Abstract
Explanations are a core component of collaboration between users and AI technology. Explanations must be designed in context, to address questions and information needs that will arise in actual use. Our approach for anticipating XAI needs includes: 1. Expert panel to identify usage situations. 2. Creation of a rudimentary design mockup. 3. Walkthrough with users in specific scenarios including edge cases. 4. Design iteration to support explanation needs that emerge.