Mapping Out the Context

Abstract: Generative artificial intelligence (GenAI) can write humanlike, context-aware messages and thus can aid individuals in their communication with others. Existing research suggests that using GenAI may be regarded as inappropriate or undesirable in certain interpersonal contexts, but acceptable in others. Yet, few studies have articulated the reasons underlying such context-dependent preferences. Taking a functional approach, we conducted an online survey ( N = 542) to explore how GenAI use might depend on an individual’s goals in interpersonal communication and to unravel the reasons why an individual may or may not use GenAI in specific contexts. Among the participants who had used GenAI in their interpersonal communication with others ( n = 240), we found that GenAI was most often used for goals of information and resource exchange. When message production for a goal requires more of one’s subjective experience (e.g., providing emotional support, showing others who you are), individuals are less likely to use GenAI in such contexts. Other contextual factors, including whether message production for a goal is difficult and whether it requires objectivity or human knowledge, were not found to predict GenAI use. These findings contribute to the theories and research in human–AI interaction and message production, while also informing the future design of human-centered AI.