If you’re an engineer, it is not AI that might steal your job. It will be an engineer who knows how to use AI to do the same work faster, cheaper or more effectively.
About the authors
Andrew Hannell is the founder of Digital Constructors and is currently working with the Australian Institute of Machine Learning at Adelaide University on a project applying computer vision to automate inspection of linear assets such as rail or road.
Professor Anton van den Hengel is the Director of Applied Science at Amazon and Director of the Centre for Augmented Reasoning at the University of Adelaide.
The recent release of ChatGPT has generated significant debate on the impact AI will have on education, business and society. The challenge is to embrace these technologies and to adapt our teaching, business or engineering methods accordingly.
Education will require a two-pronged approach. As well as specialists in AI, software engineering and machine learning, all general engineering students and practitioners will need to understand the principles of AI and how it can be applied within their field.
For example, although a civil engineer will not need to be an AI expert, they will need to understand the fundamentals of AI and how it can be applied to their engineering challenges.
Take, for instance, the characteristics of complex construction projects such as rail and tunnel infrastructure:
- Huge costs, often taxpayer money
- Take years or decades to deliver
- Significant impact on community and businesses
- Potential for accidents
- Long-term operational costs
These projects are prime candidates for technology-led innovation and AI offers vast opportunities. For a project that costs billions of dollars or takes decades to construct, even a minor design or delivery improvement will result in significant benefits.
Where to watch for AI
The opportunities do not just lie in these mega-projects and go far beyond brick-laying robots and the like. The current construction environment is particularly challenging for projects of all types and scales and there are potential benefits of AI across the board.
Artificial intelligence can help improve site safety by identifying specific activities, plant and equipment. These computer-vision models can be trained to detect any visible object, so in addition to safety, can be applied to quality, project controls or environmental purposes.
Repetitive tasks such as inspecting, counting or tracking objects to identify defects or monitor construction progress can readily be automated using artificial intelligence solutions.
The highest-value opportunities could be in optimising design and delivery engineering solutions by recognising patterns in huge and complex datasets that is otherwise impossible.
Many traditional engineering challenges are well suited to machine-learning regression techniques, where massive datasets of variables can be processed to identify correlations or optimise outcomes.
These examples include:
- a machine-learning model trained on construction programs would ‘learn’ patterns of activities, such as sequences, predecessors and successors to identify problems or optimisations, or even re-sequence an entire program
- a model trained on empirical concrete mix data would predict and optimise characteristics such as early compressive strength or chloride resistance.
One of the authors, Andrew Hannell, has helped develop a prototype machine-learning model for the Steamranger Cockle Railway in South Australia.
The model detects potential risks such as objects on and around the track such as vegetation or debris. This data is georeferenced and timestamped for analysis and identification of emerging patterns and changes over time.
Although some highly publicised uses of AI such as autonomous vacuum cleaners might appear as novelties, many of the same techniques can be applied to design and engineering problems.
For example, if a generative AI model trained on art can produce novel artworks, then a model trained on structural or electrical systems could design new structural or electrical systems. Where a language model such as ChatGPT can be used to answer general questions, a model trained specifically on engineering data would be able to answer very specific engineering questions and solve detailed challenges.
Rethinking an engineer’s role
As AI becomes more established, a major shift we see coming in the way engineers work is this: engineers will not design the thing, they will design the algorithm that designs the optimal thing.
Getting started in artificial intelligence might seem like a daunting task for a typical construction contractor or engineering consultancy.
However, many of the same analytical, mathematical or statistical skills that many engineers use on a daily basis can be applied to these problems. For those engineers working in a design field, techniques such as computational or algorithmic design have some parallels to machine learning.
To help businesses get started in artificial intelligence, the CSIRO has produced a directory of Australian organisations with capabilities to enable adoption.
By capitalising on our world-class university expertise in this field coupled with sufficient government and private sector support, Australia could become a leader to match countries such as Israel in artificial-intelligence capabilities.
One of the authors, Professor Anton van den Hengel, is active in lobbying for a national AI strategy and government investment in this critical field.
The development of sovereign AI is of critical importance to Australia’s future security and prosperity. Australia can be a world leader in AI by 2030, but strategy and investment is needed.
The time to consider how artificial intelligence will affect all aspects of our lives, including construction and engineering, is not in the distant or near future – it is now.
By tackling the challenge head on rather than following others, we can have a profound effect on our individual and national futures.