AI Could Solve America's Infrastructure Problem. Institutions Need To Let Engineers Use It.

By Jonathan Fitzpatrick, Stantec

The state of America’s crumbling infrastructure continues to be a perennial concern as the scale of the problem continually outpaces both the funding and the human resources needed to solve it. Engineers have the solution — AI systems that offer unprecedented speed and potential cost savings — but to leverage its full potential, engineers need to take on a new role — and potentially a new business model. 

AI’s Impact On The Infrastructure Lifecycle

For all the concerns around AI adoption, there is clear potential and promise for AI to be the boost American infrastructure so desperately needs, streamlining everything from planning and design, to construction, operations, and maintenance. AI has already shown its potential to dissolve bottlenecks and close long-standing efficiency gaps. Some examples include:

• Maintenance planning through predictive modeling: AI systems trained on extensive historical engineering datasets and predictive modeling techniques can swiftly evaluate "what‑if" scenarios, identify delay patterns, and optimize project timelines — delivering efficiency gains where traditional methods have stagnated.
• Extreme weather modeling and impact mitigation: Especially as the natural world becomes more complex — fires are moving faster, flooding is more extreme, and storms are bigger — AI can help to build on top of what’s already known to help predict and model infrastructure performance in this greater range of scenarios. Advanced machine learning tools, informed by decades of hydrological modeling and rich geospatial datasets, now generate real-time, probabilistic flood-risk forecasts across entire watersheds. These dynamic, data-driven insights help engineers and planners anticipate not just riverine overflow but also flash flooding — empowering communities to plan more effective mitigation and protective measures.
• Construction and planning efficiency: In the design and renewable energy space, AI-powered computational tools now accelerate the design of large-scale installations, such as solar farms, by automating tasks like terrain grading, panel alignment, and shading analysis. What once took months of iterative calculations can now be completed in days or even hours, slashing earthworks by up to half, curtailing site disturbance and carbon emissions, lowering construction costs, and maximizing energy output.

Collectively, these AI-enabled approaches have shown to dramatically reduce engineering labor and construction budgets by as much as 15% while boosting precision and resilience of the infrastructure itself. 

Rethinking Engineering Roles

Given the scale of the opportunity AI offers to improve the infrastructure lifecycle, why hasn’t there been major improvement to the infrastructure process?

Part of the issue is the nature of the industries leading the charge. Between public service, which generally owns the infrastructure, and engineering, which is responsible for overseeing its maintenance, there are few industries more cautious about the adoption of new technologies. 

On average it takes governments a year and a half longer to adopt new technologies than the private sector. While the public sector struggles with the adoption of new technology due to slow bureaucratic processes, engineers — particularly those in construction-related fields — have a natural aversion to risk and liability concerns combined with a preference for proven solutions. Given the level of calculation and accuracy required in engineering projects, engineers tend to weigh prudence and care over speed and efficiency. 

The combined impact means that public infrastructure in the U.S. can be incredibly time-consuming to build and maintain; it’s no wonder that America’s infrastructure manages only a ‘C’ grade. 

For engineers, the potential of AI also means a likely evolution of their business model. Engineering firms have traditionally based their pricing models on the number of billable hours on a given project, and most firms aren’t ready to restructure their business models to a fully AI-optimized workforce. Rather than a time-based business model, engineering as an industry will have to explore what it means to price projects based on value produced. 

But there’s also the very real concern over risk and liability — concerns held equally by engineers and infrastructure owners.

From an engineering perspective, because AI isn’t built on the same deterministic models of previous technologies, the models are “black boxes” — there is an inherent level of opacity and uncertainty that has to be considered when looking at AI-driven results. Knowing what data an AI system was trained on becomes incredibly important to understanding how valid and trustworthy its results are; just because an AI model says this bridge will stand up to a storm, doesn’t mean it actually will, especially if it was trained on weak or inaccurate data. 

Pushing For AI: Engineers Role

While data validity concerns are real, to address the current scale of America’s infrastructure needs, engineers and their infrastructure clients need to begin pushing one another to determine the right AI tools and data standards to unlock this new efficiency frontier. 

This doesn’t mean that public works employees should go out and start asking generic large language models (LLMs) to create the first draft schematics for a new water treatment plant. The purpose of AI adoption isn’t (and shouldn’t be) to replace engineers.

Instead, rather than an engineer spending dozens of hours ensuring that an asset meets all of the regulatory requirements, AI tools can check designs for compliance and safety standards while still ensuring that engineers are the experts approving the final product. 

In this world, AI becomes both the fact-checker and the question-raiser, verifying human work and offering data-based suggestions for how things like maintenance should be improved or prioritized. It doesn’t pose a threat to the role of the engineer but instead frees engineers up to do more critical thinking about how actual civil and structural design can be improved. 

Applying caution when merging new technologies with engineering expertise is critical to ensuring project safety but shouldn’t be an excuse to demur from AI. There are an estimated 2.6 trillion dollars’ worth of infrastructure repairs needed — and engineers may just be able to tackle it, if they lean into the promise of AI.

Jonathan Fitzpatrick is the Director of Digital Products at Stantec.