KCB Boiler Room Breakfast Club: Beyond AI: Visioning Design & Construction in 2030

“When AI meets design, it will always start with the vanilla middle. Push it more to the edge, and the more interesting it will get. All AI content, by definition, is derivative, but it doesn’t copy,” Michael Oh explained in his recent KCB Boiler Room Breakfast Club presentation. “There is no sense of good or bad design; right or wrong—except what you give it.”

2030 is a mere six years away, and yet when it comes to Artificial Intelligence, everything is changing rapidly. Those changes have already begun affecting the residential (and commercial) design and construction industry. Oh, Founder and President of TSP Smart Spaces, led BRBC attendees through a discussion of AI applications, beginning with defining machine learning and what came before AI— machines programmed with code—and what is happening now. In essence machine learning from large data sets so that machines (or entities) can handle functions and “perform behaviors that a person might reasonably call intelligent if a human were to do something similar.”

Specific Applications for Design and Building

The first app that Oh talked about was ChatGPT, explaining that GPT stands for Generative Pre-trained Transformer. What is striking about ChatGPT for those unfamiliar with it, is the speed to produce whatever content the user is asking for and the ability for ChatGPT to refine the output along the way. (By the way, this post has not been written by ChatGPT, although we do use it for certain tasks).

DALL-E and Firefly are two additional apps that Oh talked about. Where ChatGPT is great for content—words—these applications are made for images. Again, the speed is notable. These along with Mid-Journey have been explored and in some cases embraced by many designers and architects.

Examples of AI generated designs courtesy of Michael Oh, TSP Smart Spaces

However, in order to harness the power of AI, Oh uses the term “collaborative intelligence” (CQ). He defines this as “humans and generative AI tools working together to more quickly solve large-scale, complex problems.” Or, in some cases, maybe not so challenging but definitely with an attitude of assistance.

For developers, Oh introduced two applications, Deepblocks and Testfit, that can save time (and therefore money, presumably) for feasibility studies among other important tasks. 

For builders, Oh offered up Trimble, a company employing AI in a variety of settings, ranging from autonomous construction site machinery to more efficient workflows. Drone Deploy is another tool that is already available and in use. In fact, it has the capability to help create safer buildings by scanning worksites and noting irregularities.

Drone Deploy overhead image from their website of a residential project in Illinois

Michael Oh’s Predictions for 2030 and What to do Today

Architects - Prediction

Five people will do what twenty do today.

Designers - Prediction

Mass customization will happen; don’t compete with AI’s mastery of the “vanilla middle.”

Builders - Prediction

Efficiency and safety drastically increase.

Architects - Do Now

Think about what the other 15 could be doing, i.e. re-imagining everything client-facing.

Designers - Do Now

Seek what’s “edgy” and NOT “vanilla middle.” Also focus on everything human-centric that AI cannot do.

Builders - Do Now

Connectivity is key to enabling technology.