Building pathways for AI across infrastructure

Part of what inspires Mark Coates’ work is his curiosity regarding the technological tipping point the engineering sector is currently at.

AI tools have rapidly gone from being futuristic-sounding novelties to becoming an invaluable part of many people’s daily home and work lives. And Coates, as Bentley Systems’ vice-president for infrastructure policy advancement, is particularly interested in how this fast yet complicated adoption is playing out on the infrastructure side.

Coates took to the stage at Bentley’s Year in Infrastructure 2025 event in Amsterdam to highlight findings from the company’s white paper, The Impact of Artificial Intelligence on the Built Environment.

The white paper surveyed 130 infrastructure and building professionals on their firms’ readiness for AI and their predicted risks and returns from its deployment.

Coates said that some “78% of respondents found themselves at the early stages of adoption. Yet even at those early stages the respondents were almost 100% clear about what AI was going to evolve and bring into capability for them.”

He cited that 68% of respondents have AI policies in place and that 38% of respondents anticipate that over half of their projects will incorporate AI tools inside the next few years. “However, the sector is known for being very cautious when it comes to deploying new tools,” the Bentley man observed.

Anne-Marie Friel, a partner at global law firm Pinsent Masons, is one of the people tasked with assisting stakeholders in managing the risks associated with construction and infrastructure delivery.

Friel describes her primary focus as delivering integrated digital solutions to improve the performance of major projects and asset portfolios. For her, the white paper’s finding that 78% of survey participants said they considered themselves as ‘very early adopters’ of AI was particularly striking.

“We know that the built environment has been a relatively slower adopter compared with other industries. So, there's a huge opportunity there. But the real potential is actually when you look at connected value.”.

Providing an example, she said: “If you are able to take some of these insights from the infrastructure, but share it with other other data sets, that's where the real magic happens in terms of societal value and return on the AI investment.”

Friel is most interested in what she labels ‘bilateral use cases’. “They basically comprise a project within a project: so, you take a data set and share it with other data centres. This is how the absolutely powerful projects materialise,” she explained.

But no matter how smart the data generated and aggregated by AI tools, the idea that, for instance, a utility company will gladly share its proprietary data with a transportation department to enable a better outcome for all players remains laughable in the many countries where siloed working is part of the national DNA.

And Friel does concede that, as well as political will, there is an additional challenge involving the legal compliance side of data sharing.

Discussing ways through this murky legal and cultural landscape, Friel noted: “My job is to set up models and solutions that actually deliver outcomes. What I see missed all the time are opportunities to focus on the business models rather than the technologies.

“If you focus on the operating models and demonstrate to parties what the value is for them if they start sharing their data, you start presenting the technologies as real problem-solvers.

“I have seen projects that have floundered because they're focused on doing something clever with the technology rather than asking ‘how do we get stakeholders behind us – how do we get people wanting to participate?’”

For Friel, trust and transparency are paramount when rolling out new digital solutions. She recalled that “when ‘the AI show’ began, a lot of organisations jumped right into AI solutions and they quickly realised that without having strong foundations, deploying AI solutions can only amplify ineffectiveness without seeing the actual value”.

“We are now seeing more organisations going back to the foundations of data, process and standardisation. And that also spans data interoperability – between transport, energy and water players, for instance.”

More insights:
Digital drivers of the future energy mix and workforce
How digitalisation is pushing the envelope of conventional energy engineering

YJ Kim, the AI technical lead at engineering consultancy Mott MacDonald, is also grappling with developing strong technical foundations for a successful rollout of artificial intelligence. 

Describing her role, she said: “I build AI pathways for people so they can adopt and use it with confidence on a daily basis.”

A big part of Kim’s work involves helping organisations move out of the pilot stage, and her key advice to such organisations is not to shoot for the stars – well, not initially, anyway.

Kim explained: “If you look for the most likely wins, they are the relatively smaller risk deployments – and this makes sense from a technical perspective. Things such as document automation, large natural image processing, as well as small- and large-scale models; these are relatively mature and have been well tested across major cases not only within the built environment, but also across many other industries.”

Kim highlighted the value of AI governance among organisations saying: “We want to ensure that our AI adoption is secure, responsible and transparent. A lot of companies are now moving away from experimentation and into the practical realm. It’s important that we ensure that those solutions are technically sound, legally compliant and are well integrated.”

Echoing this point, Guy Beaumont, digital lead for infrastructure at the programme management expert Turner & Townsend, called for “a structured, set approach to move from the pilot stage”.

He observed: “Pilots have done their job. They've shown where the value is and where the high potential use cases sit. So now I’m looking to create foundation services, and this all goes back to data readiness.”

Beaumont advises organisations to get properly invested in getting their data up to scratch: “It’s about making sure data is performing to its requirements or standards and is in sufficient quality to make AI-based applications work effectively.”

He was also keen to focus on the importance of including the end user in rollout decisions. “A huge piece of the puzzle relates to user experiences. It's not just about using interfaces: it's about having to redesign day to day interactions with software.”

“There's huge focus around creating user-centred services that integrate the new environments people are working in today. And underlying that is the need for organisations to appoint the appropriate corporate sponsors: employees that take ownership of a clear AI strategy that keeps the user at the forefront.”

Bentley’s chief executive, Nicholas Cumins, used his speech to emphasise that greater rollout of digital tools must benefit all players, not only those with deep pockets. “It’s on us to democratise the use of AI, so that AI is available to all engineering firms, not just the most sophisticated ones.”

Cumins also explored the issue of data and described Bentley’s mutually beneficial approach to driving developments on this front.

He explained: “In our conversations with engineering firms, there is much concern about data and how data is being used by us. We are committed to data stewardship and of course we want our users to bring data into our systems that we can all benefit from. But with very few exceptions, we're not using any of that data to train our own AI.”

And Cumins reiterated the benefits that occur when all stakeholders can access digital tools: “In terms of principles guiding the rollout of AI solutions, we want to acknowledge the valuable role of immersion in new and ever-evolving solutions – by all players – in facilitating progress.”