Intelligent agents are part of the smartphones, smart speakers, and home robots that millions of us own and interact with regularly. However, little is known about if or how these agents should be embodied. Additionally, as conversations with agents move from heavily structured and transactional to more flexible and, at times, social, embodiment may play a more nuanced and variable role. This paper describes an experimental study that examined the effects of embodiment and quality of information on people’s perceptions of an intelligent agent. We conducted a \(3 \times 2\) between-subjects experiment in which participants completed an escape room-like task with the “help” of an artificial personal assistant that had a robotic embodiment, a virtual embodiment, or no embodiment. Participants were given either helpful or unhelpful information about the task. Findings suggest that people can attribute the quality of loyalty to an agent that belongs to someone else and may work on their behalf, and that in the context of a complex problem-solving situation that involves time pressure, people may prefer an embodied robot or a disembodied voice over a virtually embodied agent.

智能代理是智能手机，智能扬声器和家庭机器人的一部分，我们成千上万的人拥有并定期与之交互。但是，关于是否应该或如何体现这些代理人知之甚少。此外，随着与代理的对话从结构化和事务性的过渡到更加灵活的（有时是社交的），实施可能会发挥更加细微和可变的作用。本文描述了一项实验研究，该研究研究了实施方式和信息质量对人们对智能代理的感知的影响。我们进行了\（3 \ times 2 \）受试者之间的实验，其中参与者在具有机器人实施方式，虚拟实施方式或没有实施方式的人造私人助理的帮助下完成了类似逃生室的任务。给参与者有关任务的有用或无用信息。研究结果表明，人们可以将忠诚度的质量归因于属于他人并且可以代表他们工作的代理人，并且在涉及时间压力的复杂解决问题的情况下，人们可能更喜欢使用嵌入式机器人或在虚拟体现的代理人上体现的声音。