Who’s teaching professionals how to think alongside AI? Amazon senior pre-construction manager Dominic Skinner FCIOB CIAT argues that this is not a technology problem: in fact, it is a human opportunity.
Who’s teaching professionals how to think alongside AI? The recent Built Environment Futures Assembly (BEFA) report gives us a framework. The question is whether we’ve built the workforce that can deliver it.
I want to tell you about a time I looked a bit silly. A few months ago, I used an AI tool to write a request for quotation for a feasibility study. It was a lengthy document, the kind of thing I could have written myself if I’d had the time. I was busy, like everybody, and it seemed like a good shortcut.
The tool produced something that looked professional: a clean structure, correct terminology and logical flow. At first glance, it contained all the important information. I sent it out without really diving into every detail.
A consultant I work with came back with a few questions about one of the sections. I opened it, read it properly for the first time, and realised the tool had produced content that did not reflect what I wanted. I hadn’t scrutinised it in enough detail and I’d trusted the output because it looked the part.
It wasn’t catastrophic – nobody was harmed. Just maybe my ego was slightly bruised, which, honestly, I deserved. I hadn’t thought it through. I’d used the tool without really understanding it, without respecting what it was doing or its limitations. I’d treated it as an authority when it was offering me a first draft.
Once I saw that failing in myself, I started to notice it within my organisation, in some of the professionals I work alongside, across the industry. Experienced, capable people making the same mistake I’d made. Not because they were careless, but because nobody had helped them understand the difference between using a tool and thinking with one. That realisation is where everything I’m about to say comes from.
Sensible framework with a missing foundation
The BEFA published a report this month, written with Harlow Consulting and chaired by Mark Farmer, that sets out how construction professionals should approach AI outputs. The framework is clear: use it, challenge it, verify it, document it, and then decide.
Five sensible steps. I think most professionals reading that would nod along. Of course we should challenge AI outputs. Of course we should verify them. And of course the final decision should rest with the professional.
The report is valuable, timely and starts an important conversation. Where I think it falls short is in what it assumes. It assumes that:
- the professional sitting in front of the AI output already has the capability to do those five things;
- they understand enough about how AI works to challenge it meaningfully;
- they know what to look for when verifying; and
- they have a governance framework around them that supports documentation and accountability.
For some professionals, particularly those with decades of experience in their discipline, the challenge and verify steps may feel instinctive. They’ve spent years developing judgement. They know when something doesn’t smell right. That instinct is real, and it matters. It’s one of the most valuable things our industry has.
The problem is that instinct alone isn’t enough. Knowing your discipline doesn’t automatically mean you understand how an AI tool has processed your input, what it’s weighted, what data it’s drawn from, or where its blind spots are. Domain expertise and AI literacy are different capabilities – we need both.
The gap between the guideline and the ground
Here’s what this looks like in practice. Professionals are using AI, generating outputs, incorporating them into workflows and moving on. Some are doing this thoughtfully; many are not. Not because they’re careless or irresponsible, but because nobody has shown them what ‘challenge and verify’ actually means in a construction context.
“Think of it like handing someone the keys to a Formula One car as their daily driver, when they’re used to their five-door hatchback.”
What does it mean to verify an AI-generated programme when you haven’t mapped your own process into language the tool can understand? How do you challenge a design option when you’re not sure what questions to ask in the first place? How do you set the right parameters, frame the right prompts, understand the right boundaries?
This isn’t about a lack of professional skill. These are experienced people. It’s about a gap between how they work and how AI works. Most professionals have a clear process for tackling a project. Translating that process into something an AI tool can genuinely assist with – speaking its language, framing the right inputs, knowing when the output needs pushing back on – is a different capability entirely. Most people are figuring it out alone, without guidance or support.
Think of it like handing someone the keys to a Formula One car as their daily driver, when they’re used to doing the morning commute, the school run and nipping to the shops in their five-door hatchback. The machine is extraordinarily powerful. Without proper training, understanding and governance, at best you wouldn’t get it off the drive. At worst, it’s dangerous. That’s where we are with AI in construction right now: powerful tools and not enough preparation for the people using them.
A capability we haven’t developed
That moment with the feasibility RFQ was the start of this thinking for me. I’ve been developing a concept I call AI-IQ, born out of my own experience leading pre-construction teams through this transition. It describes the human intelligence required to work alongside artificial intelligence. Not whether you can operate the tool, but whether you can think critically alongside it.
