Trusting Intelligent Automation in Expert Work: Accounting Practitioners’ Experiences and Perceptions

Abstract

AI-based applications are increasingly used in knowledge-intensive expert work, which has led to a discussion regarding their trustworthiness, i.e., to which degree these applications are ethical and reliable. While trust in technology is an important aspect of using and accepting novel information systems, little is known about domain experts’ trust in machine learning systems in their work. To provide a real-life, empirical perspective on the topic, this study reports findings from an interview study of accounting practitioners’ (N = 9) trust in intelligent automation in their work. The findings underline the holistic nature of trust, suggesting that contextual and social aspects, such as participatory design practices, shape domain experts’ trust in intelligent automation. For instance, the participants emphasize their contribution to product development and open communication with the system developers. In addition, the findings shed light on the characteristics of domain experts as technology users, such as the necessity of situation-specific expert knowledge when evaluating the systems’ reliability. Thus, our findings suggest that trust in intelligent automation manifests at different levels, both in human-AI interaction and interpersonal communication and collaboration. This research contributes to the existing literature on trust in technology, especially AI-powered applications, by providing insights into trust in intelligent automation in expert work.