The AI-teration Method and the Role of AI in Architectural Design

With the advent of artificial intelligence (AI) machines are able to challenge our notion that human intelligence can be the sole source of creativity. In this study, the author has developed a pipeline in which the client inputs a selection of images that they prefer, from which the computer infers the client’s preferences using image classification and suggests three design proposals in the form of a “mood board” using approximate nearest neighbor finding and generative adversarial networks (GANs). The mood board is a way of capturing and presenting architectural ideas to a client widely used in the architectural field. The machine’s output acts as “sketches”, similar in their look and feel to human created sketches that are included in traditional, human generated mood boards. The author believes that this technology will not be a replacement for the work of the architect, but rather be a complement to it, as the architect still has a valuable role to play in “deciphering” and interpreting the computer’s output, as well as in moving forward with the execution of the project. However, it has potential to speed up and enhance the iterative design process through the computer’s generative capabilities and the architect will benefit from the abundance of new ideas to accompany his or her own. In the long run, the architectural profession may change towards serving a more managerial role that brings computer made designs to (real) life.