Risk analysis: lessons from artificial intelligence

Abstract Risk-analysis procedures have been available for many years. Risk-analysis procedures have been used to gauge the economic and technical hazards that are inherent in many commercial ventures. Most managers recognize the benefits of making decisions on the basis of probabilistic measures rather than single-point estimates. However, risk analyses are not yet common in project-oriented industries. One problem with present-day risk-analysis procedures is that those procedures that are simple enough for use by normal project personnel are too simplistic to capture the subtlety of risky situations. Those that are complex enough to capture the essence and subtlety of risky situations are so complex that they require an expert to operate them. Practically minded project personnel are reluctant to use procedures that appear to be too simplistic to yield useful results, and managers are equally reluctant to allocate scarce resources to the hiring of risk-analysis experts. To address this problem, one needs risk-analysis procedures that are able to model risky situations, but that hide their inherent computational complexity from the everyday user. Fortunately, new analysis tools are emerging that have the potential to allow complex risk analyses to be performed simply. These new tools, which are underpinned by decision analysis and, lately, expert-systems technology, may lead to powerful, yet simple, approaches to the representation of risky problems. The paper suggests a possible direction for the evolution of project risk-analysis procedures, and the tools that might be used to support these procedures.