On July 3, 2026, the School of Media and Communication (SMC) at Shenzhen University hosted an academic lecture titled “Interpersonal Communication in the Age of AI.” The lecture was delivered by Professor Bo Feng of the University of California, Davis, and moderated by Professor Guang Yang, Acting Dean of SMC. Faculty members and students attended the event.
Drawing on communication theory and empirical research, Professor Feng examined how interactions with artificial intelligence differ from human-to-human communication. She began by exploring the changing role of AI in communication—from a tool for generating content to an interactive agent capable of participating in increasingly interpersonal exchanges. She then introduced key concepts in interpersonal communication and relationship development, laying the groundwork for a closer look at supportive communication in human–AI interaction.
A central question of the lecture was how AI systems can be designed to provide high-quality emotional and informational support. Professor Feng presented two studies comparing human and AI communication. One study found that ChatGPT showed higher levels of engagement and politeness, while humans demonstrated greater reciprocal self-disclosure. Users’ level of self-disclosure had little effect on ChatGPT’s responses but was significantly associated with the quality of human responses. A second study, based on real-world conversations between users and the companion AI Replika that were shared on Reddit, found that humans tended to disclose more deeply and express stronger emotions than the AI. The findings suggest that users may often turn to AI as an outlet for emotional expression, while the AI primarily takes on the role of an affirming conversational partner.
In the final part of the lecture, Professor Feng introduced the idea of “alignment without understanding” to examine AI sycophancy. Unlike human flattery, which is often shaped by social relationships and hierarchies, AI responses can shift readily in response to user input, producing excessive agreement without genuine understanding. She highlighted this tendency as an important challenge for the future design of AI systems. The lecture offered fresh insights into the evolving dynamics of human–AI communication and prompted further discussion about the ethical boundaries of increasingly social and relational AI.
