---

How is artificial intelligence already changing the landscape of technical design and construction?

Sarah is a Chartered Construction Manager at HKA who  specialises in digital transformation and building information modelling (BIM) in construction disputes. In this interview, she considers the benefits artificial intelligence (AI) could bring to the construction industry, and why she looks forward to an automated future.

How is construction embracing the digital revolution?

Despite being an industry derided for tech resistance and old-fashioned working practices, construction is changing. As an expert witness specialising in disputes involving digital transformation and building information modelling (BIM), my existence is evidence that the industry has, at the very least, begun to embrace the digital revolution.

Before discussing the future, let us first consider our collective past: What steps towards digital transformation have we already made?

Figure 1 Design Production: AI generated images comparing paper-based design workflows to modern methods of design (Images created by Sarah Keyte)

The Design Process

When my colleague Shaun Russell, Associate Technical Director, describes his formative years as an Architect, design processes were analogue and entirely reliant on paper. Design drawings, contracts, invoices and QA/QC paperwork were all physical, paper-based records. Each office had records room filled with folded drawings, carbon copy paper, and old records.  Designers used drawing boards, set squares, scale rulers, and rotring pens to create design information.  Shaun’s early years as an architect required him to literally go back to the drawing board when clients changed their minds. Urgent communication was sent via fax.

By contrast, when I joined the construction industry a little over a decade ago, design and construction records were already typically digital by default. When clients changed their minds, technicians such as  me amended the intelligent 3D building information models, resulting in changes to drawings, design schedules, and project metadata.^[1]Metadata definition –  https://www.thenbs.com/knowledge/exploring-the-nbs-bim-object-standard-metadata-requirements Unsurprisingly, BIM adoption has grown in all sectors of the built environment, which has improved productivity.^[2]https://www.thenbs.com/digital-construction-report-2021/_download/NBS_digital_construction_report.pdf Digitalisation is understood to be fundamental to achieving the “golden thread of information”, making our built assets safer^[3] https://assets.publishing.service.gov.uk/media/5afc50c840f0b622e4844ab4/Building_a_Safer_Future_-_web.pdf

Figure 2 Document Management: AI generated images comparing paper-based information management and searches to digital (Images created by Sarah Keyte)

Document Management

The way project information is managed has changed in our lifetime. At the outset of one of my late former colleagues’ careers, he requested documents using carbon copy paper, and an office assistant would deliver the bundles of paper to his desk, placing them next to the ashtray.

By contrast, documents are typically digital by default. Nowadays, electronic document management systems have evolved to share, store, and manage large volumes of data on cloud-based platforms. Numerous project documents can be stored on cloud-based “common data environment” solutions such as Procure^[4]https://support.procore.com/products/online/user-guide/project-level/documents, Aconex^[5]https://www.oracle.com/uk/construction-engineering/aconex/, Asite, ProjectWise^[6]https://www.bentley.com/software/projectwise/, and AutoDesk’s construction cloud.^[7]https://construction.autodesk.co.uk/ Cloud-based document repositories have made remote access to numerous documents available from anywhere with internet access, making remote working possible.

Digitalisation has had a significant impact on the way designs are created and how information is managed.

But when I think about the technological developments in my colleagues and my lifetime the jukebox in my brain starts playing an upbeat song: Bachman-Turner Overdrive’s 70s hit “you ain’t seen nothin’ yet”.

What are the next key steps to meaningful change in technology that will improve our professional lives?

While I believe many issues we currently face as an industry can be alleviated by tools which already exist, without artificial intelligence (AI), they will remain crude and rudimentary.

 There are several AI tools that are already being developed and used in the construction industry, each with its own potential benefits (and risks) that AI may bring to our professions in the future.

Disclaimer: this article mentions products and services for information purposes. The inclusion of these products or services does not constitute an endorsement.

Let me begin with my definition of Artificial Intelligence

Artificial intelligence is a broad field that refers to the use of technologies to build machines and computers that have the ability to mimic cognitive functions associated with human intelligence.^[8]https://cloud.google.com/learn/artificial-intelligence-vs-machine-learning This includes the ability to see, understand, and respond to spoken or written language, analyse data, and more.

