A Brave New Blueprint: engineering and liability in a world with AI

Introduction: Community, identity…liabilityⁱ?

Since the introduction of CAD, few innovations have promised to redefine the work of consulting engineers, architects, and construction professionals as profoundly as Artificial Intelligence (‘AI’).  In its various forms AI, from machine learning algorithms and neural networks to sophisticated generative design tools, it is no longer a theoretical concept from science fictionⁱⁱ, despite our play on Huxley’s classic in the title. It is a tangible, powerful, and increasingly used tool in the modern design toolkit. Across the UK, Ireland and beyond, from One Eastside’s digital twinⁱⁱⁱ to the optimisation of hydraulic models for local flood defence systems^iv, AI is not simply enhancing productivity; it is challenging established practices and beginning to reshape what is possible. Indeed, the comparison with the introduction of CAD is a poor one. While CAD helped (and continues to help) professionals draw their ideas with greater efficiency, AI is capable of autonomously generating design options, identifying performance optimisations, and uncovering solutions that might not have been otherwise conceived of by the human designer. That last phrase alone – “human designer” – is telling of itself. Have we ever had to clarify before that the designer was homo sapiens? 

The technological wave promises much. Generative design algorithms can produce dozens of optimised structural or artistic options in the time it takes an experienced professional to sketch one^v. These can be capable of meeting complex multi-objective constraints; from materials usage, maximising sustainability^vi, measuring and reducing embodied carbon^vii, energy performance^viii through to cost^ix. AI-powered analytics can predict structural fatigue^x, automate building regulation compliance checks^xi, and manage the vast data streams of Building Information Modelling (’BIM‘) with record speed and accuracy^xii.   

This future of unprecedented efficiency, sustainability, and innovation, through to the reduction in human error, sees the shoulders of AI bearing a heavy load. Before we get too carried away on this technological wave, it is worth noting that many commentators say it’s too heavy a load. There is also a significant question mark as to the willingness of the construction industry, particularly those responsible for the physical construction, to get to grips with adopting and heavily utilising digital technologies. As reported on by the Australian Contractors Association, a global McKinsey study reported that only hunting and fishing have a worse track record than construction when it comes to the adoption of digital technologies^xiii. 

There are philosophical points too in grappling with this technology. As AI becomes more sophisticated and the tools begin to assume tasks traditionally performed by engineers and architects, what will we allow ‘it’ to do, what do we retain for the province of humanity and why? The answer to the ‘why’ lies not just in the current limitations of AI, but fundamentally in the recognition of the intrinsic value of human attributes – such as creativity, intuition, ethical judgment, and the pursuit of continual improvement – that contribute to fulfilling and meaningful work.  

Consider the contrasting visions of Henry Ford and Ove Arup. Ford saw the machine as a liberator, automating labour to achieve unprecedented efficiency and free individuals for leisure. As the great industrialist said: “man minus the Machine is a slave; Man plus the Machine is a freeman”. Arup, in contrast, championed the inherent value of work itself, advocating for a work environment that is both ‘interesting and rewarding’ and fosters a sense of purpose and fulfilment. Arup believed a fundamental principle was to find enjoyment not just in leisure, but equally in the time spent working. Put another way, we must find fulfilment and purpose in our work, rather than it being simply a means to an end.  Arup considered this distinction in his 1970s vision^xiv. 

The challenge with AI lies in reconciling these two perspectives. Can we harness AI’s power to automate the mundane while simultaneously creating opportunities for more profound and meaningful human engagement? Can we bring together Ford and Arup?  The answer lies in carefully navigating the integration of AI, ensuring it enhances human capabilities and elevates the professions, rather than simply reducing work to a means to an end, or ceding fundamental aspects of professional fulfilment to automation, risking the depersonalisation, over-reliance, or job losses that could undermine the very essence of satisfying, human centred work. 

Many commentators argue that this careful navigation, necessitates a technological tightrope walk to build a future where AI works with humans, enhancing human wisdom and innovation, rather than replacing them. 

