한국과 중국 대학생의 AI 기반 검색 어시스턴트 지속적 이용의도 결정요인 연구 : IMTBPT-REI 모델을 중심으로 = Determinants of Continuous Use Intention of AI-Based Search Assistants Among Korean and Chinese University Students

국문 초록 (Abstract)

대학생의 일상적인 정보 탐색 과정에서 AI 검색 어시스턴트가 웹 검색 엔 진을 대체하고 있다. 본 연구는 IMTBPT-REI 모델(The Integrated Model of Technology-based Behavior Prediction Theory Reflecting the Environment and Individual Differences; 환경과 개인차가 반영된 기술기반 행동 의도 예측 통합 모델)을 활용하여 한국과 중국 대학생들의 AI 기반 검 색 어시스턴트의 지속적 이용의도 결정요인을 확인해 내고자 했다. 한국 대 학생 295명과 중국 대학생 297명을 대상으로 설문조사를 진행한 결과 IMTBPT-REI 모델은 양국 대학생의 AI 검색 어시스턴트 지속적 이용의도 를 설명해 내는 효과적인 이론적 틀이라는 점이 확인되었다. 두 집단을 통합하여 볼 때, 정보 관련성 동기와 효율성 동기, 정보잡음 회 피 동기는 공통적으로 AI 검색 어시스턴트 지속적 이용의도를 강화하는 핵 심 요인으로 나타났다. 즉 학생들이 AI 기반 검색 어시스턴트를 통해 얻는 정보가 과제와 학습 상황에 정확히 부합하고, 기존 방식보다 빠르고 효율적 으로 문제를 해결해 주며, 불필요한 광고와 중복 정보를 줄여 준다고 인식할 수록 AI 기반 검색 어시스턴트를 지속적으로 이용하려는 의도가 높아졌다. 또한 태도와 주관적 규범 역시 한국과 중국 모두에서 AI 기반 검색 어시스턴트 지속적 이용의도에 정적인 영향을 미쳤다. 즉 AI 기반 검색 어시스턴 트를 전반적으로 긍정적인 학습 도구로 평가하고, 친구와 동료 학생, 교수자 등 중요한 타인이 이를 사용하는 것을 당연하고 바람직한 행위로 여긴다고 인식할수록 지속적 이용의도가 강화되었다. 반면, 인지된 반응 비용은 두 집 단에서 공통적으로 지속적 이용의도에 부적인 영향을 미치는 것으로 나타났 다. 이는 사용 과정에서 요구되는 시간과 노력, 과부하 된 정보 처리와 잠재 적 오류 부담이 크다고 느낄수록, 도구의 효율성과 편의성이 인정되더라도 지속적 사용으로 이어지기 어렵다는 점을 시사한다. 국가별 경로 구조의 차이도 확인되었다. 한국 대학생의 경우 정보 관련성 동기, 효율성 동기, 정보잡음 회피 동기뿐 아니라 문화 이해 동기가 AI 기반 검색 어시스턴트 지속적 이용의도에 직접적인 영향을 미쳤다. 그리고 보안성 동기는 유용성을, 자연어 이해 능력 동기는 용이성을, 멀티모달성 동기는 용 이성과 유용성 및 태도에 간접적인 영향을 미치는 요인으로 작동하였다. 유 용성은 지속적 이용의도를 직접적으로 예측하였고, 용이성은 유용성을 매개 로 간접적인 영향을 미쳤다. 또한 태도와 주관적 규범뿐 아니라 인지된 행위 통제와 집단 규범도 지속적 이용의도에 유의한 정적 영향을 미쳤다. 이는 한 국 대학생의 경우 AI 기반 검색 어시스턴트를 과제 수행에 도움이 되는 유 용한 도구이자, 문화 이해와 멀티모달 학습을 돕는 수단으로 인식하고 있다 는 점을 보여준다. 아울러 한국 대학생의 경우 학습 공동체 내부의 규범이 AI 검색 어시스턴트의 지속적 사용을 뒷받침한다는 점을 시사한다. 중국 대학생의 경우 정보 관련성 동기와 효율성 동기, 정보잡음 회피 동기 외에 보안성 동기가 AI 기반 검색 어시스턴트 지속적 이용의도를 높이는 것 으로 나타났다. 자연어 이해 능력 동기는 용이성과 지속적 이용의도에 정적 인 영향을 미쳤고, 멀티모달성 동기는 유용성과 태도에 정적인 영향을 미쳤 다. 또한 AI 평가와 경쟁 압력, 조직적 지원은 주로 유용성과 용이성을 매개 로 간접적인 영향을 미쳤으며, 자기효능감은 지속적 이용의도에 직접적인 영 향을, 인지된 취약성과 인지된 반응 비용 비용은 용이성과 AI 기반 검색 어 시스턴트 지속적 이용의도에 부적인 영향을 미쳤다. 유용성은 태도에 영향을 미쳤지만, 지속적 이용의도에는 직접적인 영향을 미치지 않았고, 확장된 계획행동이론 변인 가운데서는 태도와 주관적 규범만이 의미 있는 설명력을 나타냈다. 이는 중국 대학생들이 AI 기반 검색 어시스턴트를 유용한 도구로 인식하더라도, 실제 장기 사용 여부는 자기효능감, 보안성과 신뢰성, 비용과 위험 인식 등 보다 구체적인 심리적 판단에 의해 결정된다는 점을 보여준다. 이상의 결과는 AI 기반 검색 어시스턴트가 두 국가 모두에서 정보 관련성 과 효율성, 정보잡음 회피 기능을 중심으로 새로운 탐색 경로로 자리 잡고 있음을 보여 준다. 또한 양 국가에 AI 기반 검색 어시스턴트를 활성화하기 위한 차별화된 전략이 필요하다는 점을 짐작하게 한다. 한국의 경우 국내 플 랫폼을 통한 문화 이해와 멀티모달 표현, 정보잡음 최소화 기능을 강화하고, 교수자 지원과 학습 공동체 내 긍정적 규범 형성을 통해 태도와 집단 규범 을 높이기 위한 전략이 요구된다. 중국의 경우 이용자들에게 보안성과 신뢰 성을 중심으로 차별화된 가치를 제공하고, 이용자의 자기효능감을 높이기 위 한 전략과 함께 반응 비용과 취약성 인식을 낮추기 위한 인터페이스와 교육 설계가 중요하다. 한편, 양국의 AI 리터러시 교육은 AI 기반 검색 어시스턴 트를 정보 탐색의 새로운 표준 경로로 인식하도록 돕고, 학생들이 정보 관련 성과 효율성, 보안성과 비용을 균형 있게 평가할 수 있는 방향으로 설계될 필요가 있다. 적극적으로 활용할 수 있도록 지원하는 방향으로 설계될 필요 가 있다.