If you have heard of artificial intelligence, you may have also heard the term machine learning. Machine learning refers to the technologies and algorithms that enable systems to identify patterns, make decisions, and improve themselves through experience.^[9]https://ai.engineering.columbia.edu/ai-vs-machine-learning/#:~:text=Put%20in%20context%2C%20artificial%20intelligence,and%20improve%20themselves%20through%20experience

For example, ChatGPT^[10]https://chat.openai.com/ can answer questions and write content, hold a conversation, write and explain code in different programming languages, translate natural language to code, and more. Or rather, it will try to. Its success is based on the natural language prompts you feed it. It is not perfect yet; one study demonstrated ChatGPT answered more than 50% of software engineering questions posed to it incorrectly.^[11]https://arxiv.org/pdf/2308.02312.pdf

With the requisite data and sustained efforts from the tech industry, the capabilities of AI tools will improve over time.

How might AI help with design and production of design concepts?

Some designers are already using AI tools  to generate concepts. For example, AI tools like DALL-E^[12]https://openai.com/dall-e-2 and Midjourney^[13]https://www.architecture.com/knowledge-and-resources/knowledge-landing-page/artificial-intelligence-midjourney-for-architects can produce complex, detailed (and arguably ethereal) visualisations in seconds based on text prompts. The architect Wanyu utilises AI to depict what a skyscraper might look like if it were designed by Le Corbusier (1887-1965).^[14]https://www.theguardian.com/artanddesign/2023/aug/07/ai-architects-revolutionising-corbusier-architecture

Before these tools existed, visualisations of similar calibre would require significant time investments by skilled technicians, most likely using proprietary software. In short, AI tools are helping generate advanced visualisations quickly. For example, I created the image below using the prompts “create a visualisation of a futuristic skyscraper in the style of Zaha Hadid”. It was produced in under one minute.

Figure 3 AI generated image created in under one minute by Sarah Keyte

The positives are simple. Using AI tools, imaginative design concepts are only a few clicks away; I generated the image of skyscraper above in less than a minute. AI is helping to make complex parametric designs, the sort that took Zaha Hadid or Anton Gaudi years to master, easy to generate. Perhaps if Gaudi had access to the same tools, La Sagrada Familia would not be infamous for its delays.^[15]https://www.euronews.com/culture/2021/12/05/why-has-barcelona-s-sagrada-familia-taken-over-100-years-to-be-built

AI tools which produce captivating images are interesting, but they are not what the industry needs as a priority. An aspect of design which is crying out for innovation is design optimisation and quality assurance.

1. Design optimisation tools

AI tools are starting to help make design decisions. For example, ARKDesign.ai^[16]https://arkdesign.ai/ allows users to create optimised floor plans based on prompts, which adapt when a new criteria is added, while simultaneously creating useful metadata (e.g. square footage, cost).

Leading design software Autodesk introduced a “generative design” tool as part of Revit 2021.^[17]Revit 2021: Generative Design (youtube.com) The integrated tools enable designers to generate design options instantly. For example, the generative design tool can produce multiple layout options for desks based on variables such as the distance to exits, proximity to windows (i.e. views to outside), and desk count.^[18]https://www.autodesk.com/autodesk-university/article/Generative-Design-Revit-Workspace-Layout-2021

Tools such as Revit’s generative design tool can help designers make better design decisions in shorter time-frames.

How will these change the way we work?

Once generative AI tools such as Midjourney or ARKDesign.ai exist, it is arguably not particularly skillful to type prompts into an AI generator. The skill lies in asking the right questions to decipher client needs, and verifying the design to ensure it is compliant with the specification, contract, and relevant building standards. Even with the existence of these tools, design will still need to be guided by competent professionals. Machines will need to be guided by humans to learn initially, but hopefully machines will learn independently.

Figure 4 Imagining a future where AI tools help identify issues as early as possible.

2. Design quality assurance and verification

The case of Freeborn v Marcal [2019] EWHC 454 (TCC) centred around an architect producing a computer generated image,but failing to meet the most basic requirements of the Architects Registration Board code of conduct (record-keeping, ensuring a clear contract exists, keeping clients abreast of project updates).^[19]https://www.architectsjournal.co.uk/news/wonky-home-cinema-court-case-warning-to-all-architects

At present, technology does little to prevent disputes. Current design software tools are arguably rather rudimentary and do not automatically help to ensure designs are produced with reasonable skill and care.

For example, you may use clash detection tools to identify coordination issues, but designers are still required to move or change objects to resolve coordination flaws. Working for an expert witness consultancy, our technical departments have been flooded with claims relating to combustible cladding and cavity barriers. The designs may have been produced using market-leading building information modelling (BIM) software packages and yet they did not avoid these key design flaws.

In the future, AI may help verify design information. With machine learning advancements, the solutions suggested could be used to highlight and resolve design flaws.

