Few technologies are as polarising as artificial intelligence (AI). One side claims it’s the answer to all the world’s woes. Another vocal minority, including prominent engineers and scientists, believe it could doom us all. Like most things, the truth likely lies somewhere in between.
To start with, the term ‘artificial intelligence’ is a misnomer. These systems are not conscious or thinking for themselves, and they’re not self-aware. They’re only able to mimic some of the cognitive capabilities of humans because humans programmed them to behave that way.
That said, can a system that merely mimics cognition do useful work? Absolutely. AI is poised to have a tremendous impact on the way we do things, disrupting every industry, not just construction.
Machines that do one thing well
Hyperbole aside, AI experts predict the technology has numerous applications in construction, including:
- contracts, bids and reporting
- supply chain management
- project planning and scheduling
- site monitoring
- building information modelling (BIM) integration
- access to expert knowledge.
These lead to flow-on benefits such as increased productivity, reduced wastage, faster build times, higher quality and safer workplaces – a shopping list of hot topics the construction industry is trying to improve. AI systems designed to tackle these applications are already beginning to emerge.
Advanced as these systems are, within the field of computer science, they’re typically referred to as ‘weak AI’ – a system that’s very good at solving one class of problem but not very good at anything else.
Experts believe it’s a form of AI that will prove particularly useful in construction. For example, a weak AI might be able to identify workers and visitors not wearing the correct personal protective equipment (PPE) on a building site.
Provided it can reliably spot hard hats and high-vis, it doesn’t matter that it can’t write the quarterly financials. Its ability in that one regard gives it value.
Expert systems the first to appear
Another recently commercialised tool is an expert system for asset managers. It takes a detailed BIM model and associated documentation and provides a plain-language interface that can answer any question you care to ask about the building.
Weak AI excels at providing a usable interface to complex data, in this case solving a common problem – quickly and reliably finding information about a building asset.
A similar expert system in the works will have a complete understanding of the building regulation framework. When fully trained, it’ll be possible for anyone to ask it detailed plain-language questions and immediately get accurate and reliable design guidance.
These are examples of near-future AI where the technology already exists and will likely see wholesale commercialisation in the next year or so. A few years further down the track, vision systems are likely to be the next technology to see widespread uptake in construction.
One local company is already developing computer vision to identify and sort waste materials from construction sites. Overseas, another company uses vision systems to track the location and condition of machinery across large, multi-level construction sites. Another start-up is commercialising a computer vision system to monitor the quality of materials and quickly sort and remove defects.
The applications are almost endless. The tech is fast and accurate, and although it’s still not perfect, experts say it’s rapidly advancing.
Preference for the tried and true However, state-of-the-art technology isn’t necessarily the limiting factor when it comes to adopting AI tools.
Worldwide, construction is one of the least digitised industries. Take Code compliance checks, for example. Not many councils will even accept BIM models in their compliance processes, so the widespread adoption of advanced AI in the next couple of years seems unlikely, to say the least.
When councils can cope with digital models, there’s often a great deal of variability in the quality of data that they receive from architects and engineers. When the data isn’t up to scratch, it’s impossible to conduct accurate checks – no matter how clever the AI.
This lack of digital know-how has come under increasing focus in the last few years as the construction industry pushes for greater productivity and accountability.
Without significant improvements, experts foresee an industry that has access to an array of fast, accurate and skilled AI tools but is unable to use them because the quality of data produced isn’t up to standard.
The price of convenience
There are also numerous ethical and privacy considerations to deal with before the construction industry, and society in general, is ready to fully embrace AI.
Think of the system for monitoring on-site PPE. What is the scope of its capabilities? Can it only check for PPE or can it be used for less altruistic purposes? Could it, for instance, be used to monitor employee productivity or identify and track individuals?
Privacy watchdogs like the Office of the Privacy Commissioner have already issued stern warnings about the potential privacy and security risks surrounding AI. For now, their advice is largely directed towards government agencies.
However, these issues don’t seem insurmountable. It’s already commonplace to have agreements with contractors for the degree of site monitoring that will take place. Adding AI to the mix doesn’t change that. It might expand the scope of what is monitored, but experts believe AI doesn’t necessarily pose a greater risk than the systems that are in place now.
A growing technological divide
For the small construction business, all this talk about AI might seem irrelevant – just another over-hyped fad with little meaning to the bread-and-butter work of swinging a hammer.
While it’s true that AI is unlikely to have a serious impact immediately – unlike other industries – experts warn that businesses would do well not to ignore it.
It’ll begin around the periphery, mostly with small advances in productivity. Planning, scheduling, materials purchasing, invoicing and finance systems are all prime candidates. Many simplistic versions of these tools already exist today. Within a short time, these systems will become fully automated.
More complex work like writing contracts, bids and reports, while not fully automated at first, will become faster, easier, more detailed and more accurate with support from increasingly sophisticated AI tools. Full automation will follow, and the advances will continue at a growing pace.
Those who invest in AI tools will have a significant commercial advantage over those who don’t, likely driving a gap between the top companies and everyone else – an even wider gap than we already see today.
Like most things AI, such dire predictions are usually balanced by equally hopeful crystal gazing. Both sides agree on one thing though – AI means change. If the construction industry treads carefully, it could be the opportunity to transform almost everything we do.
AI jargon
- Artificial intelligence (AI) – a machine that mimics natural cognitive ability by perceiving its environment and acting to achieve given outcomes.
- Weak AI – an AI that focuses on one function.
- Strong AI – an AI capable of several different functions at once.
- Generative AI – an AI that creates output, often text or images, based on a series of human inputs or prompts.
- Machine learning – a field of computer science that enables machines to infer meaning and act based on algorithms and pattern recognition rather than explicit instructions.
- Computer vision – a field of computer science that enables machines to recognise patterns in images and video.
- Expert system or knowledge-based system – a computer system comprised of a series of rules extracted from a human expert and codified for the computer.
- Neural network – a machine architecture comprising a network of artificial neurons used to replicate specific human abilities.
- Natural language processing – a field of computer science that enables people and machines to interact using human languages.