Benefits and risks 

The benefits and risks focus area assists you to characterize the potential benefits and inherent risks of the use case, using only minimal assumptions about the AI solution given in the Use Case focus area. In general, the depth of analysis devoted to potential benefits and inherent risks should align with the breadth and complexity of the use case, and the potential scale of deployment. Iteration is expected. Additionally, since use cases are often defined by working backwards from desired benefits, risks may require more effort to explore than benefits.

Technically, benefits and risks can be assessed similarly. Both are based on the likelihood and impact of system interactions, where a system interaction typically involves a sequence of input and output pairs, which can be considered as events in the interaction. A helpful interaction has a positive impact on a stakeholder, while a harmful interaction has a negative impact on a stakeholder.

The result of an interaction can be the compounding effect of multiple events. To estimate benefits, you identify examples of helpful interactions, categorize the examples into benefits, and then use your knowledge of your business to estimate the value of each benefit. Assessing risk is trickier, because there is usually less information available about harmful interactions. As a result, you first identify categories of potential harmful interactions, qualitatively estimate the likelihood and impact severity of each harmful interaction category, and map the qualitative estimates to a risk level. In most cases, benefits and risks should be assessed for each stakeholder group.

Note that the risk posed by the AI system implementation when applied to the use case is referred to as residual risk. This is computed in Monitoring, as part of the process of deciding whether the AI system is ready for release.