RESEARCH IN DESIGN COMPUTING: AN ARTIFICIAL INTELLIGENCE FRAMEWORK John S Gero Key Centre of Design Computing Department of Architectural and Design Science University of Sydney NSW 2006 Australia email: john@arch.usyd.edu.au 1. Introduction Design computing has often been considered a subset of computer applications that assist the designer in documenting and analysing complex designs. As one of many areas in which computer applications have been developed, design computing has relied on software developers and vendors to implement and market software with the relevant features and utilities to support some aspects of design activity. The development sin artificial intelligence and cognitive science over the last decades have provided new intellectual platforms to support research into design. In this paper we consider design computing as a research area which utilises these intellectual platforms, one in which the results of the research lead to more than additional computer programs and in fact lead to a better understanding of designing and computer support for designing. Considering design computing as a research area, we identify three sets of goals: (i) to develop theories, models and methods of designing as a process; (ii) to use these theories, models and methods as the basis for the development of tools; (iii) to use these theories, models and methods as the basis for teaching. The first set of goals utilises both artificial intelligence and cognitive science as the providers of research methods to support design computing research. In order to achieve the first set of goals, it is sometimes useful to consider computational models of design as a way of simulating design processes. However, human designers can also provide the basis for developing theories, models, and methods of designing. The second set of goals looks at the implications of particular theories, models, and methods of designing when considering computer support or automation of specific design tasks. This set of goals has a more direct correlation with the majority of design computing research currently taking place at universities. The third set of goals brings this understanding of design processes to bear on how we teach design. Here again, the focus is not entirely on computer applications for design, but on the use of computational models and/or cognitive models of design to inform design teaching. The Key Centre of Design Computing at the University of Sydney carries out teaching and research in the area of design computing. There are approximately 300 undergraduate architecture students, 60 graduate design computing students, 15-20 doctoral students, and 10 academic and research staff at the Key Centre. The artificial intelligence framework for design computing research presented here is based on research that has taken place at the Key Centre over the last 25 years. Design computing research can be pursued using a variety of scientific methods, an artificial intelligence framework that we find to be both useful and distinctive is based on the following three research methodologies: (i) empirically-based research (cognitive models); (ii) axiom-based research (computational models); and (ii) conjecture-based research (computational models). Empirically-based research involves the development of experimental studies of designers that result in cognitive models of designing, which then form the basis