Communication in the Digital Realm: Examining ChatGPT's Impact on the Leader-Subordinate Trust Relationship

Abstract

Since its initial public release in 2022, ChatGPT has experienced substantial growth, bringing about significant organizational advantages such as heightened proficiency, increased effectiveness, and reduced complexity. With its immense potential and groundbreaking capabilities, it is transforming industries and empowering people to accomplish tasks more effectively. It ushers a new area in human civilization, making it an indispensable tool in today’s digital landscape. However, the use of AI in leadership may bring about pitfalls. The present study put an emphasis on the application of ChatGPT in sensitive interpersonal situations, focusing on sexual harassment. It further delves into exploring potential impacts of perceived trustworthiness and authentic leadership on the dynamics of the leader-subordinate trust relationship. Our first three hypotheses suggested that using ChatGPT to prohibit sexual harassment negatively affects the perceived trustworthiness of the leader. This trust was measured based on three factors proposed by Mayer et al., (1995); benevolence, ability, and integrity. Our last three hypotheses examined the same effect on two components of authentic leadership: internalized moral perspective (in the leader and in an email), and relational transparency. Our statistical analysis, using one-way ANOVA, supported all six hypotheses. The results indicated that when a leader uses ChatGPT to construct a message about a personal matter, their subordinates may perceive them as less trustworthy and less authentic, potentially damaging the leader-subordinate trust relationship. These findings hold significant implications for future leaders, signaling the necessity of exercising careful judgment when contemplating the use of tools like ChatGPT. They underscore the persistent importance of human engagement and deliberation within leadership roles, suggesting that certain responsibilities may not be suited for support from LLMs or other forms of machine learning algorithms. Keywords: Artificial Intelligence, ChatGPT, communication, sexual harassment, trust, perceived trustworthiness, authentic leadership, Leader-Member Exchange theory, Stereotype Content Model.