Merging Roles and Expertise: Redefining Stakeholder Characterization in Explainable Artificial Intelligence

Explainable Artificial Intelligence (XAI) strives to make Artificial Intelligence Systems (AIS) more understandable, thus tackling the 'black box' challenge. However, successful implementation requires precise identification of XAI requirements, made complex by the absence of universally accepted protocols. Given the importance of identifying stakeholders in this quest, this article proposes an innovative framework to characterize them. We compare and merge two predominant approaches: role-based and knowledge-based characterizations. The result is a novel framework, segmenting knowledge into subcategories while linking them to specific roles. This XAI Roles and Knowledge Framework offers a flexible methodology that can be adapted to the nuances of each XAI project. By providing a balance between specificity and generality, this tool aims to guide the implementation of XAI while ensuring that the stakeholders' needs are taken into account. By using this approach, XAI projects benefit from a more precise identification of needs, leading to outcomes more closely aligned with user expectations and greater transparency in AI decisions.