Episode 107 - AI in Architecture

This week we will be talking AI in architecture. This episode content meets PC2 - Clients, Users and Delivery of Services of the Part 3 Criteria.

Resources from today's episode:

Websites:

• https://www.architecture.com/knowledge-and-resources/resources-landing-page/riba-ai-report-2024

Thank you for listening! Please follow me on Instagram @part3withme for weekly content and updates. 

Join me next week for more Part3 With Me time.

If you liked this episode please give it a rating to help reach more fellow Part3er's!

Transcript

Episode 107:

Hello and Welcome to the Part3 with me podcast. 

The show that helps part 3 students jump-start into their careers as qualified architects and also provides refresher episodes for practising architects. I am your host Maria Skoutari and this week we will be talking about AI in Architecture. Today’s episode meets PC2 of the Part 3 Criteria.

As AI starts to become more and more prevalent in our everyday lives with rapid technological advancement, it has also started to become more widely used in the architectural profession as practices use it more day to day to assist with the project concept development. Leading to the discussion about the future of artificial intelligence and what it means for society generally and for the architectural profession. 

Architecture is no stranger to computational advancements with the use of Building Information Modelling, geographic information systems and other modelling software, the advancement of AI offer the potential for architects to take on far and more impactful roles than ever before.  

How can AI be used to improve and enhance day-to-day architectural practice:

Today, the global construction sector lags behind all other sectors in innovation and productivity. Compared to most other industrial and service sectors, whose productivity and performance have increased by roughly 1,500% over the past 70 years, the construction sector barely breaks a 1% improvement. This presents the architecture profession with an opportunity to leverage digital twin technologies to move from being drafting services for developers to high-valued professionals operating at a more strategic, policymaking levels, where profound decisions are regularly made which shape the performance and livability of every urban area in the world. Digital twin technologies enable architects to account for all the genomic complexities of an urban region’s DNA that represent how and why an urban region works in the way that it does. This offers significant opportunities for architects to participate in shaping the structure and form of urban regions of any size and location. This reshaping of urban forms needs to be done on both a large scale and an individual building or urban site scale, including retrofitting what has already been built.

When designing homes, workplaces, schools and communities, the architectural industry should embrace 3D modelling driven by AI, automation and data-supported software applications to solve even more complex design challenges, such as delivering more sustainable building outcomes, maximising building density and use of space without negatively impacting people’s quality of life or the environment in order to meet increasing demands from building owners, developers, citizens and municipalities, all while creating something that can stand the test of time.

The shift towards outcome-based design, powered by AI, enables architects to arrive at solutions faster and more efficiently making day-to-day management of projects easier. AI also makes it easy for architects to incorporate environmental and other contextual data into plans when optimising designs. AI not only provides real-time analytics, that fuel essential insights into operational energy, microclimate, sunlight, wind and noise, it also enables architects to test a wide variety of scenarios digitally, in a risk-free environment, to find optimal solutions within chosen parameters. 

It is envisioned that AI will serve as an assistant in the design process and routine task automation, with designers retaining their role as decision-makers controlling the creative process, as its the architect that has the real-world understanding of local specifics and needs, such as cultural and aesthetic concerns, regulatory issues of local and regional building codes or the complex multi-layered relationships with stakeholders and customers. Therefore, combining human intuition and expertise with AI’s computational capabilities allows architects to explore more possibilities for sustainable and innovative designs, leading to better-informed and more creative solutions. The integration of AI-enabled capabilities into the design process doesn’t necessarily mean that architects will no longer be required; instead, it empowers architects to focus on outcomes and winning work. 

Generally, AI within architecture refers to the application of computational techniques, algorithms and technologies to assist architects and designers at various stages of the architectural process. It is already being utilised in various ways, often without even being labelled as AI. Some of the more common areas of its current application include:  

Generative Design: AI algorithms generating and optimising design solutions based on criteria such as space requirements and aesthetic preferences. This approach allows architects to explore innovative options efficiently. 

Project management and scheduling: AI being used in Customer Relationship Management (CRM) and cloud-based project management software/systems to make predictions, optimise schedules, carry out risk assessments, and more. 

