Trust in Human-AI Interaction: Scoping Out Models, Measures, and Methods

Abstract

Trust has emerged as a key factor in people's interactions with AI-infused systems. Yet, little is known about what models of trust have been used and for what systems: robots, virtual characters, smart vehicles, decision aids, or others. Moreover, there is yet no known standard approach to measuring trust in AI. This scoping review maps out the state of affairs on trust in human-AI interaction (HAII) from the perspectives of models, measures, and methods. Findings suggest that trust is an important and multi-faceted topic of study within HAII contexts. However, most work is under-theorized and under-reported, generally not using established trust models and missing details about methods, especially Wizard of Oz. We offer several targets for systematic review work as well as a research agenda for combining the strengths and addressing the weaknesses of the current literature. CCS CONCEPTS • Human-centered computing • Human computer interaction (HCI) • HCI theory, concepts and models

Trust and Reliance in Consensus-Based Explanations from an Anti-Misinformation Agent

“…We used the Trust in Automation scale [14], a 7-point Likert scale consisting of 12 items: 7 for trust and 5 for distrust. Although developed for automation, it is also the most commonly used measure for AI [30]. Trust and distrust can exist simultaneously and are different concepts [19,28].…”

“…Trust in automation is defined as "the attitude that an agent will help achieve an individual's goals in a situation characterized by uncertainty and vulnerability" [18, p. 54]. This definition is based on a strong foundation of empirical research and frequently referenced in the context of AI [30]. Lee and See [18] define trust as an attitude and distinguish it from trust as a behavior.…”

“…Still, these concepts were developed for automation. Indeed, trust and reliance have not been well distinguished in research on AI-based systems and have been grouped together under the term trust [30]. Trust is typically measured in isolation as a subjective attitude through questionnaires and interviews [12,16,34].…”

“…Given the lack of established models of trust in AI and particularly XAI [30], we used Hoff and Bashir's model for automation [11,13]. Our instantiation of the model for our XAI agent and construction of hypotheses is in Figure 1.…”

“…A pressing question is what forms of explanations are effective at enabling lay end-users to trust XAI systems [2,30,33,34]. In social media platforms, lay people make judgments and decisions based on evaluating and integrating reports from multiple informants [4,31].…”