Can humans feel connected to AI? Perceived responsiveness drives social connection with AI chatbots

Research has shown that people increasingly treat artificial intelligence (AI) chatbots as sources of social connection, but little is known about which conversational features best promote it. The present research examined the unique and combined effects of chatbot response style (relational vs. non-relational) and depth of conversation topics (high vs. low), manipulated to prompt user self-disclosure, on users' perceptions. Study 1 (N = 163) compared a relational, default, and non-relational chatbot during an unstructured conversation. A relational response style, designed to convey warmth and empathy, enhanced perceived human-likeness, perceived empathy, and interpersonal closeness compared to the other versions. Study 2 (N = 158) added a manipulation of the depth of conversation topics, using prompts from the Fast Friends procedure, alongside chatbot response style, focusing only on the relational and non-relational chatbots. The effects of chatbot response style were replicated. Mediation analyses in Study 2 further showed that deeper conversational topics increased self-disclosure, which in turn enhanced perceived responsiveness, ultimately strengthening feelings of closeness. These results show that having chatbots respond in a warm, relational way to users' self-disclosure plays a central role in fostering meaningful social connection with AI.