Building information modelling (BIM): AI is prevalent in existing BIM practices. For example, AI algorithms can analyse BIM models and provide valuable insights. It is also responsible for detecting clashes between architectural, structural and mechanical systems, thereby reducing errors and conflicts during construction.

Digital twins: Digital twins enable architects to simulate an existing building at maximum accuracy. This can be achieved through point cloud models of existing spaces, which can be used to create BIM models, or by providing a collaborative platform for architects, clients and stakeholders, where they can view and interact with a virtual model. 

Energy efficiency and sustainability: AI algorithms can optimise building performance and predict energy use, daylighting, thermal properties, and more. 

Technology interactions: AI can help architects and engineers to conduct structural analysis, analyse loads and optimise designs for maximum strength and safety. There may be a happy time when AI will produce building control drawings and specifications! 

So those are some examples in which AI can be utilised in architectural practice.

What is the data behind AI adoption in architecture practices:

A survey carried out by the RIBA revealed that a significant number of practices have started to use AI in at least of their projects approximately 41%. If the profession does not continue to adopt, adapt to and lead new technologies, it might not only fail to reap the benefits of innovation but also leave unguarded significant areas of current and future work and, therefore, revenue. In the recent survey, it has identified that 11% or practices consider themselves as leading digital innovators, with a further 19% identifying themselves as early adopters of digital innovation. This suggests around three in ten practices actively look to develop their practice offering through leading in digitisation. A near-majority of 47% believe their digital maturity is around where most are, suggesting a willingness to adopt new digital tools that are already of proven value. At the other end of the range, 17% are late adopters of digital innovation, and 5% resist digital innovation altogether, preferring traditional techniques. This resistance might be due to a lack of resources or in-house skills. Alternatively, it might be that for practices that take on particular types of work, digitisation has a more limited role. 

Most practices already use Building Information Modelling and the survey sought to explore and better understand the extent to which practices work with well-formed data and BIM protocols. Some adopt ISO 19650 when it comes to BIM, which is a series of international standards that describe a collaborative and consistent approach to information management for built assets. It forms part of the UK BIM Framework. The series includes concepts and principles (ISO 19650-1), the asset delivery phase (ISO 19650-2), the asset operational phase (ISO 19650-3), information exchange (ISO 19650-4) and information security (ISO 19650-5). Survey findings suggest there is a range in the extent to which practices create and maintain building models that comply with ISO 19650. A quarter of the respondents (26%) always create and maintain models that comply with ISO19650, a slightly higher proportion (28%) sometimes do, while 12% rarely do and just over a third (34%) never do. 

Small practices are significantly less likely to create and maintain ISO19650 compliant models, and large practices are significantly more likely to. Twelve per cent of small practices with between one and ten staff always create and maintain BIM models that comply with ISO 19650 and 62% never do. This ‘always’ figure rises to 43% for large (50 to 99 staff) practices, and 50% for those practices with 100 or more staff. 

This analysis may prove to be a foundation for the future development of AI within the profession. 

How aware of AI are architecture practices:

Currently, 2% of practices are using it on every project with two-fifths using it on at least the occasional project. Almost all architects have at least some knowledge of AI. The majority (51%) assess themselves as having a basic knowledge, and a third (32%) as having a practical knowledge of AI. A small percentage have advanced knowledge (6%) or are recognised authorities (2%). In contrast, fewer than one in ten (9%) architects have no knowledge of AI. Significant numbers of practices are using AI for at least the occasional project.

Generally, architects’ views about AI vary, in that they are both positive and negaive, with a common view being that AI is a threat to the profession; that it will destroy jobs by automating some or all of the roles architects fulfil. Architects are evenly split on this possible future, with 36% agreeing that AI is a threat to the architectural profession, 34% disagreeing and 30% neither agreeing nor disagreeing. A future without the profession is being seriously considered by some, suggesting architects need to engage with and shape the future of AI now. If AI can easily and cheaply output plausible design imitations that can be readily passed-off, the creative foundation of the profession may become vulnerable. However, AI also has a transformative potential. Modern buildings are becoming increasingly complex in their design, construction and maintenance where AI can be of extreme assistance. A near majority (49%) of architects agree that this complexity means the profession needs more and better digital tools, including AI. Practices could gain a competitive edge here by being early AI adopters, but this potential is likely to need investment to realise. Just one in five (20%) practices have invested resources in AI research and development. The majority (69%) have not.  

