An `AI in Design' View of Design


It appears from the literature on Engineering Design, and from interactions with engineering faculty, that there tends to be ``two cultures'': one concerned with the study of design from a domain point-of-view (e.g., Mechanical Engineering or Architecture), and the other concerned with the study of design processes and knowledge of designing.

This split echoes two prevailing views in design pedagogy. One is that design is an ``art''. This view tends to deny that the process of design can be studied, or taught, tends to focus on teaching domain knowledge, and argues that design must be learned from experience. As computational design tools, methods and literature (some influenced by AI) become increasingly used by designers (and taught to undergraduates) the strong form of this view is seen less often.

The other view, while acknowledging the vital role of experience in learning how to design, and the obvious need for domain knowledge, accepts that design can be studied, analysed, reproduced and taught. The area of AI in Design is consistent with this second view.

AI is viewed with some suspicion by many designers. This is probably because design is seen as an Engineering discipline, closely tied to analysis, to equations, and to deep domain knowledge. In contrast, AI is seen as ``soft'' science. AI researchers who study design are sometimes criticized (often correctly) for not having an adequate grasp of the required domain knowledge (e.g., material properties). Not knowing about the domain details is sometimes incorrectly associated with not being able to understand or know about the design process. There is also a strong sentiment that unless you can do design, you can't study it.

However, AI in Design researchers are interested in many of the same issues as engineers, and often base their work on studies of designers, on their knowledge and their activity [Dixon et al. 1988].

The focus of AI in Design studies is often on how and when the domain knowledge is used, rather than what it is [Chandrasekaran 1990]. There is a strong need for material that can be used to demonstrate and teach this distinction. This would be useful for both Engineering and Computer Science students.

Up : Next

/Papers/IJDC/sect1.1.html Thu May 22 22:13:24 EDT 1997