Would You Like to Visit My World? Cultivating Perceived Equality in Human-Agent Interaction via Observable Social Life Spaces

Abstract

Most AI agents remain confined to an instrumental "command-execution" model, resulting in unequal, one-sided interactions. While recent works attempt to build relationships through hidden memory backends, these invisible processes often fail to break the instrumental bias. In this paper, we argue that true relational equality requires agents to have an independent, observable existence. We introduce the \textit{Observable Life Spaces} paradigm, where agents inhabit a continuous virtual environment, engage in daily activities, and form social relationships that users can directly observe. Through a mixed-methods study ($N=24$), we demonstrate that only when agents are endowed with a socialized life space that is visually observable to humans can the perceived equality during interaction be significantly enhanced ($p = 0.015$). Our findings suggest that visually representing an agent's social life space can effectively shift the human-agent dynamic from a purely instrumental relationship to one characterized by perceived equality.

Figures (2)

• Figure 1: System architecture of the Observable Social Life Spaces for Interactive Agents. In the lower panel, the system is inherently grounded in four sociological design principles: Social Identity, Relationship, Environment, and Time Equality. The middle section illustrates how the agent continuously accumulates experiences via a Dual-Track Memory Mechanism, processing both direct user interactions and autonomous life events across two conditions (Unobservable and Observable Modes). Note that the visual depictions of these two modes are illustrative schematics designed to concisely convey their conceptual equivalence, rather than literal interface screenshots. Finally, the upper panel presents an illustration of the continuous 2D environment. It highlights the core operational modules, including behavior planning, path execution, and social triggers, which collectively endow the agent with an autonomous existence. By unifying these parallel tracks, we shift the relational dynamic from purely instrumental utilization toward a sustained, equal partnership.
• Figure 2: Four progressive application scenarios for the Observable Social Life Spaces paradigm. The upper left panel visualizes the life spaces of agents during both working and rest hours. The upper right panel illustrates the integration of life spaces across multiple mobile devices to enable terminal interconnection. The lower left panel depicts physically embodied agents remaining active in their virtual life spaces during idle state. The lower right panel demonstrates the unconstrained action flexibility of agents independent of physical rules in both VR and AR environments.