Original post on LinkedIn here.

“Can a machine design?”

This is a research question which design researcher and educator Nigel Cross had been exploring as early as the late 1960s. To Cross, the assumption motivating this question as an appropriate research direction is that it would, in turn, allow insights into the cognitive processes of human design activity, rather than “simply” trying to replace human design. As Cross explains:

“We might not necessarily want machines to do everything that human beings do, but setting challenges for machines to do some of the cognitively hard things that people do should give us insight into those things and into the broader nature of human cognitive abilities.”

Cross positions aspects of his prior research as demonstrating “that, in principle, a machine can do some things that many human beings regard as a uniquely human attribute.” On this basis, Cross can argue for a vision for designer-computer collaboration whereby we give machines “a sufficient degree of intelligent behavior, and a corresponding increase in participation in the design process, to liberate the designer from routine procedures and to enhance his decision-making role.”

Cross, then, proposed that we can learn important things about the nature of human design cognition through looking at design from a computational perspective. At the same time, he points to writing by Terry Liddament which challenges the computationalist paradigm in design research.

As Liddament explains, while the notion that some aspects of human cognitive activity can be encodable in some specifiable set of explicit, unambiguous instructions of the type that could be produced in the form of computational rules need not be a controversial assumption, we should be cautious not to conceptualise what is in effect a mechanical process as a _cognitive_ activity. Such a conceptualisation implies the reduction of cognitive activities to a specifiable set to do with algorithmic reasoning.

Such reductionism, Liddament argues, is not a viable strategy in the design domain. On the one hand, designers are typically “concerned with the particular and unique, with the actual concreteness of the specific objects they are handling and with the specific requirements set by the design brief or problem, rather than with generalisations.” On the other hand, designers typically encounter domains which may not be amenable to reductionist explanation; in such ill-specified or ‘open ended’ domains, human designers are required to impose and modify constraints as they develop their designs. Designers, in other words, are required to “frame” the problem space in which they design—this is a challenge AI research is still grappling with, and one that seemingly cannot be adequately solved by the very reductionist assumptions driving developments in AI.

Link to Cross’ 2001 article “Can a Machine Design?”