184 BUILT ENVIRONMENT VOL 46 NO 4 CITY INFORMATION MODELLING: DIGITAL PLANNING FOR SUSTAINABLE CITIES et al., 2008) and City Information Model (CIM) on the urban scale (Gil et al., 2011; Stojanovski, 2013, 12018). The availability of more powerful hard- ware in the early 2000s, along with new soft- ware that made the development of software tools easier, created the conditions for solutions that could be applied in practice (Braach, 2002a, Coates and Derix, 2008; Hovestadt, 2009). Finally, the development of visual program- ming tools and their integration into CAD systems (such as Grasshopper for Rhino 3D) made automation in urban design accessible to regular designers. However, the software solutions are either problem-driven and adapted to specific workflow challenges or lack a proper theoretical basis. They fail to put related approaches into perspective. The situation is comparable with the mid-1970s The development of software for automat- ing urban and architectural design began in the 1970s. The first publications on computer- aided architectural design (CAAD) included sections on methods for optimizing designs (Maver, 1971; Mitchell, 1977; Negroponte, 1970, 1979). Generative methods such as Shape Grammar (Stiny and Gips, 1972; Stiny and Mitchell, 1978) or methods for urban analysis (Hillier and Hanson, 1984; Negroponte and Groisser, 1970) followed. However, generative approaches and automation were rarely used in architectural and urban design practice and did not feature in CAAD software until the beginning of the 2000s. At that time, development focused on digital drawing and 3D modelling (Rooney and Steadman, 1988), and in turn led to the concept of a digital Building Information Model (BIM) (Eastman Levels of Automation in Urban Design Through Arti ficial Intelligence A Framework to Characterize Automation Approaches REINHARD KOENIG, MARTIN BIELIK, MARTIN DENNEMARK, THERESA FINK, SVEN SCHNEIDER and NORBERT SIEGMUND One of the most important consequences of digitalization and the progress of artificial intelligence is automation in all areas of life. In this paper we investigate the automation of urban design. Based on four levels of automation, we provide a conceptual framework for the classification and comparison of various urban design automation approaches and consider the scope of their applicability and the division of tasks between humans and computers. The proposed framework is trialled using two demonstration projects. Finally, we discuss the technical needs and possibilities for increasing urban design automation, as well as the implications these are expected to have for the profession of urban designers and architects. Contact: Reinhard Koenig  reinhard.koenig@uni-weimar.de