Decision Analysis for Management Judgment, 6th Edition

18Alternative decision-support systems and the role of AI

Introduction

In this chapter, we present an overview of an alternative way of aiding decision-making based on human judgment: linear modeling. Linear modeling (or bootstrapping) involves building a statistical model of a person’s judgments or predictions and subsequently utilizing the model instead of the person. The method is then compared with decision analysis and scenario planning. We go on to consider when it is safe to rely on snap judgments based on intuition, rather than deep reflective thought and analysis, and whether it is possible to subtly nudge people toward choosing better options. Earlier chapters have shown how artificial intelligence (AI) can complement decision analysis by, for example, suggesting attributes that people should consider when making decisions. AI tools are being increasingly used by businesses, and here we investigate in general how AI can support human decision-makers and the limitations of using these tools. We conclude by offering some final words of advice on how to apply decision analysis.

Statistical models of judgment

Most statistical models of judgment involve the use of a formula which is designed to represent, as closely as possible, the way that an expert arrives at their judgments. It is then possible to replace the expert with the model in the hope that this will yield improved estimates or decisions.

For example, the following formula might represent how a sales ...