Elliott School professors Alexa Alice Joubin and Susan Ariel Aaronson, along with their co-authors, have published "The epistemologies of trust: Conflicting worldviews in the Trustworthy AI discourse” (full text) in AI and Ethics, a leading journal in the field. As the second author, Joubin argues that trust resides in deep social connections between humans and between humans and non-human agents. Therefore, trust is fundamentally a human-centered, relational concept. Trust is not a simple property of a machine, but a dynamic between the user and AI applications. Stakeholders should construct trust-inducing environments with parameters that encourage trustworthy behaviors. Trust emerges from participating in processes and through exchanges among people or entities. This worldview focuses on the “social contract” of AI including the risks of anthropomorphism.
This article represents their ongoing interdisciplinary collaboration and grew out of their Oxbridge-style conversations as members of the GW Trustworthy AI initiative.