Published online by Cambridge University Press: 29 November 2013
The vulnerabilities inherent in choosing an innovative material for a design can make for technological catastrophe. The points of weakness are not dependent on the technical nature of the innovation, but instead are intrinsic to human problem-solving and decision-making. We present a computational approach for managing and containing this fallibility using ideas from cognitive science and artificial intelligence, in particular, understanding how the boundedly rational behavior of humans and organizations leads to specific kinds of errors that affect the choice and use of materials.
Materials choice in an industrial setting is often the result of a long chain or network of circumstances whose origins lie in many domains—from traditional engineering practice to the very latest innovation, from the commercial to the technical. In turn, every materials-selection decision has a long chain of consequences that are difficult to compute. This lengthy chain of consequences of a single decision gives rise to numerous points of error. What is particularly troublesome about these errors is not that they arise because of incomplete scientific engineering knowledge, but rather that they occur in spite of the fact that all (or at least, almost all) of the technically relevant information is available and sometimes even known to the technical personnel on the project. We will illustrate this approach by using a historical example of a major materials failure. Advances in database-systems design present an opportunity for integrating the ontology of material attributes with properties data. This may enable the design of more appropriate validation procedures required in proving a material for an artifact.