With AI dominating headlines and reshaping industries, many are wondering: what does this mean for construction?

Will AI transform the way we cost, design, and deliver buildings – or are we just caught up in another hype cycle?

To get a grounded perspective, I sat down with long-time colleague Nick Clements, Director and founder at YourQS and a self-proclaimed “sceptical advocate” of AI. In this candid discussion, Nick shares his insights, frustrations, and cautious optimism about AI’s real role in construction – both now and in the future.

(1) Getting It Wrong Before – And Learning From It

David Mitchell: Nick, you’ve said before that you’re a sceptical advocate when it comes to AI. What’s behind that?

Nick Clements: I’ve lived through two big tech revolutions – the internet and BIM – and I misjudged both. I underestimated the internet and overestimated what BIM would achieve. So now, with AI, I’m wary. The potential is clearly massive, but I’m more interested in what it’s actually doing today, not just what it might do tomorrow.

(2) AI vs. Algorithms – The Mislabelled Revolution

David: There’s a lot of “AI” being pushed in software tools. Are you seeing genuine innovation, or just rebranding?

Nick: Mostly rebranding. At YourQS, we have tools that generate detailed costing at the push of a button. But underneath, it’s structured logic – not AI. Lots of “if this, then that” statements. They’re complex and clever, sure, but that’s not the same as machines learning from experience, which is the core definition of AI.

(3) What True AI Looks Like – and Where It’s Limited

David: So what’s your benchmark for real AI?

Nick: True AI learns from data and creates new rules or outputs without being explicitly programmed. Tools like ChatGPT fit this model – they identify patterns across huge datasets and generate their own responses. But they also have serious limitations.

David: Such as?

Nick: First, they’re only as good as the data they learn from – if the data’s biased or poor quality, the output will be too. And second, they lack creativity. AI can’t come up with something it hasn’t seen before. It can remix, but it can’t truly innovate the way people can.

(4) Real-World Use Cases – What Works, What Doesn’t

David: Have you seen any AI use cases in construction that genuinely impressed you?

Nick: Documentation is where AI shines right now. I had a concussion check-up recently, and the clinic’s AI created a brilliant summary of the session. It picked out key topics across the whole conversation and turned them into a coherent report. That’s useful.

David: What about estimating – the core of what you do at YourQS?

Nick: That’s the million-dollar question. People ask, “Can AI cost my design?” We’ve looked into it. For example, we saw tools that could identify a light switch symbol across a drawing – which is helpful – but they still need someone to tell them what that symbol means, where the relevant specs are hidden, and what it all costs. That’s where AI still falls short.

(5) Testing the Tools – and Finding the Limits

David: Have you trialled any AI tools yourselves?

Nick: We tested one from France that converts 2D drawings into 3D models – walls, rooms, all that. It looked promising, but the clean-up took as long as building the model manually. Splitting mixed-use walls, correcting window heights – it’s not there yet. Honestly, ArchiCAD’s had better automation for decades.

(Nick Clements: image via LinkedIn)

(6) The Deep Seek Effect – Why the Hype Matters

David: The Deep Seek AI launch made headlines recently. Does that shift the landscape?

Nick: It might. What spooked the market is that Deep Seek trained a large language model on older, cheaper chips – reducing costs significantly. That raises real questions about the business model for some of the big players. But again, it shows the scale involved – training a large AI model still costs millions. That’s not feasible for every company or use case.

(7) IP, Bias, and Oversight – The Hidden Risks

David: Are there risks that construction professionals should be aware of when using AI tools?

Nick: Absolutely. First, everything you submit to an AI tool becomes part of someone else’s training data – maybe without you realising. Second, we don’t know what biases are baked into the models. And third, there are real questions around IP. If an AI’s response is based on copyrighted data, who owns it?

(8) So, What’s the Verdict?

David: So is AI going to revolutionise construction?

Nick: Not overnight. It’s not going to take over everything – but it will have a big impact. Your job may not disappear, but it will change. We’re already seeing AI as a great productivity aid – summarising meetings, generating captions, identifying elements in images. But it still needs human oversight. That’s not going away any time soon.

(9) Final Thoughts

David: So should we believe your predictions?

Nick: [Laughs] My track record with big tech trends is 100% wrong so far – so maybe take this all with a grain of salt. But we’re not sitting on the sidelines at YourQS. We’re experimenting, testing, and learning. AI will evolve, and so will we.