Rational Reasons for Realistic Reliance (R^2R^2)

This document seeks to illuminate the dangers of using the term 'trust' in relation to artificial intelligence (AI). The concept of 'reliance' is introduced as a suitable replacement. Several informal thought experiments are used to distinguish trust and reliance from one another, and to propose possible use cases for the latter. The goals of the paper are to address concerns about widely adopted paradigms surrounding AI, explain the dangers posed by said paradigms, and introduce methods to mitigate potential harm.