THE INFLUENCE OF CULTURAL FACTORS ON TRUST IN AUTOMATION

Abstract

Human interaction with automation is a complex process that requires both skilled operators and complex system designs to effectively enhance overall performance. Although automation has successfully managed complex systems throughout the world for over half a century, inappropriate reliance on automation can still occur, such as the recent malfunction in Tesla autopilot mechanisms that resulted in a fatality. Research has shown that trust, as an intervening variable, is critical to the development of appropriate reliance on automated systems. Because automation inevitably involves uncertainty, trust in automation is related to a calibration between a user’s expectations and the capabilities of automation. Prior studies suggest that trust is dynamic and influenced by both endogenous (e.g., cultural diversity) and exogenous (e.g., system reliability) variables. While a large body of work on trust in automation has accumulated over the past two decades, a standard measure has remained elusive, with research relying on short, idiosyncratically worded questionnaires. These challenges are exacerbated for measuring trust in automation in non-Western cultures because most research has been limited to North America and Western Europe. To determine how cultural factors affect various aspects of trust in and reliance on automation, the present research has developed a cross-cultural trust questionnaire and an air traffic control simulator that incorporates a variety of scenarios identified from a review of relevant literature. The measures and tasks have been validated by a crowdsourcing system (Amazon Mechanical Turk), as well as through experimental studies conducted in the U.S., Turkey, and Taiwan, with approximately 1000 participants. Over various phases of data collection and statistical evaluations, a final 18-item Universal Trust in Automation (UTA) instrument was identified that satisfies the stringent tests (including reliability and validity tests and measurement invariance analysis), indicating that the instrument is robust across national cultures and is effective in capturing both predispositions to trust and trust that evolves through use of a system. The findings reveal substantial cultural differences in human trust in automation, which have a significant impact on the design, implementation, and evaluation of automated systems to make them more trustworthy in determining the appropriate trust calibration for optimized reliance across cultures.