Abstract

Research on trust in Artificial Intelligence (AI) agents, often based on the model suggested by Mayer and colleagues, has shown that trust in the specific AI involved is an important intermediate element in a decision-making process about whether and how to use that AI. The recent advent of artificial companions (ACs) and similar technologies that foster emotional bonds between users and AI (e.g., social robots) presents an arguably different context: people interact with ACs and similar technologies also to form personalized long-term emotional bonding, much as they do with close human friends. Addressing this shift, we suggest revisiting the distinction between an “economic exchange” and a “social exchange” as defined in early research on trust, predating the current Mayer et al.-based consensus, in order to rethink the role of trust in these new AI contexts. An “economic exchange”, on the one hand, deals with short-term specific transactional relationships where the parties involved know what is expected of all sides – but where the lack of outcome guarantees and the risks involved make trust an important enabler of the transaction (i.e., trust as an intermediate element). This resembles much current research on AI adoption. A “social exchange”, on the other hand, is centered on creating long-term emotional bonding that is built through acts of non-obligatory and unnegotiated reciprocity. We discuss the nature of this distinction and its implications. Based on this, we outline several avenues for future research. We also emphasize the importance of including endocrine signatures, particularly oxytocin.

DOI

10.17705/1jais.01025

Download

Share

COinS