Out-Law Analysis 3 min. read
02 Nov 2023, 2:30 am
Artificial intelligence (AI) offers the opportunity to fundamentally change the way the built environment is created.
The implementation of AI solutions is not only a competitive advantage issue. It is also an important part of the construction industry’s response to other fundamental challenges that are re-shaping the way in which projects are designed, built and operated, including decarbonisation and net-zero demand.
There are a number of emerging and potential use cases for AI in the design and construction phases of the project asset lifecycle. The construction industry needs to seize the opportunity.
Generative design, powered by AI, provides a range of suitable solutions to choose from based on defined engineering challenges, which can then be refined by the user to suit their needs.
Instead of trying to keep constraints and parameters in mind during the design process to arrive at a solution, designers can tell the software what the requirements and constraints are – for example around strength, agility, cost and performance – and generative design then determines possible combinations based on those requirements.
With AI's advanced algorithms and machine learning capabilities, designers and architects can unlock innovative solutions, optimise energy efficiency, enhance structural integrity, and streamline the design process.
Generative design has proven benefits. Software company Autodesk reports that using generative design can result in a 30% reduction in material costs and a 40% decrease in construction time. Arup estimates that AI-assisted structural analysis can reduce design time by 50% while maintaining or improving structural performance; and the US Green Building Council estimates that AI-driven sustainable design practices can reduce energy consumption in buildings by up to 30% and lower carbon emissions.
Building information modelling (BIM) is a set of 3D design and modelling software tools already familiar to the construction industry, which help visualise construction designs from many angles. Enhancing BIM with AI capabilities can unlock additional benefits including allowing for quantity take-offs – the extraction of information around required materials.
BIM based quantity take-offs offer more accurate results. They also enable contractors to improve efficiency, lower costs and improve the overall quality of design, construction and operation.
However, while the advantages of digital quantity surveying based on a digital model of a project are obvious in terms of accuracy and time savings, some of the challenges around the transition to digital begin to emerge. For example, the people issue: how does the industry attract the talent that it needs, and how does the industry incentivise and reward that talent? The required skills go beyond ‘pure’ digital: if the BIM model is incorrect, but the people involved lack the basic quantity surveying skills to recognise the error, who is ultimately responsible for the accuracy of the BIM model?
With such enormous potential, why does construction remain one of the least digitised industries and slow to adopt new technology? There are generally two reasons.
Most machine learning systems operate on a ‘black box’ basis - they don’t explain the reasons for the conclusions they reach, and the algorithms used to make decisions are often proprietary and not easily understood. This might well lead to construction professionals not being able to understand why and how recommendations are made. Some academic commentators suggest that AI’s potential in construction will be limited unless an “explainable artificial intelligence” approach is taken. This is something of an evolving concept in which the AI model is developed to present its explanation in a way that makes sense to the human observer, with algorithmic transparency and on which simulations can be run.
Investment costs associated with an AI enabled system can be high and construction companies struggle to understand what the benefits are and how they can be realised. That makes investment decisions difficult - in particular, with so many contractors delivering work through multiple subcontractors, reaching a common view can be challenging, which makes it difficult for the industry to coalesce around one “accepted” AI enabled solution. That raises the possibility that an investment could quickly become redundant. Given the profitability challenges experienced across the sector, it’s not surprising that organisations can be slow to invest for the future.
From a legal perspective, the development of AI-enabled critical path programmes which can be found in the marketplace would minimise disputes and thus eliminate the need for expert programming witnesses. However, if the conclusions are not explainable, and the why and how of those conclusions cannot be tested and challenged, why should those whose commercial interests are dependent on that outcome accept it? And if each member of the supply chain is operating different programming algorithms – possibly demonstrating different outcomes – how does that assist? There’s a clear role for governments and policy makers here.
At this stage, there may be the emergence of a two-speed industry, where winners are emerging and the barriers to entering the fast lane are getting higher. There may start to be information asymmetry, where one party has access to AI solutions while the other does not. There may also be consolidation across the supply chain and the alignment of the supply chain around specific segments with AI solutions specific to that segment.