It’s the ability to ask a good enough question, because the quality of any AI output depends entirely on the quality of the thinking going in. It is knowing how to structure your process so AI can genuinely assist, rather than just generating something that looks plausible. It’s understanding enough about how these tools work to know where they’re strong and where they need a professional checkpoint. It is the governance instinct to document your reasoning and maintain accountability for the decisions you make, even when AI has informed them.
The BEFA five-step framework is essentially AI-IQ in action. It describes the behaviours we need. Describing the behaviour, though, is not the same as developing the capability. Right now, we are not developing it systematically: not in universities, not in CPD and not in professional competence frameworks.
The CIOB published an AI Playbook in 2024, which is a positive step. It offers guidance, but not a mandate. There is no requirement within MCIOB or FCIOB competence frameworks for AI literacy. No CPD obligation requires critical evaluation of AI outputs as a professional skill. We’ve described the exam paper without teaching the course.
The cognitive risk we’re not talking about
“If we don’t design the learning alongside the automation, we will produce professionals who can operate tools, but cannot evaluate outcomes.”
Research published in January 2025 by Michael Gerlich at SBS Swiss Business School studied 666 participants across multiple age groups and found a significant negative correlation between frequent AI tool usage and critical thinking abilities. The study identified cognitive offloading, the tendency to delegate mental effort to AI tools, as the primary driver of that decline. Younger participants showed the highest dependence on AI tools and the lowest critical thinking scores.
Think about what that means for construction. Much of how professionals develop judgement is through analytical work: estimating, comparing, evaluating, making mistakes and learning why. If AI performs that layer, and nobody ensures the person is still thinking it through, we don’t just risk unreliable outputs today; we risk a generation of professionals who never develop the instinct to catch a problem before it becomes one.
The BEFA report acknowledges that AI may reduce time spent on document production, analysis, and routine processing. It notes this will raise questions about the role of junior staff. I’d put it more directly. If we don’t actively design the learning pathway alongside the automation, we will produce professionals who can operate tools, but cannot evaluate outcomes.
The opportunity in the challenge
This isn’t just a risk, though; it’s an invitation.
The construction industry needs 41,200 additional workers per year between now and 2030, according to the CITB’s latest Construction Workforce Outlook. Too few people are entering the sector and too many experienced workers are leaving. Productivity improvements haven’t closed the gap.
AI changes that equation if we approach it thoughtfully. Not as a replacement for capability, but as something that makes construction careers more attractive, more accessible, and more varied. If we develop AI-IQ alongside domain skills, we create a career proposition that speaks to people who might never have considered construction. For example, career changers who bring critical thinking and governance experience from other industries; or younger people who are already fluent in technology and need mentoring to apply it wisely.
The experienced professionals in our sector aren’t obstacles to AI adoption, they’re a bridge. They’re the ones who can pair their hard-won instinct for when something doesn’t stack up with a newer professional’s fluency in technology. That partnership is where AI-IQ develops. Not in a training course, but in the relationship between experience and energy, between wisdom and fluency.
“This starts with investing as much in the professionals who use these tools as we’ve invested in the tools themselves.”
A challenge worth rising to
The BEFA report gives us a sensible framework. Now we need to build the people who can deliver it.
That means professional bodies developing competence standards that include AI literacy and critical evaluation, not as optional CPD, but as core professional capability. It means organisations mapping where AI sits in their decision-making processes with clear checkpoints and named accountability. It means investing in the relationship between experienced and newer professionals, because that’s where the real capability transfer happens.
This isn’t a technology problem – it’s a human opportunity. The industry that built the world around us is more than capable of building the workforce to rise to this moment.
I started this piece telling you about a time I looked silly. A small moment, easily forgotten. I could have moved on and never thought about it again. Instead, it raised questions I couldn’t leave. What does it actually take to think well alongside these tools? How do we develop that capability in ourselves, in our teams, in our industry? I don’t have all the answers yet, I’m still working them out.
What I do know is that this starts with people, not technology. It starts with investing as much in the professionals who use these tools as we’ve invested in the tools themselves. Technology should amplify human potential, not replace it. I believe construction can show other industries what that looks like in practice. It starts with each of us being honest about where we are, and committing to closing the gap.
