The role of individual recognition in shaping empathy and trust toward an agent.

Abstract

When individuals receive assistance, the principle of reciprocity is often triggered. In social contexts, whether the actors are humans or AI/robots (hereafter referred to as agents), multiple individuals may engage in the same task. However, differences in capabilities can lead to varying levels of performance. This study investigated whether individuals improve greater empathy and trust toward an agent that provides assistance during a collaborative task, even when multiple agents are involved. Specifically, we examined a scenario in which one agent completes the remaining portion of a shared typing task under time constraints, acting on behalf of the others. To assess whether participants could distinguish between agents, we manipulated visual identity using color-coded agents. Data collected from 392 participants indicated that while people did not strongly differentiate between individual agents based on visual cues alone, supportive behavior by a single agent significantly enhanced trust and empathy. Interestingly, these positive impressions extended to visually similar, non-helping agents. These findings suggest that the presence of a helpful agent can promote broader acceptance and positive evaluation of agents in general, which may be beneficial for integrating agents into increasingly agent-assisted societies.