The Trust Architecture: Designing AI Systems That Earn Explainability-Based Trust

Key Points

• The aim is to propose a trust architecture that enables AI systems to earn user trust based on explainability.
• Developed a framework incorporating key design principles for explainability in AI systems.
• Analyzed user interaction patterns to identify trust factors related to explainability.
• Evaluated the proposed architecture through case studies and user feedback.
• Demonstrated that enhanced explainability significantly improves user trust in AI systems.
• Identified key design features that correlate with increased trust ratings.
• Showed that users prefer AI systems that provide clear and understandable explanations of their decisions.