From Communication Anxiety to Psychological Reliance

Abstract: Grounded in self-determination theory (SDT), this study investigates how communication anxiety and psychological need satisfaction relate to the use of Large Language Model (LLM)-based chatbots, specifically ChatGPT, for communication tasks (e.g., writing emails, drafting messages, and preparing scripts). Using a three-wave panel survey of participants from the United States and South Korea, we examined both between- and within-person dynamics. Results showed that individuals with higher trait-level communication anxiety tended to use ChatGPT more frequently for communication tasks, whereas within-person fluctuations in anxiety did not predict greater use. ChatGPT use was positively associated with the satisfaction of basic psychological needs – autonomy, competence, and relatedness – and this need satisfaction, in turn, was associated with greater psychological reliance on ChatGPT. Importantly, need satisfaction further mediated the relationship between ChatGPT use and psychological reliance at both within- and between-person levels. Cultural comparisons revealed that US participants exhibited stronger momentary effects, while Korean participants showed more stable, trait-based patterns. These findings highlight the role of LLM-based chatbots as need-supportive communication tools that may become integrated into users’ communication practices over time, with implications for theories of AI-mediated communication, cross-cultural research, and the design of human-AI communication systems.