The purpose of this study was to investigate how perceptions of an agent, such as trust and likability, may change depending on how the virtual agent paraphrases the human user’s utterances during a conversation between a human and a agent. To achieve this goal, an experiment was conducted with a virtual agent. In the experiment, three levels of paraphrasing were prepared: “no paraphrasing,” “simple paraphrasing,” and “paraphrasing with care.” Subjects were asked to rate their trust, likeability, and intention to use before and after the virtual agent paraphrased their speech in conversation, and the results were compared between the three levels of paraphrasing. Subjects were 68 male adults. The results showed that cognitive trust was higher when the virtual agent simply paraphrased than when the agent did not paraphrase. On the other hand, trust, likeability, and intention to use did not increase when the agent paraphrased with care compared to the other levels. The results of this study show that having virtual agents paraphrase appropriately can increase users’ cognitive trust in agents, which may provide design guideline for human-virtual agent interactions.
各種検索
サポート
ヘルプ