Proactive Interaction of Artificial Intelligence Agents in Intelligent Assistive Systems: Mechanisms and Impacts

The evolution of artificial intelligence (AI) has driven AI Agents to shift from passive response to proactive interaction, a critical advancement for optimizing user experience and operational efficiency. Current research predominantly emphasizes functional implementation, while studies on user attention and related task performance of proactive interaction mechanisms remain relatively scarce. This research systematically examines three core characteristics of AI Agent proactive interaction (environmental perception, decision support, and task execution) through an integrated methodology combining eye-tracking, Information Processing Efficiency Index, and Self-Assessment Manikin scales. Key findings reveal: (1) Proactive environmental perception with visual-textual cues significantly enhances user attention and task efficiency; (2) Provision of decision-making rationales strengthens user trust and satisfaction; (3) Proactive information delivery improves both user satisfaction and information processing efficiency. As the first study to jointly analyze AI Agent interaction mechanisms from psychological and cognitive perspectives, this work establishes theoretical foundations and actionable guidelines for designing intelligent assistive systems.