EUCA: the End-User-Centered Explainable AI Framework

• DOI: 10.48550/arxiv.2102.02437
• Publication Date: 2021-01-01
• Authors: Weina Jin; Jianyu Fan; Diane Gromala; Philippe Pasquier; Ghassan Hamarneh
• Keywords: Relevance; Layperson; Bridge (graph theory); Rapid Prototyping

The ability to explain decisions end-users is a necessity deploy AI as critical decision support. Yet making explainable non-technical relatively ignored and challenging problem. To bridge the gap, we first identify twelve end-user-friendly explanatory forms that do not require technical knowledge comprehend, including feature-, example-, rule-based explanations. We then instantiate prototyping cards in four AI-assisted decision-making tasks, conduct user study co-design low-fidelity prototypes with 32 layperson participants. results confirm relevance of using building blocks explanations, their proprieties - pros, cons, applicable explanation goals, design implications. forms, proprieties, supports (including suggested process, templates exemplars, associated algorithms actualize forms) constitute End-User-Centered framework EUCA, available at http://weinajin.github.io/end-user-xai . It serves practical toolkit for HCI/AI practitioners researchers understand requirements build end-user-centered AI.