Key takeaways for civil infrastructure from ‘The Impact of Artificial Intelligence on the Built Environment’ white paper

By Colleen O’Hanlon

Artificial intelligence isn’t sitting on the sidelines anymore – it’s moving fast into the built environment. The Impact of Artifical Intelligence on the Built Environment, a white paper which surveyed more than 130 senior leaders across infrastructure and building globally shows that AI adoption is well underway and accelerating. The findings, published in October 2025 by Bentley Systems, Pinsent Masons, Mott MacDonald, and Turner & Townsend, make one thing clear: AI is going to reshape how civil infrastructure is designed, built, governed, and valued.

Mark Coates, Bentley Systems Vice President Infrastructure Policy Advancement, said that AI in infrastructure was no longer a thought experiment.

‘Almost half of the organisations we surveyed are already trialling AI or using it in day-to-day operations, and around a third expect more than half of their projects to rely on AI for design, engineering, and construction within just three years. The question is no longer if we adopt AI, but whether we can do it safely and at scale,’ he said.

‘What struck me is how practical the early use cases already are. Organisations are using AI to automate contract and document workflows, to speed up design and engineering, and to improve cost estimation, forecasting, and project scheduling. These are not science-fiction pilots; they are very real productivity plays on live projects.’

For an industry historically known for being slower to change, the survey highlights both huge opportunities and serious challenges as well as showcasing the significant pick up in pace of AI adoption and implementation. Here are the three big takeaways civil infrastructure leaders should be turning their attention to:

The survey revealed organisations are already using AI to automate contract and document workflows, to speed up design and engineering, and to improve cost estimation, forecasting, and project scheduling
Source: Shutterstock

1. AI adoption is real – and it’s starting with documentation, design, and engineering

One of the strongest signals from the survey is that AI isn’t just theoretical anymore. Most organisations are already trialling AI or using it in parts of their operations. Nearly half are applying it to automate documentation, and 40% are using it to optimise design and engineering.

Documentation-heavy processes- contracts, change requests, compensation events- are proving to be early wins. That’s no surprise in civil infrastructure, where paperwork and claims create major overhead and delay. AI is starting to ease that burden.

On the technical side, the sector is leaning into generative design, multi-factor optimisation, and engineering productivity tools. Forty percent of respondents are already using AI systems to optimise design processes, and generative AI (both generic and bespoke) is becoming mainstream.

The pattern is clear: organisations are deploying AI where it delivers measurable gains without compromising safety or regulatory obligations.

2. Business models are shifting toward data-enabled, integrated, outcome-driven delivery

Early AI adoption is about efficiency, but executives expect deeper disruption ahead. Forty percent anticipate a significant impact on their current business model, and nearly a quarter are already restructuring to prepare.

AI will accelerate the move from fragmented, document-driven delivery toward integrated, data-centred value chains.

The whitepaper points to several trends:

Civil infrastructure has long relied on siloed processes – designers hand over to contractors, contractors to operators- each with separate tools and risk positions. AI exposes the inefficiencies in that model.

As AI matures, data continuity across the asset lifecycle becomes economically essential. This will reshape contracting, collaboration frameworks, and procurement strategies.

What this means for civil infrastructure:

Mark Coates, Bentley Systems Vice President Infrastructure Policy Advancement
Source: Bentley Systems

3. Governance and risk management aren’t keeping up – and they’ll make or break AI’s success

Despite growing adoption, governance and risk controls are lagging. Mark said that the biggest brake on AI adoption today is not enthusiasm, it is trust in how data will be used.

‘Leaders told us they are most worried about data sharing risks such as IP, privacy, cybersecurity, and commercial sensitivity, followed closely by the complexity of integrating AI into existing systems and a lack of internal skills. If we want AI to scale, we have to solve those three issues together,’ he said.

The survey found:

Data-sharing risk- IP, privacy, cybersecurity, and commercial sensitivity- is the top barrier to adoption.

In a safety-critical industry, that’s a red flag. Firms are enthusiastic about AI, but many lack frameworks to ensure:

The survey also shows that while most organisations allow suppliers and contractors to use AI, many do so without robust oversight.

What this means for civil infrastructure:

Current and future use cases for AI.
Source: The Impact of Artificial Intelligence on the Built Environment, Pinsent Masons, Bentley Systems, Mott MacDonald and Turner & Townsend, September 2025.

The bottom line: AI is here – but the real transformation is still to come

In just three years, one-third of organisations expect more than half their projects to use AI for design, engineering, and construction.

To prepare, firms are prioritising investment in:

Civil infrastructure has a rare chance to break through long-standing productivity barriers and modernise how assets are conceived, delivered, and maintained.

Success will depend on how proactively organisations:

Mark said the research suggests a quiet but important shift.

‘Historically, many firms did not treat their project data as strategically important, because the value was hard to see. The success of data driven tools in finance and insurance has changed expectations. Leaders now understand that if they want to capture the upside of AI, they first need to treat their data as a core asset, not an afterthought,’ he said.