As architects in new age of design we were interested in challenging the idea of what digital design is in the world of AI. With that said we wanted to delve in the realm of complex design process which isn’t absolute or definitive when it comes to decision making while substantiating the relationship of part to the whole. While doing so, we explored digital consolidation of architecture through the process of AI and supervised learning, using a folly as a vehicle. Questions like what is good and what can be perceived to be better was something we were interested in exploring further.

As computational designers working with various parameters, the idea to manipulate all of them to create an “ideal” one is challenging task. A defined script helps us generate unbiased model configurations and along with saves the parametric data for each, creating a huge library. This library gives us an array of possibilities to choose from, but this might be cumbersome task.

By introducing a supervised learning system, the library of data can be labelled as a set of target and non-target objects based on users’ own design sensibility. The machine learns the data parameters from the target data and return a set of preferred ones, learning the humans’ discernment. This new output set is what the “ideal” configuration should look like, saving our time and helping us approach the “better”. This outputted is what is then post processed by the user to create a folly.