Human-AI Interfaces are a Central Component of Trustworthy AI

Abstract

This chapter demonstrates the crucial role that human-AI interfaces play in conveying the trustworthiness of AI solutions to their users. Explainability is a central component of such interfaces, particularly in high-stake domains where human oversight is essential: justice, finance, security, and medicine. To successfully build and communicate trustworthiness, a user-centered approach to the design and development of AI solutions and their human interfaces is essential. In this chapter, we explain how proven methods for stakeholder analysis and user testing from human-computer interaction (HCI) research can be adapted to human-AI interaction (HAII) in support of this goal. The practical implementation of a user-centric approach is described within the context of AI applications in computational pathology.