LLM기반AI 서비스사용자경험비교연구 : ChatGPT와 Clova X를 중심으로

Abstract

n the first half of 2023, Large Language Model (LLM) based on AI services have explosively grown and now influencing various aspects of society, including education policy, corporate practices, and the cultural utilization of technology. Consequently, there is a demand for the evaluation and improvement of user experiences tailored to LLM service products. This research compares two LLM services (ChatGPT 3.5 and Naver Clova X) across five themes and their respective sub-elements: 'Feedback and Articulation,' 'Intuitive Design,' 'Context and Purpose,' 'Engagement and Ethics,' and 'Naturalness.' An empirical study involved conducting online surveys with 80 domestic respondents, who were selected through pre-survey system. Then there were in-depth interviews with 8 participants among them. The results were analyzed using radial charts and tables for each theme and summarized with in-depth research comments, resulting in the creation of a User Experience Journey Map. The research revealed that the utility and accuracy of responses are more important than language proficiency. Both services needed to ensure the safety of personal information use, particularly in the aspect of 'Presence and Ethics'. Furthermore, various discussion points related to different usage stages (access and login, usage, re-access, and re-use) were raised to establish intimacy with users. This study aimed to provide user experience measurements specific to LLM services, along with concrete improvement suggestions, and to pave the way for various subsequent studies in this direction.