Thinking, fast, and artificial: processing fluency in AI-mediated relational maintenance planning and its impact on relationship perceptions

Previous research suggests that the ease (i.e., processing fluency) of generating future relational plans leads to perceptions of greater relational closeness and commitment. Can generative artificial intelligence (GenAI), when used to help users generate relational maintenance plans, increase fluency and thereby enhance users’ perceived closeness and commitment in friendships? Does the anthropomorphism of GenAI matter? Drawing upon prior work on processing fluency and extended cognition, we conducted a 2 (plan quantity: generating three vs. 10 plans) × 3 (agency: internal retrieval vs. anthropomorphic AI vs. nonanthropomorphic AI) online experiment (N = 1,397). After establishing the ease-of-retrieval effects on friendship closeness and commitment during internal retrieval, we found that both nonanthropomorphic and anthropomorphic AI (vs. internal retrieval) increased fluency, through which they further enhanced perceived friendship closeness and commitment. Despite taking a longer time to operate, the anthropomorphic AI induced a greater level of fluency than the nonanthropomorphic AI. We discuss theoretical implications for the ease-of-retrieval effect in interpersonal planning, AI-mediated communication, and human–machine communication.