Interactive Visualization for Fostering Trust in AI

Summary

Artificial Intelligence (AI) and other computational processes continue to influence decisions across a wide range of applications including healthcare decisions, vehicle navigation, data science, and others. This Dagstuhl seminar reflected on some of the challenges inherent in the goal of increasing the interpretability of these systems, and when applicable, increase the trust people put into them to make decisions. The seminar participants discussed the complexity of trust itself, and how the concept is multi-faceted, and likely outside of researchers in technology and computer science to fully define. We discussed an inter-disciplinary research agenda, as well as a manifesto that should help frame this direction going forward.