Equally importantly, from the perspective of a publication dealing with insurance, this algorithmic revolution brings with it a shadow of uncertainty and risk that strikes at the very heart of professional accountability. It’s an accountability that we are very close to, having spent the last 90-odd years thinking about how it should be insured. The traditional model of professional liability is predicated on the judgment, skill, and care of a professional person. As our clients know, our business is linked inexorably to theirs: we are in the business of helping professionals understand, manage, mitigate and transfer their assumed professional risks. The use of AI begs many fundamental questions. This publication attempts to consider some of them. Most fundamental for us, however, is what happens to the established order of professional liability when a critical design decision is influenced, augmented, or even wholly delegated to an algorithm?  

If an AI-optimised beam fails, a generative-designed facade leaks, or an AI-audited M&E system proves non-compliant, who is legally and financially responsible…or who should be?  Is it the architect who trusted the tool, the structural engineer who signed off on the output, the software developer who coded the algorithm, the data provider whose flawed data may have corrupted the AI’s analysis, or the client who mandated the use of AI in order to cut costs?  Or does AI just add another layer of complexity to an inevitable multi-party dispute?  Unusually for one of our publications, do we need to ask where moral and ethical responsibilities lie?   

This article will explore this new reality. We will first consider the specific applications and inherent risks of AI in the architectural and engineering sectors.  We will then take a look at the legal quagmire this has the potential to create, particularly concerning the established ‘standard of care’.  We will then consider the critical role of Professional Indemnity (‘PI’) insurance—how current policies are likely to respond to AI-related claims, why the existing framework is under existential strain, and how the insurance market must adapt to remain relevant and effective. Finally, we will consider how the future of underwriting might change and what the crucial questions insurers might begin to ask to distinguish responsible AI adopters from those courting disaster.   

The broader point, alluded to in our sub-title, is that the dawn of AI is about more than liability. Careful thought needs to be given to its impact on our communities: our communities of professionals and their broader ecosystems. Who will they be collaborating with in future and how?  How will new professionals be brought into the industry and what will their roles and careers look like?  Equally, its impact on our professional identities too. What will being an ‘engineer’ or ‘architect’ mean in the second half of the 21st century?  We have long argued that the status of the professions needs to be raised in the public imagination^xv. AI presents many opportunities and challenges to that aim and this must be a key focus of the major engineering and architectural institutions.   

Most of these questions are for another day and another author. For now, we hope this series of articles serves as a useful backdrop to our forthcoming series of roundtables we are planning to host with our clients on the subject where we will ponder: in an age of intelligent machines, how can we ensure the sustainable insurability of the professions?  

References

ⁱApologies to Aldous Huxley 
ⁱⁱArchilabs.ai
ⁱⁱⁱ Skyscraper in Birmingham to make use of digital twins
^iv FloodAI
^v The architect’s guide to generative design: what is generative design?
^vi Fatima Alsakka et al., Digital twin for production estimation, scheduling and real-time monitoring in offsite construction, Computers & Industrial Engineering Volume 191, May 2024
^vii Zhi Xian Chew et al., Generative Design in the Built Environment, Automation in Construction, Volume 166, October 2024
^viii Phattranis Suphavarophas et al., A Systematic Review of Applications of Generative Design Methods for Energy Efficiency in Buildings, Buildings 2024, 14, 1311
^ix Developing a Generative AI platform for construction cost management 
^x W Schneller et al., Artificial intelligence assisted fatigue failure prediction, International Journal of Fatigue, 155, 2022
^xi Dayou Chen et al., Automated fire risk assessment and mitigation in building blueprints using computer vision and deep generative models, Advanced Engineering Informatics, Vol.62, Part A, October 2024; Advanced Engineering Informatics Volume 62, Part A, October 2024
^xiiBIM and AI Integration Shaping The Future of Construction
^xiiiAustralian Contractors Association, Disrupt or Die Transforming Australia’s construction industry, 2024
^xivhttps://www.arup.com/about-us/corporate-reports/ove-arup-key-speech/
^xvConstructing Change: Evolving the status quo or time to reset?