A preliminary design framework for human-friendly AI interaction: grounded in anthropomorphism

People converse with AI as they would with friends in daily life, collaborate with it, and form personalized relationships. In response to these emerging forms of interaction, this study clarifies the concept of Human-Friendly Interaction with conversational AI and develops the Human-Friendly AI Interaction Design Framework (HFAI-DF). This study positions anthropomorphism, a foundational phenomenon in human–AI interaction, as a theoretical bridge and reviews literature on anthropomorphism theory and related research on human cognition. On this basis, this study interpretively derives and organizes how anthropomorphism-inducing stimuli can be applied to AI agents, systematizing them into 30 AI behavioral items that enable human-friendly interaction. Subsequently, a survey was conducted and exploratory factor analysis performed to identify six dimensions as a preliminary structure for a design framework for AI’s human-friendly interaction. These dimensions are: Self-Aware Interaction, User-Adaptive Responsiveness Interaction, Context-Aware Reasoning Interaction, Socio-Emotional Interaction, Creative Collaboration Interaction, and Proactive Assistance Interaction. Within the evolving AI research landscape focused on interaction, this study extends the human-centered AI concept toward human-friendly interaction that operates in an organic manner comparable to human–human conversation, grounded in an understanding of human cognitive functioning. The proposed framework provides both AI providers and designers with an initial structure for shaping and improving the quality of human-friendly interaction.