Why we can’t afford to ignore AI

Despite all the hype, our industry has been relatively slow in apprehending AI. Adoption remains low, use cases vague, and legacy methods stubbornly resilient. Yet AI has the potential to transform productivity - and if we are to meet the manifold challenges of the climate emergency, such a boost is indispensable. For our industry to harness this potential, we must begin working together more effectively to identify opportunities and minimise risks.

Imagine: it’s 2010. The rollout of 4G, improving smartphones, and e-commerce is converging in Uber’s ride-hailing gig-economy breakthrough. Except rather than Uber, the technology is instead in the hands of the construction industry - and, instead of Uber as we now know it, the industry builds a different app per location. Internal digital teams argue amongst themselves about whose app is best and at a company  level, with each prime contracting construction firm wrestling over which of the many apps should take primacy in each airport. 

Customers have to download a different app in each airport. Tribal digital teams in each region compete on whose Uber app is best. The real scaling necessary to get taxi drivers to use the app and grow the concept in each country is completely missed. We are left with long queues, pricey cabs, and the quintessential struggle for exact change at the end of the journey. The opportunity is lost; moreover, developments emerging from the original breakthrough are also lost. Uber Eats is never realised, and we all go hungry on a Saturday night in. And perhaps even Netflix isn’t there to entertain us.

It’s a crude analogy, but blended with an important truth. The construction sector is fragmented. It’s well known that this fragmentation obstructs productivity, hinders progress, and embeds unnecessary inefficiency. But the emergence of AI, and the cluster of machine learning technologies around it, is exposing our industry’s lack of integration at a deeper level. Simply put, harnessing AI is much, much harder in a divided, fragmented, and siloed industry. It’s riskier, it’s more expensive, and the benefits are less certain.

If construction were a fringe industry, with only a trivial relationship to the global economy, this might be an acceptable quirk. But construction is vital to almost every major challenge of the 21st century, from meeting the needs of a growing population to decarbonisation and climate change resilience. In our industry, if we don’t move fast, there are big consequences.

It isn’t due to a lack of enthusiasm. In the past 18 months, AI has skyrocketed up the agenda. Many firms now have a compelling story to tell about how they’re planning to deploy it in service of diverse use cases. Yet because every company is largely going it alone, the sheer volume of rework is sapping momentum and wasting precious capacity.

Moreover, if every company strives to design and deploy its own version of the next big thing, a crowded market lacking interoperability ends up suiting no-one - especially not our customers. Comparative advantage is lost, economies of scale recede, and valuable data remains locked inside walled gardens. And all the while, our whole sector loses a once-in-a-generation opportunity to leap forward, at a time when the world needs it most.

Robots and rigidity

Adjacent industries can show us how to implement innovation at pace. Earlier in my career, I was responsible for the deployment of robotic arms on a production line for a major manufacturer of cars and aeroplanes - applying a suite of cutting-edge tools to manufacturing of high tech consumer goods. In the motor industry, application of novel technologies was advanced, to the extent that it was part of the working culture. When I worked in the tech sector itself, the influence of cultural factors became even more obvious. In construction, a risk-averse mindset helps to uphold standards of care and responsibility. Unfortunately, it can also discourage breakthroughs. Whilst I never want to cross a chasm on a wobbly bridge, we all seem to use technology at pace to drive convenience and outcome, tolerating the odd glitch on the way.

Yet in the context of climate change, traditionalism may well be the higher risk option. Business as usual cannot deliver the efficient collaboration-at-speed necessary to win the race against a rapidly changing climate and the increased volatility associated with it. For once, brute force won’t be enough to blunder our way through.

By 2030, it’s estimated that around $130 trillion will be invested globally in improving capital infrastructure and moving towards renewables. That’s a seriously hefty sum - but it doesn’t absolve us from improving our processes. New schemes are conceptually challenging; permitting and consent is lengthy and labour-intensive; shifts in weather patterns, population, and regulations are challenging major projects to rethink their ways of working, from design to deployment and decommissioning. Even a sum like $130 trillion fails to guarantee momentous progress towards the infrastructure and energy systems we urgently need in the very near future.

As well as efficiency, AI is indispensable for another reason. The staffing crisis in construction is getting worse. We are struggling to fill vacancies across a range of critical roles, and the problem tends to be more acute in burgeoning fields like data science. If we don’t use AI to tackle human capital deficits, we may not have enough people to transform our industry. Overheating, resiliency, decarbonisation: the inefficiency of legacy methods, combined with a lack of human resources, is seriously undermining our capacity to protect our existing physical infrastructure from the ravages of rising seas and volatile climate.

Impossible to inevitable

The good news is that change is happening. From identifying how best to make assets more resilient to damage, to where best to allocate capital investment, AI is accelerating our learning and decision-making across a variety of complex areas. With so many variables, and so little time, we have no choice but to quickly and collectively learn how best to deploy machine learning and trust its outputs.

Our priority must be to focus on the most convincing use cases. Optimising and protecting critical national infrastructure amid increasing floods, droughts, and storms is as uncontroversial as they come. For example, machine learning enabled the Federal Emergency Management Agency (FEMA) to rapidly assess over 146,000 structures, identifying over 30,000 for inspection - accelerating resilience interventions and saving FEMA millions in the first year alone. Successful applications can then serve as proof of value for broader needs to scale across the industry at large. As well as mitigating catastrophic effects, the broad penetration of AI in construction will lay the foundations for the next generation of innovations.

To see what’s possible, we should look beyond our own industry, and adopt best practices in sectors where AI adoption is rapid. Engineers love to create, but can’t match the countless billions tech has already poured into generative AI, let alone its ongoing investments. And thankfully, we don’t need to. By working together, pooling our efforts, and making use of what’s already available for us, we can avoid reinventing the wheel - or indeed the ride-hailing app - and focus on what we do best: building a better world.