It seems like you’re trying to use AI to accelerate construction. Would you like help?

This creates a familiar but underappreciated problem: Baumol’s Cost Disease. As productivity rises elsewhere, wages rise across the economy. Ultimately, we all fish for talent in the same labour pool. And so, sectors that cannot automate at the same pace see their costs increase without corresponding gains in output. Construction has lived with this dynamic for decades. AI risks intensifying it.

The implication is that AI will affect construction even if the sector does not meaningfully adopt it. If productivity accelerates in finance, technology and professional services, but not on building sites, then construction becomes relatively more expensive by default (and let’s be honest, the sector is hardly unfamiliar with cost overruns as it stands). More cost pressure, thinner margins and greater difficulty justifying investment follow as a matter of course.

When we look at where AI’s benefits are likely to accrue, the picture becomes clearer. At the risk of sounding a tad Marxist, the answer is always capital. Capital owners – particularly purveyors of software, data platforms and intellectual property – are well-positioned to capture value. Organisations delivering physical projects are at a disadvantage unless they can use capital to unlock genuine productivity improvements. Office-based roles that involve routine reading, writing and analysis are more exposed to automation than many onsite roles, at least in the short term.

This is not an argument for panic. Construction has a few characteristics that slow both the upside and downside of AI adoption. Projects are often poorly incentivised to invest in long-term digital capability. Physical labour remains difficult to substitute. And physical assets will continue to attract capital if they generate yield.

There is also a deeper issue that economics tends to gloss over: construction projects are traditionally cost-driven rather than value-driven. AI tools promise long-term efficiency gains, but require substantial upfront investment (not to mention cultural change). In a low-margin, risk-averse sector, that investment is hard to justify, particularly when benefits are diffuse, uncertain or realised after contractual responsibility has moved on. The result is a familiar cycle of underinvestment, partial implementation and disappointment.

These and other factors will mean that the pace of change will likely be slower in this sector, but also shelter it from the full impact of disruption. Seen in this light, AI looks less like a silver bullet and more like a new layer of infrastructure. Increasingly, it acts as a user interface over existing systems: helping people search, summarise, coordinate and document work. In that sense, AI may be closer to Excel than to science fiction.

For construction organisations, the opportunity is not to chase magic, but to reduce friction. Information management, document control, scheduling, reporting – the work around the work – remains stubbornly inefficient. These are precisely the areas where AI can deliver incremental but meaningful gains. AI will not fix the structural challenges of construction. But understood properly – as an economic force rather than a technological fantasy – it may help the sector navigate them more intelligently. That, at least, would be a good place to start.

For a deeper dive on this topic, join Ian Gordon and Dr Patrick Owen, asset information manager at Tideway London, at the Inspire Stage on day one of DCW.

AI was used in the preparation of this piece.

Register to attend DCW.

Keep up to date with DC+: sign up for the midweek newsletter.