Computer vs. Architect: Death Match
I like to think it’ll happen like this:
Joe the architect rolls into the office at 10:15am on Monday morning, his regular time, eager for the news. Joe is the Director of Fancy Studio, global architecture and design firm, and the winner of this year’s design awards is being announced this morning. Fancy Studio is once again up for the coveted Australian Institute of Architects Annual Award for Best Project, and Joe is smiling confidently to himself while sipping his french press and waiting for his computer to boot. Sure enough, there’s an email waiting from the AIA. Excitedly, Joe clicks and opens.
‘We are pleased to announce,’ the email opens, ‘that the winner of the 2020 AIA award for Best Project goes to AI Architecture’. Joe stares at his screen in disbelief, and irritatedly yanks at his turtleneck.
‘Steve,’ he says to one of his associates sitting nearby, ‘who the hell is AI Architecture?’
Somewhere across town, Linda, the Director of AC/BC Quantity Surveyors, is putting down her phone in disbelief.
‘Michael,’ she says to one of her associates sitting nearby, ‘we lost the Department tender.’
‘Wow,’ Michael responds. ‘I thought we had it in the bag. We’ve been working there for 10 years! Who did we lose out to?’
‘Someone called AI Quantity Surveying. Apparently their fee was half ours,’ she said, looking glumly out over the sea of workstations in her office.
Across the CBD the same story was playing out: DSTY Engineering lost a tender to AI Engineering, the local Council made a decision to use AI Building Surveying for all its projects.
In a small office on the city fringe, another professional services firm was being disrupted, however this one in a different way. Instead of uncomfortable turtlenecks or calls from global HQ for an unscheduled strategy meeting, the 10 person team at AI Construction Services was being disrupted by another drinks session celebrating yet another project win over the big, established firms. And while they cracked beers and poured bubbly, in another room their small data centre full of computers hummed away, working on winning that next award.
Ok, maybe it won’t happen exactly like that, but AI — Artificial Intelligence — is coming, coming for our jobs. There are lots of jobs here in the property and construction world, upwards of 10% of all jobs in Australia in most years, and some are going to be up for grabs.
Sure, you might be thinking, people have been saying for years that we’ll all soon be replaced by computers and robots, we’re all still here and my company can’t hire people fast enough. And you’d be mostly right. In his excellent article What’s Next in Computing?, Chris Dixon wrote that ‘AI has a long history of hype and disappointment. Alan Turing himself predicted that machines would be able to successfully imitate humans by the year 2000.’
So while Turing and others have been forecasting big changes and the demise of work as we know it for decades, changes which haven’t yet come to pass, developments we have witnessed in our lifetimes — the rise of computers as everyday tools, digitsed information and the internet, and mobile phones — have all played their part in bringing us to now, which may be the beginning of a ‘golden age’ in AI… and ultimately to the point where it makes some of us not as useful as we once were.
To be clear, we’re not talking about robots. While bricklayers may be preparing their torches, pitchforks and water guns to chase away bricklaying robots when they turn up on the jobsite, robots and AI are inherently different. AI is where a computer is programmed to use some input information to generate a output. One recognisable example is speech recognition. You ask Siri for directions to the nearest petrol station (the input), and (most of the time) she spits it out for you. For Siri to be able to do this requires the computer to have a huge amount of data for both the speech, and the output directions. Considering the potential options for speech instructions and then the outputs — think directions, sports scores, math… whatever it is that a morally square and sober person might ask Siri — and huge means huge. One reason for the current advances in AI is the widespread availability of large data sets which are part of what make it all possible.
AI also incorporates machine learning. It’s a boring, misused, and oft misunderstood thing, machine learning, but its advances are another part of what’s driving cool stuff that’s happening with computers. Think of it basically as a computer tracking human inputs, adding them to its huge data set, which then makes the computer’s outputs better over time.
A good example is is Netflix and its recommendation engine. The computer is always watching what movies you select, recording this data, and then adjusting future recommendations accordingly. Spotify works the same way with music. Coupling improvements in machine learning, availability of data sets, and also the decreasing price of computing power, and we’re getting closer to having AI that starts stealing work from our architects, engineers, even project managers like myself who don’t actually produce much of anything.
So who really needs to be keeping an eye on whether or not the boss is setting up a data centre in the spare office? Here’s a short list, starting with those who can breath easy for the moment, and moving down to the levels where one might want to start spending some time updating their LinkedIn profile.
Architects. Architecture is driven by inspiration, by the creative muse. It strives to create a sense of place, and responds to the environment and the context. I’ll pause for you to gag, but there’s an element of truth to the bullshit, and creativity is known to be one area where AI is particularly challenged — it doesn’t fit the straightforward input->output model very well. But the good people at Google are working on systems tailored to creating music and art. And even if it’s currently garbage and the art looks like something you made with the program Paintbrush back in the 80’s, what’s important is that They are working on it.
That said, there are many current examples of AI and algorithms being used to take past experience, say your preference in music or books, and making suggestions for new music or books based on previous information. So if you had a system backed up by enough architectural concept design data, it’s not hard to envision it allowing you to provide some input based on what you are wanting to build, say a two story house, an example of style of architecture you are looking for and other parameters, and the computer would spit out some concept design options based on a million similar examples.
