ChatGPT 충성도가 실제 사용에미치는영향 = A Study on the Effect of Loyalty to ChatGPT on the Structural Change of Factors That Decide the Actual Intention to Use

다국어 초록 (Multilingual Abstract)

The GPT (Generative Pre-trained Transformer) model, especially the conversational AI model of ChatGPT, which emerged from the rapid development of artificial intelligence technology in the 2020s, especially in the field of natural language processing (NLP), is being used in various ways, supplemented and developed in the form of conversations with users. This is providing users with a new experience and greatly changing the way existing services are provided. In addition, many companies are making great efforts to enhance flexibility in their organisational structures and create a creative and innovative corporate culture in order to respond to the rapidly changing economic and social structures, and the use of ChatGPT within companies is also showing an explosive increase in this process.
ChatGPT is considered to be able to analyse and explain the effects of applying the IS model because it embodies the characteristics of an information system (IS). As a result, there is a growing need to analyse the benefits of using ChatGPT using IS models and to analyse the impact on user satisfaction, benefits, and organisational performance (DeLone W. H. and McLean, E. R, 1992). Nevertheless, most of the current research on ChatGPT in the social sciences is focused on the benefits of ChatGPT for users or companies (MI-NA CHU, 2023; Hyo-Jung Kim, 2023), and there has been no analysis or research on what changes will occur as its use becomes more familiar and its use spreads. Therefore, it is necessary to analyse how the factors that determine the use of ChatGPT change when ChatGPT users become familiar with its use and form a certain level of loyalty.
The characteristics of the ChatGPT service, which were set as research hypotheses, are found to have a significant positive impact on actual use, except in the case of service quality, and this is also true for analyses using centralised values.
On the other hand, the results of the multiple regression analysis model that considered loyalty as an interaction variable with the moderating variable showed that both loyalty and interaction were not significant.
On the other hand, the moderating effect analysis of Process Macro showed that loyalty has a moderating effect on all characteristic variables, and has a significant interaction with most variables except for perceived usefulness.

목차 (Table of Contents)

• 목 차 ⅰ
• 표 목 차 ⅲ
• 그림목차 ⅳ
• Ⅰ. 서론 1
• 제 1 절 연구 목적 1
• 제 2 절 주요내용 및 구성 4
• Ⅱ. 선행연구 및 이론적 배경 7
• 제 1 절 ChatGPT의 개념 및 변화과정 7
• 제 2 절 이론적 배경 12
• Ⅲ. 연구모형 및 가설 19
• 제 1 절 연구모형 및 가설의 설정 19
• 제 2 절 변수의 조작적 정의 27
• Ⅳ. 실증분석결과 32
• 제 1 절 기초통계량 분석 32
• 제 2 절 타당성 및 신뢰도　분석 34
• 제 3 절 연구가설의 검정결과 38
• 제 4 절 실증분석 결과의 종합 44
• Ⅴ. 결 론 45
• 제 1 절 연구 요약 45
• 제 2 절 연구의 시사점 46
• 참고문헌 47
• ABSTRACT 58
• 부록 [설문지] 60