How Do Humans Identify Human-Likeness from Online Text-Based QA Communication?

摘要

This study aims to clarify a person's impressions during the course of a conversation. In conversations between a person and a chatbot, we evaluated the impressions formed on reading text created by a human and text created by a chatbot. In terms of question/answer relevance, it was found that chatbot-created answers could not relate to or meet the expectations of questions and that human-created answers, while not necessarily meeting the expectations of questions, had relevance.