Maybe. Maybe not.

I was talking with an Australian Builder a few weeks back. He was telling me how he is using AI for his pricing and it cut his time spent down from 40 hours to 20 hours. We average 8.9 hours on residential builds. Our process is not AI, but it is highly automated.

Am I worried AI will replace us? Yes. But it is more because I’m a worrier than what I believe. I was around at the early days of BIM and was one of the many giving presentations about what BIM was going to do and how everything was going to be BIM.

Does the BIM story inform us?

BIM hasn’t panned out that way. There isn’t a technical reason why BIM hasn’t delivered on the expectation. It can do what it promised. There just happen to be many other reasons why the BIM workflows we envisaged haven’t come to fruition. This isn’t to say that BIM hasn’t added value, but it has the potential to add much more.

There is every chance that AI will follow the same path as BIM. Huge potential but never quite delivering on the hope for a bunch of entirely valid reasons.

Will AI help BIM

In fact, you could argue that AI needs BIM so AI may help BIM realise its potential. Like any computer system, GIGO (garbage in, garbage out) applies. BIM by its nature creates structured and good quality data. Feed this into AI and will you get better results.

Augment rather than replace

One of the things that I think people are realising with AI is that is an augmenter not a replacer. It can help us be more efficient, but it can’t do what we do without us. Gartner are predicting that “AI productivity gains as cost savings will be eclipsed by competitors that aggressively reinvest those gains into innovation, modernization and upskilling.” They expect that 30% of the people laid off by AI will be replaced with more expensive people.

That is an American scale corporate view. A more NZ scale construction sector view is that the time saved by using AI on tasks will be replaced by a need to focus on another part of the process. The good thing is with AI taking away some of the low value drudge work it should allow us to focus resources on areas that add more value.

The “try harder” problem with AI

An Auckland professor told a me a trick — when AI gives you an answer, respond with “try harder”. Generally, this will result in an improved answer.

But if AI is so smart, why didn’t it do that in the first place? It is this variability that is an issue for me a as QS.

We achieve our productivity by using algorithms and the power of computers. This means that it is 100% repeatable, unlike AI. Provide the same inputs, you always get the same outputs. Our AI research is not focusing on replacing our people who create the inputs but augmenting them. Making them faster, checking for gaps or errors.