User Experience Research on the Differences in Linguistic Representation of Request about Informal Language from AI Assistant

Abstract

Background This research is focused on situations before artificial intelligence (AI) is familiar with users through a cumulative conversion. This research aims to find the most positive user experience relevant in the case of speech on a situation when AI asks the user to converse through the informal word ‘Banmal’, which is a culture spoken in Korea. Methods For the test, the language was sorted not only by honorific, but also informal words based on the perspective of user experience (UX) writing. The test also divided language through question and declarative markers. The prototype was designed as four cases, which were possible to indirectly experience the request from AI, based on the four factors as above. After using the prototype, an experiment was conducted to evaluate the user experience and service factors on a 5-point Likert-type scale. Results Case A, which is based on a polite word and a question, was the most positive in the user experience. The gender difference is meaningful as a regulation effect on the experiment. In the male group, the difference between Case B, which is based on a polite word and an assertive sentence and Case C, which is constructed by an informal word and a question, had a meaningless effect on the factors of usability, trust, intimacy, and satisfaction. On the other hand, in the female group, the difference between Case D, which is based on informal word and declarative sentence and Case C, had a meaningless effect on the same factors as above. Conclusions This research asserts that the tone from AI based on an informal word and a question to the user before AI asks the user, results in the most positive experience. AI, which is a service provider from the perspective of stakeholders, is judged that the use of honorifics is possible from the perspective of users as customers will be positive for users.