What happens to infrastructure design when AI has a say?

By 19/06/2018Robotics

What happens to design and consultancy in a world in which computers are doing most of our thinking? At the upcoming Australian Engineering Conference, Tim Chapman will fill us in.

Up until recently, computers and digital assistants have simply done what humans have told them to do. Even apparently complex and ‘intelligent’ computer systems have only been able to carry out tasks that they have been pre-programmed for.

For example, online insurance systems have offered immediate quotes. Collaborative robots have helped with tasks on factory floors. Doctors have utilised technology to conduct surgery more safely and with greater precision.

But today, computers are beginning to think and learn and make their own decisions without a human programmer telling them what they should decide, said Tim Chapman, Leader Infrastructure London Group with Arup. It’s a brave, new world, indeed.

“For instance, think of how a doctor might spot moles, and how they might describe the factors that would lead to them thinking that a mole might be cancerous,” Chapman said.

“Perhaps it’s pink around the edges, or rough, or it has a changed shape or begun to bleed. Nowadays you could show a computer 10,000 moles, with a description of whether they went on to become malignant or not. Then you could leave that computer to work out the characteristics of each one by various methods, and trust it to recognise moles that are likely to become cancerous in the future better than the best doctor can.”


Big data


What does all of this have to do with engineering? Plenty, Chapman said. As engineers are people who have been trained to conduct complicated calculations in order to build safe structures, their jobs will likely be enormously altered by artificial intelligence. In fact in some cases they already have been.

“Right now a graduate experiences a series of reasonably mundane tasks in order to pick up their trade,” Chapman said.

“The graduates that we employ do work that is very similar to what I used to do many years ago when I was in their shoes. I didn’t have a computer on my desk when I was at their stage, but actually the tasks are fairly similar and their rate of acquiring knowledge is very similar to what mine was. But because of the automation of various tasks they now sometimes can’t learn the basics of the trade, how things really work. This potentially means they’ll become less capable professionals.

“The really big question is, how does a graduate learn their trade at all in five years’ time, when digital has really kicked in?”

Certain types of engineers will see their roles being automated sooner than others, he said. This particularly applies to roles that are less multi-disciplinary, such as the purer forms of structural engineering, which can be more easily automated or “digitised away”.

Other parts of the profession that are highly multi-disciplinary, such as station design, will be far more difficult to replicate with computers and robots because of the sheer number of processes that must be balanced, managed and optimised.

“Crudely, if you’re already working or being trained, you’re probably okay,” Chapman said.

“If you become a graduate in three or four years’ time, you might not be. And in some fields, such as transport planning, things have already changed hugely. That’s a profession where the prolific nature of Google data enables people to draw much better conclusions than you’ve been able to historically draw about how roads work. Therefore it is already having an enormous influence over how roads are designed, the capacity of networks and where improvements are required, etc.

“If I was a transport planner halfway through my career, having been immersed in the old world, but I actually have 20 more years of the new world ahead of me, I might be very afraid right now unless I was fast learning about the new world and all its magic.”

It’s all very dystopian, but Chapman said what we’re actually seeing is a natural push towards utilising technology to become more efficient and productive, as is occurring in every other industry. Some firms will win and some will lose, but the entire industry will be transformed in ways that people haven’t previously considered.

Society itself will be transformed. It is these all-encompassing industrial and societal changes that Chapman would like to highlight to his audience at AEC 2018, and specifically around how engineering design, construction and consultancy could change.

Of course, there is also positive news. Society will be offered better solutions more cheaply across all industries. Entirely new types of businesses will enter the engineering realm, as they have done in retail (Amazon), accommodation (Airbnb), taxis (Uber) and insurance (Friendsurance). These businesses will offer engineers career opportunities they’d never previously imagined.

“There are a whole lot of new providers of data coming in who are unconnected with the old ways,” Chapman said.

“They might not have the same level of skill in terms of how things are engineered, but they have a huge amount of skill in terms of how data gets managed and applied.”


Slippery slope


Chapman offers an example from geotechnical engineering, his original line of work.

“In order to work out how stable a railway embankment is, I’d dig a whole lot of boreholes along the embankment to find out exactly what it’s made out of,” he said.

“I’d test samples from those bore holes. I’d do complicated analyses to work out the most critical slip circle by which the embankment might fall down, etc. But in the future, somebody might buy data sets that could show, for instance, whether the ground is wet or dry. That will have a correlation to stability.”

“Some of the data might show the slope angle using laser or LIDAR scans of the surface. It could show whether the embankment has moved by three centimetres, or not, over the previous five years. This could be done with data sets and without any engineering knowledge.

“On a Friday afternoon, a job that would have taken me six months and $200,000 to do traditionally, could be started and finished – and possibly even more reliably than it would have been done before!”

In terms of engineering design, consultancy and construction, humans will always be relevant, Chapman said. But their relevance will change enormously. Those who are prepared for such changes, including individuals and organisations, will reap massive rewards.

Tim Chapman is a keynote speaker at the upcoming Australian Engineering Conference, which will focus on AI, robotics and the future of engineering. To register, click here.  

Leave a Reply