Making better decisions earlier with responsible AI use

The construction industry needs to get better at making the right decisions early, and AI can help facilitate this if used in a safe and responsible way, Ben Brittain, director of public affairs at the Association for Consultancy and Engineering (ACE), told GIRI members at the autumn members’ meeting. 

ACE is a trade association representing consultant engineers, and Ben’s presentation focused on two connected ACE initiatives – it’s upcoming Design early, deliver better campaign, in partnership with GIRI, and the ACE Responsible AI Framework. “These two things might seem unrelated, but they are in fact two halves of the same story.”

Ben argued that it is essential that a safety-critical sector such as construction deploys AI in a safe and responsible way. He referenced findings from a report published by ACE and Autodesk at the end of 2025, which showed that 68% of UK business leaders are increasing their technology investment. “It's not a niche interest anymore. It’s board level. We're seeing productivity gains of up to 40% within businesses deploying AI and automation, and up to 25% fewer project delays. So, there are gains to be made from AI.”

However, he added that AI should augment engineers, not replace them. “That's crucial. The opportunity for productivity gains is real, but so is the responsibility that comes with it, which is why we built the framework.”

The opportunity

Ben noted that the construction sector grew only 10% between 2002 and 2022 compared to manufacturing's 90%. “We have been standing still while other industries have been transformed. We cannot allow that to continue.” 

Research from McKinsey suggests that AI and automation could unlock $228 billion a year in the US architecture, engineering, and construction sector by 2030 – and another $126 billion across Europe –  and that as much as half of the non-physical work in architecture and engineering is automatable in some form. “It is important to stress that this doesn’t mean it should be automated without oversight. That's the bridge into the framework. But the size of the prize is why we are leaning into this rather than waiting for it to happen.”

This is where the Design early, deliver better campaign comes in. The new campaign will be run partnership with GIRI in 2027. It will highlight the £25 billion a year lost to avoidable error in UK construction. “That’s more than we spend on maintaining the entire combined road and rail network every single year lost to mistakes that we, as an industry, could have prevented. And critically, GIRI’s research traces most of this back to failures in the project formation and design stage.

“That's the whole logic of this campaign: if the errors are rooted in the earliest decisions, that's where the intervention must happen. That's where the work must happen. That's what we must fix.”

Design early, deliver better

Ben explained that the campaign will have four pillars:

• front-loaded decisions through strategic partnerships, 
• equipping teams with better tools like AI-assisted design and predictive modelling, 
• establishing common ground among industry stakeholders around ‘get it right first time’, 
• and measuring value through pilot data.

Where does AI fit into ‘design early’? Ben explained that there are four applications of the ACE framework that cover this: 

• Generative and parametric design, so teams can explore more options before anything is locked in. 
• Predictive modelling and monitoring, which can surface structural and maintenance issues earlier.
• Analytics, scheduling and risk forecasting to enable programme and cost risk to be designed out rather than discovered later.
• Environmental and sustainability modelling at concept stage. 

“Each of these is powerful precisely because it acts early, which is also exactly why it needs those guardrails,” said Ben.

The ACE Responsible AI Framework 

The ACE framework has been developed in partnership with members and the Department for Business and Trade. “It is practical, non-prescriptive, and designed to scale up from a five-person practice to a 5,000-person business. This means that SMEs can use and have confidence in it, as can the big consultancies. Not all firms are at the same level with their AI deployment, and this framework reflects that and ensures it's a stepped journey for them.” 

Ben explained that the framework prescribes a ‘value test’ that should sit in front of any AI use. “Before any detailed governance, we ask members to apply a simple value test, and this comes down to four questions.” These questions are:

1. Does this AI use have clear value? 
2. Do the benefits outweigh the professional, legal, and reputational risks? 
3. Can you explain why and how AI was used, and what it has produced? 
4. Would your client regulator or professional body be comfortable with this use? 

“If the answer to any of those is unclear, that's a signal for the business to pause. This is not a box-ticking exercise. It's the first thing in the framework because everything else only matters once you've established the use cases worth pursuing in the first place.”

The framework then sets out eight core principles for responsible AI use: transparency, accountability, safety and reliability, security and privacy, fairness and integrity, competence, human oversight and environmental responsibility.

Accountability, said Ben, is particularly important. “This means that a named, qualified individual remains professionally responsible for the output of AI. And human oversight requires meaningful review before anything is relied on. We know that the deployment of AI is innately risky, and the compute and energy cost is real. For firms with net-zero commitments, AI belongs in the same conversation as any other sustainability decision. So ACE has taken the active decision to embed environmental considerations in the framework.”

As not every use of AI carries the same risks, the framework classifies risk by the consequences of failure, not by activity type.  For example, structural safety-critical work requires senior sign-off and mandatory client disclosure, whereas using AI to format documents merely requires the tools to be on the business’s approved list.  “The point is proportionality. One model doesn't fit all.”

Implementing AI governance

Ben explained that training requirements under the framework are graded by role, ranging from foundational awareness for all staff to strategic governance for leadership. “There is the expectation in this document that members are serious about training their staff on AI. That's fundamental and critical for safe deployment.”

Most critically, he added: “Liability rests with the engineer, not the tool. You are the human in the loop. AI can accelerate the work. It cannot accept professional responsibility for it.”

Two halves of the whole

Design early, deliver better and the AI framework are not competing initiatives but rather two campaigns with a shared purpose, said Ben. “Design early, deliver better is about making decisions sooner before change gets expensive. The responsible AI framework says that when you use AI to help make those decisions, do it with a value test, clear principles, and proportionate controls. Put together, the outcome we're after is fewer errors, safer AI adoption, and better outcome for clients, for our profession, and for the public we ultimately serve.”

Become a GIRI member.