IMPACT OF LANGUAGE MANIPULATION ON TRUST IN ARTIFICIAL INTELLIGENCE-SUPPORTED DECISION-MAKING IN HUMAN MACHINE TEAMS

Abstract

Decision support among leaders and within teams is at the precipice of defense when responding to and preventing armed conflict. As technology and use of technologies advance in aiding decision support, understanding dynamics between humans and artificial intelligence (AI) becomes even more critical. Existing research on decision support within human-AI teams identifies trust as a central factor affecting the performance and interactions amongst team members. Existing research indicates humans often encounter challenges in delegating decision-making authority to subordinates, and this hesitancy extends to decision support systems and AI-based solutions, largely due to a lack of trust in these technologies. A previous experiment investigated how time and interim assessments influence trust when an AI system is the sole source of decision support. This thesis expands on that prior experiment by manipulating the language used, to examine how the use of explicit trust language in intermediate decisions on AI-provided advice for decision-making influence trust.