AI and digital tools can empower civil engineers, not replace them | New Civil Engineer
At the ICE/CIWEM Yorkshire and Humber Flooding and Water Quality Conference, one message cut through: AI and digital technologies are already strengthening engineering practice—and they should remain tools in the hands of engineers, not substitutes for expertise. In a closing session on digital and AI in flood resilience and water quality, engineers and data scientists shared real-world examples that track the field’s evolution, from early predictive models that supported the Thames Barrier to the machine learning systems now informing the Boston Barrier’s operations.
Machine learning that manages water, not magic
Machine learning excels at inductive reasoning—spotting patterns in data to make informed predictions. Just as we infer rain from cloud formations, ML models can parse unstructured data to forecast rainfall and river conditions, delivering insights at speed and scale we couldn’t achieve manually.
This isn’t theoretical. Bias-correction models now refine water forecasts used by the Thames Barrier, sharpening operational decisions. In Boston, layered ML systems blend tide, surge, and wave predictions to guide barrier closures with greater confidence. These data-driven augmentations help operators act earlier and more precisely, especially when the margin for error is small.
Visualisation is advancing too. AI-enhanced imaging translates complex hydraulic and structural models into accessible, realistic visuals of flood defences before construction. That clarity helps stakeholders—from communities to clients—understand design intent, risks, and trade-offs, enabling more informed engagement and approvals.
Digital-by-default delivery
Parametric modelling is reducing iteration time and uncertainty across civil schemes. By encoding design rules and dependencies, engineers can test how a single variable—say, crest height or slope angle—ripples through an embankment design, optimising for safety, cost, carbon, and constructability. The result is faster progress from concept to site without sacrificing rigour.
Data-driven triage is also proving its worth in high-stakes scenarios. A panellist described approaches used during the RAAC school building crisis, where analytics narrowed a universe of more than 500,000 assets down to 245 priority inspections. That strategic focus saved time and resources while directing attention to the highest-risk locations—an approach applicable across asset management, from bridges to flood defences.
Protecting public health in real time
AI-powered models are improving how we communicate water quality, particularly for open-water swimmers. By combining environmental data, operational inputs, and predictive intelligence, platforms can provide near real-time advice on bathing conditions and likely contamination events.
In New Zealand, the Safeswim Auckland programme delivers minute-by-minute updates on water quality and swimming conditions, protecting public health by showing current status and forecasting future risk. This kind of dynamic service is a template for cities seeking to build trust and reduce exposure to pollution events.
Use with care: humans stay in the loop
Across the panel, the consensus was clear: AI must be deployed responsibly, with engineers firmly in control. That starts with transparency—practitioners should understand how their models work and how they reach recommendations. This knowledge helps detect anomalies, manage bias, and set realistic confidence bounds around outputs.
Ethical decisions must remain human-led. Models should be monitored from training through deployment, with robust validation against ground truth. And AI should augment, not default. If a conventional method solves the problem with sufficient certainty, use it. Engineering judgement—context, experience, accountability—must anchor every decision.
What’s next: scaling responsibly
AI and digital tools promise faster delivery, safer assets, and more resilient communities. But scaling these systems raises new infrastructure questions, notably the energy and freshwater demands of large data centres. The sector must weigh the benefits of advanced analytics against their resource footprint and pursue efficiency by design.
The Institution of Civil Engineers has long championed innovation, and the takeaway from Yorkshire and Humber was unambiguous: AI should extend the capabilities of civil and infrastructure engineers, not replace them. By embedding digital tools thoughtfully into workflows—paired with strong governance and human oversight—we can deliver projects that are quicker to build, easier to operate, and better for people and the environment, while keeping engineering expertise at the heart of every choice.