So what do the practices that do use AI think about it:

43% agree that AI has improved efficiency in the architectural design processes, while 24% disagree. This is the one area where current users of AI, on balance, see a clear benefit. More respondents disagree (39%) that AI currently enhances accuracy in modelling and simulations than agree (26%). Only 11% agree that AI enhances the accuracy of specifications, while a majority (49%) disagree. 

This lack of perceived benefit might be because current AI tools are not yet sufficiently developed, or because most architects lack the training and guidance needed to make the most of them. Or perhaps this is typical in the early stages of the adoption of a new technology. The benefits to practice are expected to increase over the next few years. Only 24% agree they have successfully integrated AI into bid creation, project management or scheduling, while a substantial 48% disagree. Twenty-one per cent agree that they have employed AI in environmental sustainability analysis, while 46% disagree. This level of integration is set to rise. Just 7% agree that AI has led to staff reductions and 61% disagree, suggesting that AI adoption is not significantly affecting practice staff levels.

A majority do not see the existential risks to the profession and employment coming in the next two years, but a significant amount do. Thirty-six per cent agree that AI will lead to staff reductions, while 30% disagree and 34% have no clear view. The view on the potential threat to the profession is finely balanced, with 35% agreeing that AI is a near-term threat to the profession, 36% disagreeing and 29% equivocal. 

At what stages of the design process do practices use AI:

The most common use is for early design stage visualisations, with 6% using it always, 22% often, 60% sometimes or rarely, and only 12% never using it for this purpose. This use of AI may help clients to see the possible resolutions of their brief more clearly, through detailed immersive visualisations. 

Twenty-one per cent of practices use AI for generative design always or often, while 31% use it sometimes, 16% rarely and almost a third (32%) never. AI-based generative design has the potential to fundamentally change the design process, allowing architects to rapidly create new and innovative designs, which can be explored, analysed and then refined to meet the brief in new and better ways. 

Parametric design has been a common feature of design tools since the early days of BIM and 3D design. AI has the potential to optimise and extend the use of parametric design, by drawing on wider and more complete datasets, allowing workable choices of building elements and systems to be algorithmically generated. Forty-three per cent of those using AI in some way always, often or sometimes use it for parametric design, while 17% rarely and 40% never use it for this purpose. Forty-three per cent never use AI for model generation, while only 2% always do, with 12% using AI often here, 22% sometimes and 21% rarely. 

An overwhelming 61% do not employ AI for specification writing. Creating the specification is possibly one of the least-loved parts of the design process and it might offer a significant opportunity for automation through AI, which could improve accuracy, material and product choice, and consistency with other sources of design information. 

Only a minority of those who use AI use it for construction product and material selection and analysis, building performance simulation, standards and regulatory compliance checking or environmental impact modelling. An example of how AI can be used is in delivering accurate and speedy regulation and compliance checking quickening planning application progress while helping to ensure that buildings are safe, accessible and sustainable.  

AI could also potentially assist with the business of architecture which are under significant pressure to remain profitable, there may be value in exploring the potential of AI to pick up project administration, leaving architects free to develop client relationships and create buildings and focus on fee-earning work. 

What are practices views on the future of AI in architecture:

Majority of practices (54%) agreed that in two years’ time AI will have been adopted in their practices, although a quarter (25%) disagree. The remainder (21%) are equivocal. This anticipated adoption is not quite matched by investment, with 41% anticipating that their practice will invest in AI research and development. A majority also agree that AI will be used to carry out environmental sustainability analysis (57%) and that it will improve efficiency in architectural design (57%). The majority also expect AI to enhance accuracy in modelling and simulations (49%). A significant minority expect AI will be integrated into bid creation and project management (41%) and will come to enhance the accuracy of their specifications (40%). Many of the comments from the survey also expressed the view that they forsee the role of AI will always be limited. For example, that AI could never be well-suited to considering the cultural, historical and social factors of design, nor the intricate spatial, structural or regulatory complexities of the design process, nor to make the subjective judgments of aesthetics and client preference. 

