Urban morphology and artificial intelligence

Abstract

This commentary aims to concisely introduce artificial intelligence and urbantech for urban morphologists. We are in a midst of a new revolution in machine learning with ‘neural nets’ capable of understanding human speech and written language and analysing content on images and videos. The neural nets can semantically parse scenes on images recognizing objects, creating scene graphs, and describing content with text. However, specialized neural nets for urban morphology do not exist. Neural nets can recognise artefacts from specific historical ages or learn about architectural styles only if they are supervised by experts. To create urban morphological architectural intelligence that can help with morphological research or morphologically-informed urban design practices, urban morphologists need to translate their analytics and practices into software specifications. Creating specialized neural nets for urban morphology requires expertise in software engineering and programming effort and seems far in the future, but the International Seminar for Urban Form and Urban Morphology can play a profound role in debating urbantech, needs for intelligent tools and reaching to computational science and technology. Only through coordination and finding synergies the revolution of artificial intelligence will influence urban morphology as urbantech.