The Role of Shared Mental Models in Driving Knowledge Complementarity: Enhancing Human–AI Team Effectiveness

With the rapid advancement of artificial intelligence (AI), especially the widespread adoption of large language models (e.g., ChatGPT), the role of AI in team collaboration is undergoing profound transformation. Prior research suggests that improvements in team effectiveness largely depend on the knowledge complementarity among team members and the development of shared mental models. However, the underlying mechanisms through which these factors operate in human–AI teams remain insufficiently understood. Grounded in shared mental model theory and transactive memory system theory and informed by a cognitive complementarity framework within human–AI teams, this study develops a cognitive mechanism model for team collaboration. A 2 (team type: human–human vs. human–AI) × 2 (knowledge complementarity: high vs. low) × 2 (team climate: positive vs. negative) between-subjects experimental design was used to examine how knowledge complementarity affects team effectiveness and member satisfaction via shared mental models, with a focus on moderating effects. A total of 128 participants were recruited to complete a collaborative promotional writing task with the theme “modern communication of classical Western artistic ideals.” The participants co-created a promotional text (within 300 words) with either a human teammate or an AI system powered by the DeepSeek architecture. The results indicate that in human–AI teams, high levels of knowledge complementarity significantly enhance both team effectiveness and member satisfaction. Shared mental models serve as a mediating mechanism, and a positive team climate further amplifies these effects. These findings contribute to a deeper understanding of cognitive structures in human–AI interactions and offer theoretical and practical guidance for the deployment of large language models in collaborative tasks.