Practices generally view AI as a positive tool that can assist address certain challenges in the construction industry. 65% of respondents think that AI will have a positive effect on the productivity of the construction industry, and only 10% think it will have a negative effect. Half think that AI will have a positive effect on collaboration between architects and other professions, and only 14% think the effect will be negative. On balance, the effect of AI on project collaboration is anticipated to be positive, with 48% expecting AI to improve project collaboration and just 13% feeling the effect here will be negative. Perhaps because architects collaborate well together already, 49% think Al will make no difference. But even here, the balance is for AI to have a positive effect, with 31% believing it will be positive for collaboration between architects, and 20% that it will be negative.

The survey has also found that AI will have a positive effect on both design innovation (54% positive) and design creativity (48% positive). 44% of they survey respondents also anticipate AI having a positive effect on architectural education, with AI having the potential to offer tailored on-demand learning and immediate feedback. 

What concerns do the majority of architecture practices have with the prevalence of AI:

The most significant areas of concern are with regards to fees and employment. Without adequate fees, there will be fewer jobs in architecture, and ultimately no profession. Only 15% of respondents think that AI will have a positive effect on fee income, and a clear majority (56%) believe the effect will be negative. A significant minority (46%) anticipate negative effects on employment opportunities and only 22% see positive effects here. 

Also, AI brings with it some ethical concerns regarding, the risk of inadvertent plagiarism, how to attribute ownership of and charge a fair amount for work, and the changed relationship between the architect and others where machine-generated design and communication is interposed between them. A proportion of practices (21–27%) rank the ethical concerns as ‘significant’. 

Some other key challenges that AI may pose within the architecture industry include: 

Overreliance on certain aspects of AI: Insufficient human oversight could lead to unchecked biases or errors in AI-generated designs, or to designs that unintentionally imitate copyrighted material. To mitigate this risk, architects should remain critical of everything that AI produces, leveraging AI as a tool to enhance, rather than replace, human expertise and creativity. 

Legal and insurance considerations: Given the advancements of generative design iterations and technology interactions, it is important for architects to remember that AI is not an entity that can be held liable. Architects hold PI insurance and assume liability for all information produced as a result of the use of AI. 

Accuracy of the data used to train AI: The integration of AI in architecture has several attendant risks relating to the data used to train AI. For instance, biased, inaccurate or incomplete training data can lead to designs that introduce societal biases, result in inaccuracies in predictions and give rise to privacy concerns relating to sensitive data. To mitigate these risks, a diverse and transparent process of data selection should be applied with rigorous validation processes. 

Despite AI’s capabilities, there remains a need for human oversight and validation to ensure the accuracy, reliability and ethical integrity of the outcomes it produces. This dual responsibility requires architects to embrace AI as a valuable tool while also maintaining an observant stance, double-checking and verifying its outputs to mitigate potential errors or biases. If we take this approach, we can ensure responsible AI adoption while driving innovation in architecture. 

To sum up what I discussed today:

• AI is slowly becoming a more prominent tool in our day-to day lives and AI technologies have the potential to transform traditional approaches to design and construction, offering architects powerful tools to enhance efficiency and creativity and we are seeing more and more practices using such tools. 
• A survey carried out by RIBA has found that it is envisioned for AI to serve as an assistant in the design process and routine task automation, with designers retaining their role as decision-makers controlling the creative process, given the architect has the real-world understanding of local needs, regulatory issues and navigating the complex multi-layered relationships with stakeholders and customers.
• AI could also assist with the business of architecture which are under pressure to remain profitable, with the potential of AI to pick up project administration, leaving architects free to develop client relationships and focus on fee-earning work
• Architectural practices are still in the process of adapting and understanding AI with some practices already using AI tools to assist with their design process as a regular design development tool whilst others refuse to adopt such technologies, indicating the profession still has some way to go with integrating such technologies into their regular work streams but the future indicates that AI will be a positive and beneficial tool moving forward that can assist to address certain challenges in the construction industry.