“The Enemy That Kills You, Doesn’t Look Like You”
Read time: 5 minutes.
This morning was the ICE Strategy Session: Covid-19, Artificial Intelligence, and the future of the Civil Engineer. During the session, chaired by Rachel Skinner, Prof. Richard Susskind spoke about how AI and ‘the future’ might affect industries. His thoughts then provoked discussion amongst a board of construction industry experts.
Richard’s presentation focused on several key themes which I’ll summarise below:
The Mind Set
Richard explained that if, as an industry, we don’t want to be left behind, we need to change the way we approach problem solving. We shouldn’t be thinking about how we can do what we currently do faster, better or cheaper but change what we do all together to achieve the end goal.
An example Richard gave was the decline of the rail industry in the US. He suggested that their decline can be attributed to the rail industry’s inability to comprehend that they weren’t in the ‘rail business’, they were in the ‘transportation business’, and the customer didn’t care how they got there. Another example was given where the MD of Black & Decker explained to new employees that they weren’t in the ‘selling drills’ business, but in the hole making business, and if they find a better way to help customers make holes in walls that doesn’t involve a drill, then they need to adapt, quickly. The focus should be outcome above all else.
The Disrupter
The discussion then moved on to, “well ok, so who is it going to be that shakes things up”. After the CEO’s of several international construction firms (Balfour, WSP and others) discussed how we might adapt as civil engineers, Richard suggested that “the enemy that kills you, doesn’t look like you”. He went on to explain that whilst we might be able to disrupt and innovate internally, this is typically a very hard thing to do and usually it’s external people/companies who really disrupt industries.
So who might these disrupters be? Richard suggested that with the development of AI, data is king, and so it might be data analysts and scientists, not civil engineers, who create this disruption and lead innovation in construction. Mark Naysmith, CEO of WSP, supported this theory when he explained that as a consultancy, they’re employing more and more graduates with computer science degrees and even creative degrees such as art and music. Mark didn’t say if this meant he was consequently employing less engineers, I expect not, but perhaps this is us conceding as an industry that we, with all of our structural theory, might not be all of the solution to the problem. Those that don’t embrace this might be left behind.
Something that wasn’t mentioned in the strategy session, but I think reflects the sentiment, is what Dominic Cummings was trying to do in No. 10. We are familiar with the news that rather than employing civil servants and staff with politics degrees, he was employing data analysts and computer scientists to help run the government. Now, the success of his strategy would be a contentious debate, but to me the parallels are stark as we see an industry (politics) realising that the disrupter (the innovator) might not look like a politician.
My Opinion
It’s my opinion that the civil engineer, and what we design and build, will remain the solution to the outcomes required by society for at least the next 100 years. Whether that being transport infrastructure to enable trade and movement, energy infrastructure to power our homes or the high rise buildings to home us where space is a commodity. However, the longer we wait as an industry to ‘self-disrupt’ and innovate, the less control we’ll have over an industry that we regards as ‘ours’. This has already been witnessed with the modernisation of project management. The Bragg and many other reports of the 80s/90s that reviewed the technical and commercial practices of the civil engineer, saw the engineer’s remit transfer to project managers, quantity surveyors and schedulers. It’s my opinion that the longer we wait to self-disrupt, the higher the chances of what remit remains (such as risk management, construction sequencing, design and quality assurance) could become the remit of yet another specialist, perhaps data analysts and computer scientists.
Further Reading:
The session was recorded and should be available on the ICE’s website in a couple of days.
The board members were:
Prof. Richard Susskind OBE: A British author, speaker, and independent adviser to international professional firms and national governments, specifically on the use of AI.
Mark Neysmith: CEO of WSP UK and a member of the Global Leadership Team.
Simon Adam: Head of commercial for Crossrail 2.
Stephen Tarr: Managing Director of Balfour Beatty.
Suzannah Nichol: CEO of Build UK.
Chris Young: CEO of Tony Gee.
Holdfast Training Services Trial Blog 
Jordan,
Fascinating stuff and and a refreshing level of self-awareness from the ICE. Data science, AI and machine learning have proliferated over the past decade, moving out of academic circles and into the indsutrial mainstream and I sincerely believe that it will completely revolutionise working life across all industries within our working lifetimes.
I don’t think that our days are numbers as engineering/construction purists though. Despite the radical changes to the “how” we do things, the “what we do” and “why we do it” will utlimately remain unchanged in my opinion. Data scientists are capable of producing unprecedented levels of insightful information, but they still require a direction of travel – that’s where I see us coming in. When you break it down to it’s core – the end goal will always be 2 things:
– making decisions
– managing risks
So we definitely do not want to be resistant to change – it is imperative that we buy in, build an appreciation of emerging technology and methodologies and figure out how it can be manipulated in a way that can add value to the decisions we make and the risks we manage.
