Looking at it from the standpoint of an architect, we can tell the “AI” we are querying to design a space, give it the code we are working with, share the desires of the end users and clients, program, and any constraints we know of, and let it go. It will compute the input and then come back with many iterations, but the most mathematically perfect solution might also be terrible from an aesthetics or comfort standpoint. Those results will still need to be reviewed and then worked upon by the designer.
(Photo credit: Autodesk)
The one thing that is certain is the usage of AI or generative design will become common in many industries as it will enable designers to do more, not less. Designers already use tools that cover some aspects of these features. For instance, energy analysis AI gives options or advice on glazing sizes or materials. There are also structural analysis algorithms that recommend sizing and shapes for different loads.
As designers, the tools we have at our disposal are both varied and powerful, yet many of them are still in separate silos. The promise of generative design or more advanced AI tools is they can look at many paths at once, rather than one at a time.
Several years ago, Airbus showed how they used generative design tools to create a new, lighter partition which was just as strong as the traditional one, and could save tons of CO2 per year if implemented. The design has been updated and refined over time using similar tools focused on the fabrication process as well as streamlining the approval and airworthiness qualifications. However, it was determined that a new fabrication process would be needed, and the generative design tool was called upon to help with the design of the partition as well as streamlining the workflow.
Think of this example like the work triangle we as designers consider when laying out a usable kitchen space. But now consider that triangle also includes multiple workers, parts, tools and moving components, and a difficult triangular-shaped site footprint to work with and more.
Notably, one significant aspect of AI that sets it apart from generative design is it can learn what is better. Consider this: your photos being uploaded to some services are automatically tagged as buildings or cats. Someone had to “teach” the computers how to “read” what the photo contained so it could work on its own.
As we progress with these technologies, we can teach them what we as the designers prefer, more light at the expense of HVAC efficiency for example. With these tools being used more, the algorithms within AI will become more and more refined, offering better results and improved building blocks for our design projects.