Here are open questions that we would like to answer collaboratively with community members :
- Do you use AI systems to support decision making in your operations ?
- In which context do you see that an explanation could be useful ?
- For instance, to fill-in a lack of confidence in the machine-generated recommendation, or to leverage the ability of learning by experience for new joiners…?
- Now assuming there in an explanation enriching the AI system output, when could we say that explanations become useful in the operational process ?
How do we know that the explanation brings a positive impact to the human end user ?
Can it be measured by the type of actions that end-user decide to take after having read the explanation ?
It is very likely that the utility of an explanation will be related to the profile of end-user and how it relates to the user’s perception of AI systems