Engineering intelligence - balancing the power of AI with human expertise

Across the infrastructure engineering sector, we are witnessing a shift in priorities. For decades, success was measured by the ability to deliver major capital projects, building new roads, bridges, rail lines, and essential utilities. Today, the emphasis is increasingly shifting towards how effectively we manage, operate, and sustain those existing assets to provide resilience in service provision. With public and private sector organisations facing tighter budgets and ageing infrastructure, the question we ask increasingly is how we can get more out of what we already have. This plays out against a backdrop of threats to UK CNI from increases in both extreme weather events and, as described in the Strategic Defence Review, state-backed cyber-attacks. This shift is contributing to a growing challenge around operational expenditure – how to manage, maintain, and optimise existing assets in a way that balances performance, reliability, and safety with the need to reduce ongoing costs and carbon emissions.

We are all aware that recent extreme weather events have placed a huge strain on critical national infrastructure, exposing weaknesses in both physical assets and the systems that manage them. Since 1980, average temperatures have increased by around 0.8 - 1°C, and all ten of the warmest years on record have occurred since 1990. More frequent heatwaves are placing a huge strain on infrastructure. Railway tracks buckle, road surfaces soften, and overhead power lines sag in extremely hot conditions – all of which affect safety and performance. Increased storm activity also poses significant challenges to our infrastructure systems, including water and transportation networks.

AI has clear potential to deliver transformative impact to our sector, and the need is pressing. A characteristic of CNI management is that change in operations and procedures can be difficult to implement due to the interplay of regulation, safety, technical standards or institutional conservatism. This attitude is important to prevent potential failures from compromising safety. But we also often miss the opportunity to drive improvement or exploit new technologies because we overestimate the potential impact of one wrong AI led decision versus the many human decisions based on little more than intuition.

The greatest opportunity for the application of AI in infrastructure is fundamentally transforming how operators interact with data and systems in their organisations to support decision making. This is as much about the business models and procedures within which we expect AI to operate as it is about the technology itself. Use of agentic AI to accelerate navigation of complex, legacy systems has tremendous potential to increase capacity of valuable human capital; providing greater insight to support fast-paced operational decisions. Despite this, the infrastructure sector still lags other industries in scaling innovation, often constrained by complex systems, legacy processes, and risk-averse cultures.

As AI becomes more embedded in the infrastructure engineering sector, its success in the medium term will depend not on the extent of its autonomy, but on how well it combines with human intelligence and expertise. AI lacks the context, experience, and ethical judgement that we demand from human experts. In safety-critical environments such as transport, energy, and utilities, those qualities are non-negotiable. A predictive AI model may flag that a bridge requires maintenance intervention, but only an experienced engineer can interpret that insight, assess the broader system implications, and approve the most effective and proportionate response.

To maximise the impact of this new technology, organisations must change the way that decisions are made to re-engineer the relationships between people and AI. This requires a balanced, risk-based approach – one that combines technical innovation with human assurance and governance. This means being braver in how we test and apply new technologies, but starting at manageable levels, i.e., pilot testing, controlled trials, and incremental scaling to build confidence and trust. A systems thinking approach will play a crucial role here. By viewing infrastructure as an interconnected whole system rather than a collection of individual assets, organisations can better understand where AI can add the most value and human oversight is least needed, with a view to then scale from there.

We have seen AI application gain traction where well focused pilots deliver clear value before progressing to more challenging problems. What's more, in each of these examples, human oversight remains the constant. Engineers, data scientists, and operators validate the insights, ensure they align with the context of wider strategic objectives, and provide the ethical and contextual lens that machines cannot. This clarity around analytics assurance such as verifying data quality, testing algorithms, and maintaining transparent processes, ensures AI strengthens decision-making rather than replacing it.

Ultimately, the organisations that will lead this new era of engineering intelligence are those that treat AI not as a replacement for human expertise, but as an extension of it. This must involve the organisations changing the way they operate and the way they interact with the technology. When paired thoughtfully, human and artificial intelligence can transform operational performance, enable smarter decisions, and unlock new levels of safety, efficiency, and resilience across the infrastructure on which we all rely.

This article first appeared on New Civil Engineer.