Interplay between algorithm aversion and betrayal aversion

Abstract:
Trust serves as a cornerstone in both human-human interactions and interactions involving human-AI interfaces. Unraveling the underlying psychological mechanisms that govern trust is crucial for enhancing the quality of these interactions. Extensive research within the realms of social psychology and behavioral economics has revealed that individuals tend to exhibit lower levels of trust towards computers and AI, particularly in scenarios involving natural risks, attributed to a heightened sensitivity to the prospect of betrayal. Conversely, studies stemming from the engineering domain have reported instances of reduced trust and a propensity for aversion towards AI agents, a phenomenon coined as algorithm aversion.

This intriguing paradox underscores the importance of comprehending the fundamental constituents of trust directed towards intellectual agents. This study adopts a behavioral experimental approach to delve into the intricate interplay between betrayal aversion and algorithm aversion. By carefully examining how these two distinct aversions interact, we aim to shed light on their combined influence on trust formation. This research not only addresses a critical gap in the literature but also holds the promise of fostering a deeper understanding of trust dynamics in human-AI interactions.