The buzz of AI in the building industry

The Buzz(word) of AI in the Building Industry

AI is often used as a catch-all buzzword without due consideration into the costs and benefits of introducing it. Akin to the many conversations around ‘blockchain’ technologies in 2019, the attractiveness of introducing ‘AI’ into our designs often lacks the real backing and understanding of these technologies in the first place.

Indeed, graduate engineer job requirements offer us an interesting insight into the direction of the industry, particularly in the larger consultancies. Over the past few years, we have seen preferences for candidates to be competent in programming languages (python being the big winner here), knowledge of parametric design and, more recently, in AI.

Often the big motivator for this is the fear of ‘being left behind’. However, I feel it is slightly more nuanced than this.

Fig 1. Utilising Revit API can unlock so much potential

It’s all sales-talk I suppose

Something that ‘AI’ has had, unlike other newly developing technologies in construction (BIM for example), is that it has a wide-reaching audience and adoption, not just limited to the building sphere.

This is technology that can be both extremely complex yet easily understood by the end user. Whilst there is some level of ‘magic’ behind the big machine running the code, we have Chat-GPT/Google Bard/Copilot to interact with, providing a humanistic interface and not just a wall of code.

It is understandable, therefore, why people might refer to all automation as AI because they view it as a simplified, meaningful and expandable approach to improving efficiency and intelligence in design; something in which traditional coding does not easily produce.

Am I anti-AI? Of course not! But there are few considerations when trying to utilise machine-learning on your automation tasks.

Fig 2. ChatGPT can do everything for me right?

The balance of cost & time

Preparing any significant and meaningful model requires careful and costly preparation of data, a significant amount of trial and error along with the computing and deployment costs. Is your product really AI if it just simply automates a problem? A better way to consider this would be asking the following three questions:

  • Can this program take inference from the data provided and present it in a meaningful way?
  • Will I benefit from the ‘hidden information’ that my deployment will provide?
  • Can this simply be done with traditional logic-based processing?
Additionally, what are you aiming to achieve by utilising AI technologies, is it for a more accurate design with a reduced number of errors? Is it for a more ‘intelligent’ product for the client? Or is it a cost saving measure?

What are the next steps for JDA?

I think it is important for us to remember the basics and approach this from a simple angle: What are the problem areas and which areas will have the best outcome from any changes?

For example, we are currently running a project which is looking to ensure data integrity over its shared systems, improve system integration and centralise project data in databases.

Being new to the company has given me a fresh insight into our processes, observing how people work and interact with our systems and software (not the easiest thing to do with remote working).

Fig 3. Me holding in tears whilst trying to fix never-ending errors

The long and short of it: we are creating amazing new ways of sharing data across our projects, reducing replication errors and increasing our efficiency on how we create schedules, calculations, schematics, etc, all using traditional programming.

Oh and, of course, we have some great deep-learning models (Using a YOLO detector) in the pipeline which will add a new capability and interaction into our designs – further improving our workflow.

In conclusion, this just goes to show that however much we wish to avoid any hype or buzz around AI, it slowly does encroach upon our day-to-day lives. Even we are still caught up in the development of AI solutions – something in which will hopefully increase the quality and capability of our technical work but also with a hint of not wanting to be left behind…