I’d personally like to see the PET course (both disciplines) evolve in the future to place more emphasis on data science and machine learning – happy to discuss further if any of the PEW staff are reading this!
Scott, you’re absolutely right.
Interestingly, the discussion moved from AI and machine learning to automation which seems more likely to impact us in the nearer future. For example, we’re already using basic automation for hard-core compaction and boring machines. I suppose the turning point will be when we can trust machines to do certain jobs more safely and more effectively. It’s all about sentiment, by which I mean that whilst the technology is nearly there, driverless cars won’t take off until the sentiment changes and we trust the machine to get it right… more than we get it wrong.
So with that, the natural question is, how many of those kind of jobs like compacting (which has put a plant operator out of work) or boring machine automatic alignment (which has put whole teams of surveyors out of work) are there?
On your final point, John Moran and I have spoken to no end about the requirement for the wing to ‘keep up with times’. Digital applications like Revit aren’t just “nice to haves once the principles have been mastered”, in my opinion it’s becoming more of a case that even with all the fundamentals, without the mastery of these complex computer applications, the rest of the world will run away without us!
By the way, the civil wing has for the last 2 years now invested in teaching the use of Revit and other Autodesk applications, apparently the standard of end of module projects are improving as a result. So the urgency is certainly understood – I think!
Possibly some good news on the software front. As part of the budgetary cycle, 170 has screened (asked) for money to transition some of the existing Bde’s AutoDesk software licences to REVIT. This is not a guarantee of funding because it is essentially a £100k increase on what was agreed last year and the MOD is under considerable financial pressures despite the budget uplifts announced in the press.
As ever, there are wider issues to consider (XWETs) including; training, the hardware spec of the existing technical standalone laptops isn’t suitable for REVIT and cloud data services are problematic when you can’t connect to the internet (expeditionary operations).
TEDDs will remain the PQE(C) software toolkit because it has the ability to factor calculations based on observed damage (used during Kosovo I think for technical bridge capacity assessments) I’m not sure if Autodesk products have this feature. Like many software packages it can convert the output into different national codes and languages which is handy when working with coalition partners.
Scott, good comments and I agree.
I found that the supervisors I have worked with have years of experience behind them and could safely and accurately recreate anything that they had done before. This is their expertise. My job as the site engineer was in spotting what was new for the Supervisor and his team, understanding and communicating it and then working through a way forward that was safe.
Making decisions and managing risk is absolutely the core of our job on site.
The good relationship that site engineer has with their supervisors will hopefully break down the ego barrier that apparently comes with the job description. A supervisor that will acknowledge the limit of his experience readily is safer and thus better to work with.
Jordan, Scott, further to your points and to illustrate some of the benefits and issues
The number of sets of hands at the frontline in Heathrow uniform is actually quite scarce and so anybody with a toolbox in hand is more often than not in a contractor uniform. As a result, Heathrow engineering is always trying to employ the use of error/anomaly identification to save time and effort and prevent the need for so much ‘routine’ maintenance and turn to more targeted interventions. This is happening in two main areas, mains water provision and electrical power, the latter which i have been involved in.
In short:
– Data science company handed 4 years worth of historical electrical energy use down at switchboard-level. For info, each switchboard can have about 100 bits of kit attached to it all told, everything from an XRay machine to a single 3-pin plug with a kettle plugged in.
-Data science company crunches the data and spits out a model which (according to them) has the significant factors listed for electrical energy use (think passenger numbers, weather, time of year etc etc).
-This model is then ‘run’ and the actual energy use going forward is compared against their predictions.
-When there is a deviation, a red flag appears (essentially) and we know that switchboard 123 is using more energy than it ‘should’ be.
NOW, the so what. The data science team live and work in Madrid, there is a national lockdown ongoing and very little cash to send contractors around with a Fluke to start fault finding.
Point being, there is still a chasm between identifying what isn’t ‘normal’ and producing executable outcomes (‘have a look at this AHU’ or ‘that XRay machine is probably about to melt’ etc.) And this is where I have found myself a lot of the time playing the middle man, with a rough knowledge of both parts of the equation (or could be argued not enough of either).
I will be interested to see how long it is until Heathrow and industries in a similar place employ in-house ability to run maintenance and diagnostics on a data-driven system-wide basis rather than simply a library of maintenance routines specific to each asset.