When AI pushes back: The impact of AI dissent on user knowledge innovation in online knowledge communities
摘要
With the rapid integration of artificial intelligence (AI) into online knowledge communities (OKCs), understanding how AI feedback style shapes users’ cognitive responses and knowledge-related behaviors has become increasingly important. This study examines how AI dissent—AI-generated feedback that challenges user viewpoints—affects user knowledge innovation within OKC interactions. Grounded in the Stimulus-Organism-Response (SOR) framework, we propose a serial mediation process in which AI dissent functions as a cognitive stimulus that elicits cognitive dissonance, thereby enhancing cognitive flexibility and ultimately fostering knowledge innovation. In addition, we investigate the moderating role of AI anthropomorphism in this process. Across two experimental studies and a qualitative inquiry, the findings show that AI dissent significantly promotes user knowledge innovation through sequential effects on dissonance and flexibility. Importantly, this effect varies by anthropomorphism level. This study advances theoretical understanding of AI communication behaviors in OKCs, enriches research on anthropomorphism and cognitive dissonance in user-AI knowledge exchange, and offers actionable design implications for intelligent knowledge systems aiming to cultivate user learning and innovation.