Trust in A.I.: similarities with and differences from trust in humans

Abstract

Successful interactions between human and technology, especially A.I., are marked by trust. Trusted technology is used more adequately and appropriately. As human-A.I. interactions increase, understanding how trust is established and maintained between humans and their A.I. partners becomes more crucial. In this project, I investigated the impact of competence and reciprocity on human-A.I. compared to human-human trust. Results show that A.I. are trusted less than humans only when one interacts with both humans and A.I. sequentially. Even then, humans are trusted more than A.I. only in early rounds and when returns are not low. Therefore, patterns of human-A.I. trust are very similar to, but have few key differences from, patterns of human-human trust.