And this hope may not be unrealistic; US-based CodeComply.AI^[20]https://www.codecomply.ai/ is on the machine learning journey to ensure designs meet building regulations. While this tool works with 2D design information, my hope is that in the future, 3D modelling software packages will be able to warn users of design flaws and deviations from standards to help produce a feasible and safe design.

However, even if software did offer to automatically resolve coordination or compliance issues, software vendors typically limit liability in their terms of use. For example, if parties rely on software and find the software contained a defect, the standard Autodesk License and Services Agreement limits Autodesk’s liability (to the extent permitted by law) to “(a) to attempt reasonably to remedy the breach or (b) to refund the amounts received for the affected subscription and terminate such subscription.”^[21]https://www.autodesk.com/company/terms-of-use/en/general-terms This is accompanied by extensive disclaimers.

In short, there is unlikely to be legal recourse for over-reliance on software. Designs will still need to be driven forward by competent professionals, but the machines could be used as an extra layer of quality control. Post-Grenfell, if our software could warn us about risks and flaws, this functionality could improve productivity, quality, and make the built environment safer for everyone.

3. Digitalisation and skills barriers

It is a commonly held understanding that poor digital skills are holding back BIM adoption in the construction industry.^[22]https://www.bimplus.co.uk/poor-digital-skills-hold-back-bim-adoption/

When speaking to designers trained during the 20^th century, many opine that building information modelling and analysis software is complex to learn. However, tech barriers may soon be lowered. Autodesk is developing AI tools which allow users to verbally describe objects to generate 3D models.^[23]https://arxiv.org/pdf/2110.02624.pdf At present, the results are rudimentary (the research paper shows 3D voxelized (Minecraft-style) models, but it’s not impossible to imagine a future whereby model generation does not rely solely on skilled technicians.

I like to imagine the benefits this may bring to experienced engineers who lack digital skills and the time to acquire them; they may no longer need to rely on digitally-capable technicians to produce design information. This democratisation of BIM software could allow intelligent 3D design information to be generated by anyone, regardless of digital capabilities.

4. Construction Record Keeping

Max Abrahamson remarked “A party to a dispute… will learn three lessons (often too late); the importance of records, the importance of records, and the importance of records.”^[24]Max W Abrahamson, Engineering Law and the ICE Contracts, 4th Edition Applied Science Publishers 1979.

This is sensible advice for any construction professional, but not all project records are equal. For example, site photos can be useful factual contemporaneous evidence, but their reliability varies. Is the photo date stamped? Is the location and subject matter clear?

While claims often benefit from site photos, the next step in visual site evidence is arguably site scans.

In my past life working on live construction sites across the UK and Europe, I used a matterport camera to scan the sites and take records. When claims arose, our records were clear and organised.

Similarly, Oculo.ai^[25]https://www.oculo.ai/ uses hard-hat-affixed cameras to generate 360 scans of project sites. Scans are geolocated, ensuring the locations of the records are clear. If a claim arises, past scans are easy to locate. With 360 cameras, scans are less likely than photos to result in evidence being outside the field of vision. In my experience, scans are easier to navigate and use than still images. However, recording data is only one aspect of what Oculo.ai and similar companies are attempting to do.

What does the future look like?

In the words of Bill Gates, we overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten.

Soon, AI is likely to develop to give us new, powerful digital tools. In the next two years, I hope to see improvements in tools to help make our professional lives easier, allowing us to work smarter, not harder. Designers (and perhaps their insurers) might use AI to check that their design is efficient, safe, and viable.

I hope the industry will embrace these tools as we might embrace a new limb. What they bring may be amazing, but we should anticipate that when we start using the new tools, they will inevitably be somewhat clumsy.

But perhaps meaningful changes to our professional lives will take a decade. AI developments can only happen if we as an industry embrace technology, test, and give feedback to improve it. Those ignoring AI do so at their own risk; businesses that want to stay competitive must embrace AI and find ways to integrate it into their operations.

AI will arguably not steal the jobs of competent professionals in the construction industry (yet), but it may significantly reduce the time it takes to complete tasks that many of us dislike (e.g. administration). If we can harness the potential of AI tools, we may be able to strike a better balance and become healthier as an industry. 

Improvements in productivity could allow us to focus on important challenges, such as focusing on reducing the environmental impact of the construction industry.

This publication presents the views, thoughts or opinions of the author and not necessarily those of HKA. Whilst we take every care to ensure the accuracy of this information at the time of publication, the content is not intended to deal with all aspects of the subject referred to, should not be relied upon and does not constitute advice of any kind. This publication is protected by copyright © 2024 HKA Global Ltd.