The chapter “Components of an Explanation for Co-Constructive sXAI” examines the fundamental components that constitute explanations within the framework of social explainable AI (sXAI). It defines key concepts such as the explanandum (the entity being explained), the explanans (the manner of explanation), the explainer (the provider), and the explainee (the recipient), and it explores their interactions. The chapter emphasizes the complexity of explanations, highlighting the dynamic nature of the explainee’s evolving understanding along with the contextual factors affecting the explanation process. It advocates an approach to co-constructed explanations in which the explanandum and the explanans adapt to the explainee’s needs, allowing roles to interchange. This contrasts with traditional XAI methods that assume a static, one-way knowledge transfer. By focusing on the conceptualization of co-constructive explanation, the chapter aims to inspire more effective and human-centered AI systems, setting the stage for future research in and the following chapters on social XAI.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Components of an Explanation for Co-constructive sXAI

  • Anna-Lisa Vollmer,
  • Heike M. Buhl,
  • Rachid Alami,
  • Angela Grimminger,
  • Axel-Cyrille Ngonga Ngomo

摘要

The chapter “Components of an Explanation for Co-Constructive sXAI” examines the fundamental components that constitute explanations within the framework of social explainable AI (sXAI). It defines key concepts such as the explanandum (the entity being explained), the explanans (the manner of explanation), the explainer (the provider), and the explainee (the recipient), and it explores their interactions. The chapter emphasizes the complexity of explanations, highlighting the dynamic nature of the explainee’s evolving understanding along with the contextual factors affecting the explanation process. It advocates an approach to co-constructed explanations in which the explanandum and the explanans adapt to the explainee’s needs, allowing roles to interchange. This contrasts with traditional XAI methods that assume a static, one-way knowledge transfer. By focusing on the conceptualization of co-constructive explanation, the chapter aims to inspire more effective and human-centered AI systems, setting the stage for future research in and the following chapters on social XAI.