This is probably not feasible for more complex projects where lots of user interaction is necessary, but such a system could revolutionise the mass market long tail of architecture, and provide semi-bespoke designs to the masses. This is probably a fair way off, though, so architects can continue sleeping in for the moment.
Construction. As I noted above, manual, repetitive labor like bricklaying is in trouble, but that’s from the robots. Looking specifically at AI, the biggest impacts will be in the boring old construction office with software system improvements. One company called Construction AI is has developed a system which can automatically interpret civil design drawings, contours and levels, which is often a quite laborious task fraught with inaccuracy.
Perhaps the biggest impact AI will have on the jobsite is with procurement, and the humble yet important truck drivers who are responsible for the many deliveries coming to jobsites each day. There are a lot of drivers out there, and they are in serious and short term (5–10 years) trouble from self-driving trucks. On the whole, though, the main labour force is pretty safe for the near future and can go back to their RDO.
Project Managers. One of the strengths of AI is to complement human activities and improve productivity, particularly where the work is repetitive and runs through a fixed process. Project managers just so happen to churn out some fairly repetitive work, such as meeting minutes and the evil Project Control Group (PCG) reports. AI may be able to help, and while this will lead to some rationalisation of roles that will come with productivity increases, I am all for it.
Advances are continuing in speech recognition, and there’s a product currently on the market called Wordsmith that turns data into written text — well written text — so much so that you’d probably not realise that it was done by a computer. You’ve probably even come across some recently and not even realised it. Wordsmith has been working for the Associated Press on financial news releases, with Yahoo! on sports blurbs, and more.
At some point, you may be able to keep a spreadsheet full of project data, feed it into Wordsmith once a month, and it will spit out your PCG report. So instead of writing pages of tedious reports that no one reads, project managers may actually have time to manage something. Bring on the future!
Engineers. Engineers are facing an uphill battle in the long run. This is a profession that’s been in the midst of disruption for a long time now. First computers started making some of the heavy math and calculations required of engineers unnecessary. Computing advances, digitisation, and the internet then allowed some engineering work to be offshored, where a project in Sydney would be serviced by people in Sydney, but all the number crunching and design work would be completed in places like India where wages are a fraction of those in Australia. The good news is that the offshored jobs may now be coming back home. The bad is that those jobs will now be done by computers.
The bulk of engineering work in construction involves taking the architect’s design and designing a structure or electrical design or hydraulics to fit within that design. Input archtectural parameters around space, layout, dimensions plus some engineering parameters— >output engineering that fits those parameters, backed up by calculations (something the computer already does). This is not far fetched.
Quantity Surveyors. To be blunt, Quantity Surveyors, the guys who count things and produce cost estimates, are stuffed. QS’s generally do three things. They measure, meaning they take a design and tell you that the building is 500 square meters in area, there are 2 doors, and 3 toilets. The latest type of computer aided design called Building Information Modelling, BIM for short, is already on the cusp of making meauring irrelevant. Essentially, parts of a building are added to a design as categorised elements. Thus a toilet is a toilet, and not just a circle on a page. Because of this, when a design is done properly your computer can take the completed design and automatically tell you how many toilets there are.
QS’s also produce cost estimates, which lists out in detail all of the parts and pieces of a building, the cost of each piece, which then sums to a total cost for the whole building. Similar with measuring, the production of cost estimates is already heavily supplemented by computer estimating systems. In the near future, backed by enough data with granular detail, and QS firms love their historical cost data, the computer will take the quantities it just developed, put costs to each one, and tally up a total, all while taking into account factors inherent in the data like project location, time of year, and market conditions.
Finally, QS’s produce cost reports. Cost reporting is essentially bookeeping, something that’s currently starting to be automated using AI. Coupled with Wordsmith for a nice write up, and hopefully sometime soon I can gain back that couple of hours I spend each month reminding the project QS to turn in their reports. There are some black box issues with this future QS-less state, because no one generally likes or trusts a computer output where it is not 100% clear where all the numbers come from, but as the accuracy is proven, and if the information could be backed up by calculations, the trust level will grow. Eventually, good AI and increased availability of data will allow builders, developers, and others to do QS work in house simply, accurately and efficiently, making QS’s obsolete.
A recent survey found that while a majority of Americans think that in the not too distant future robots and computers will do the most of the work of humans, a majority also don’t think it will be their job to be the one that goes. Thus it’s easy to see why the good people in the building industry would be quick to dismiss my musings.
It’s also true that for the near future AI systems will not be replacing entire roles or processes. They more so excel at complementing, enhancing and improving on existing activities. That being the case, productivity improvements and marginalisation in some tasks mean that professional services firms will be able to do more with less, so some of us will be asked to find the door.
Crack a beer while you can, though, because AI Construction Services won’t be grabbing awards and winning tenders any time soon. But AI is coming, in fact some of it is already here. Considering change in the building industry is like steering the proverbial battleship, if you haven’t started thinking about and planning for how AI is going to effect your role or your business, maybe you should.
More Reading:
- 5 Things AI Can do Better Than Humans — The Huffington Post
- A Strategist’s Guide to Artificial Intelligence — Strategy+Business
- What Artificial Intelligence Can and Can’t Do Right Now — Harvard Business Review
- What’s Next In Computing? — by Chris Dixon
More Listening:
- How Safe Is Your Job? — Freakonomics Radio
- When Luddites Attack — Planet Money episode 621 (and also episodes 622, 623, and 624)
