Relational introspections with machines: a social-cognitive framework of trait perception, relational schemas, and trust in intention to use AI

This paper proposes a framework for understanding people’s relational introspections with machines. Through a social information processing lens, we argue that such introspections begin with relative perceptions of two basic traits: communion (prosocial intentionality) and agency (goal-pursuit capacities) for both machine and user. These relative perceptions are then interpreted through relational schemas in the short term and regulated by trust in the long term––jointly shaping relational behaviors toward the machine (e.g., intention to use). We surveyed three US samples of various AI-technology users (college students, N = 438; MTurk participants, N = 1,023; Prolific participants, N = 726, matched to census demographics). Results consistently showed that (a) intention to use such a machine was driven by self-agency followed by machine-communion, whereas machine-agency and self-communion had no direct impact, and (b) the relative importance of these trait perceptions was mediated by certain relational schemas and moderated by trust across respondents.