Expert systems and the control of risk in engineering

Engineering entails the collection, use and dissemination of information in order to make things, and it is a risk business in which decisions commonly have to be taken in the absence of firm information. The knowledge concerning the current state of a system (e.g. proposed or existing factory, power station, transport network) and it´s response to a set of actions may be incomplete, uncertain, or even contradictory. The traditional engineering approach has been to develop models to describe the system behaviour in terms of those variables which are readily quantifiable (e.g. load, temperature, pressure, strain). However, this approach only addresses a part of the problem since it implies a view of engineering as a purely technical process. Research into failures, in particular, suggests the need for a broader understanding of engineering as a socio-technical process. This paper discusses some of the possible implications of this change of emphasis on the nature of future expert systems in risk control