Measurement is the easy part

Every few months someone asks me if AI is going to replace quantity surveyors. It is a fair question, and I get why people ask it of me in particular. I spend my days writing software and the rest of my time in construction research. But the question is too big to answer as asked. "Quantity surveying" is not one task. It is a bundle of very different tasks that happen to share a job title.

Two kinds of work

Roughly, the work splits into measuring and judging. Measuring is taking off quantities from drawings, building up rates, checking the arithmetic in a bill, comparing tender figures line by line. It follows written rules, it repeats, and two competent surveyors should land on the same answer. That is exactly the kind of work software is good at, and takeoff tools working straight from drawings and BIM models already do a lot of it.

Judging is different. Valuing a variation when the contract is unclear. Advising a client on which procurement route suits their risk appetite. Settling a final account with a contractor who sees the job differently. Reading what the client actually cares about when the brief says one thing and the budget says another. None of that has a single right answer you can check against a rulebook.

The tools are getting good at counting. The profession gets paid for deciding what the count means.

What we set out to find

That split is what our latest paper is built around. With Femi Olorundare and Rotimi Samuel Ayodele, I looked at quantity surveying functions in Abuja and asked practitioners which of them they see as exposed to AI and which they see as holding up. The full results are in the article, published in the Global Journal of Engineering and Technology Research. I would rather you read the findings there than have me compress them into a headline.

The line moves

Here is the part I keep thinking about as an engineer. The boundary between measuring and judging is not fixed. Anything that gets written down precisely enough eventually becomes a rule, and anything that becomes a rule can be automated. Ten years ago, interpreting a 2D drawing set was skilled work. Now a model can read one well enough to start a takeoff. So the safe position is not to find a task that is "judgement" today and hide in it.

What I would tell a young surveyor, and what I tell myself:

• Get properly good at the parts of the job that involve people, contracts and risk. That is where the value is moving.
• Learn enough about the tools to check them. An estimate you cannot audit is a liability with a nice interface.
• Keep your measurement fundamentals anyway. You cannot judge a number you do not understand how to produce.

Measurement is the easy part. Not because it is trivial, but because it is the part we already know how to hand over. The harder and more interesting work is everything that comes after the count.