EUCA: A Practical Prototyping Framework towards End-User-Centered Explainable Artificial Intelligence

Fig. 1. End-user-friendly explanatory forms in the EUCA framework. The explanatory forms are shown on the right grids, each accompanied by a prototyping card example across four tasks used in the user study. The forms are a familiar language to both AI designers and end-users, thus overcoming the technical communication barriers between the two. The 12 explanatory forms are grouped into four categories to explain AI’s prediction on a new data point (the red dot in the leftmost 2D feature-space plot), or the model’s overall behavior: Explaining using features, examples, rules, and supplementary information. These categories correspond to the different aspects of showing AI’s learned representations at the feature, instance, and decision boundary level, indicated in the plot.