detected and optimised through comparison of good and bad practices. AI systems can also be tailored to the specific laws and guidelines of relevant jurisdictions, while producing suggested improvements at an accelerated pace.
  • Predictive maintenance: AI can analyse historical performance data of various machinery and equipment to alert supervisors to items that may be failing, require recalibration, or need replaced.
  • Building tailored large language models (LLMs): LLMs are comprised of information from multimodal sources that can be prompted to generate content. Organisations within the construction industry can use feedback to “shape use for specific building or safety problems they face”, said Hurley, with each project possibly being siloed with all information in one space for later integration.
  • Strengthening supply chain resilience: Generative AI can be leveraged to improve the supply chains of construction projects through readily establishing and rectifying “bottlenecks, skill shortages, and material costs and availabilities”, said Hurley. Additionally, new technology can also be used to help achieve environmental, social, and governance related goals by efficiently locating and sourcing from more sustainable partners.
  • Data driven decision making: Generative AI can analyse vast volumes of data and records — with a greater degree of accuracy — at a rapid pace. Using data to solve construction issues can improve key tasks, such as material selection, decarbonisation, optimise waste management, and facilitate easier cross-department collaboration.
  • Digitised site: Digitalising site images for upload into generative AI systems can allow construction managers to review progress remotely. The use of drones to monitor site development can help store data that tracks projects easier and quicker.

Risks associated with Generative AI use

Generative AI has the ability to expedite and improve the accuracy of many tasks and processes in the construction industry. However, these developments are not without threats — requiring appropriate mitigation.

  • Data: Presenting large amounts of data can pose risks to organisations leveraging AI technology. The standard of information provided is particularly crucial, as the “quality, diversity, and scale of data can have significant impact on AI performance”, said Butler. Additionally, it is crucial before uploading data that businesses implement strong data governance policies and look to actively mitigate threats surrounding data poisoning, privacy, security, and IP theft.
  • Hallucination/confabulation: ‘Hallucinations’ are a pertinent issue stemming from poor data and involve generative AI producing nonsensical or erroneous outputs. Typically, hallucinations occur when AI attempts to “infer patterns that do not exist or encounters scenarios outside of its training data”, Butler stated.
  • Threats from bad actors: The rapid digitalisation in the construction industry may have occurred at a rate that has outstripped security improvements. Cyberattacks have the potential to disrupt operations across the board and region and could lead to “loss of profits, clients, and reputation”, Butler said. Additionally, the development of cybercrime-as-a-service also enables the possibility of nontechnical people committing cyberattacks.
  • Human error: As “generative AI remains prone to human error”, according to Butler, businesses using new forms of technology will require new controls that help differentiate AI and human outputs. New threats surrounding the use of AI may surface as its adoption becomes more widespread, making the governance of AI crucial from a risk perspective.
  • Changing regulations in AI: The evolving regulatory landscape has the potential to create new risks for construction organisations. It is important to remain agile in this area and understand regulations are subject to change as “AI laws are nascent”, said Butler.