다국어 초록 (Multilingual Abstract)

AI-based search assistants are increasingly replacing conventional web search engines in university students’ everyday informationseeking processes. Drawing on the IMTBPT-REI model (The Integrated Model of Technology-based Behavior Prediction Theory Reflecting the Environment and Individual Differences), this study examines the determinants of continuous use intention of AI-based search assistants among Korean and Chinese university students. Survey data were collected from 295 Korean and 297 Chinese students. The results show that the IMTBPT-REI model provides an effective theoretical framework that explains a substantial portion of the variance in continuous use intention in both national samples. When analyzing the two groups collectively, information relevance motivation, efficiency motivation, and information noise avoidance motivation were identified as common core factors strengthening the continuous use intention to use AI search assistants. Specifically, students' intention to continue using these tools increased when they perceived that the information provided by AI assistants precisely aligned with their academic tasks and learning contexts, solved problems more quickly and efficiently than traditional methods, and reduced exposure to unnecessary advertisements and redundant information. Furthermore, attitude and subjective norms exerted a positive influence on the continuous use intention to use AI search assistants in both South Korea and China. This indicates that the intention to continue use was reinforced when students evaluated AI assistants as overall positive learning tools and perceived that significant others—such as friends, peers, and professors—regarded the use of these tools as a natural and desirable behavior. Conversely, perceived response costs were found to have a common negative impact on continuous use intention in both groups. This suggests that if students perceive the required time and effort, information overload, or the burden of potential errors during the usage process to be too high, it becomes difficult for that usage to translate into long-term commitment, even if the tool's efficiency and convenience are acknowledged. The study further identified significant differences in the path structures between the two countries. For South Korean university students, not only did information relevance, efficiency, and information noise avoidance motivations influence continuous use intention, but cultural understanding motivation also exerted a direct effect. Additionally, security motivation predicted perceived usefulness, while natural language understanding (NLU) motivation influenced ease of use. Multimodality motivation acted as an antecedent indirectly affecting ease of use, usefulness, and attitude. While usefulness directly predicted continuous use intention, ease of use exerted an indirect influence mediated by usefulness. Beyond attitude and subjective norms, perceived behavioral control (PBC) and group norms also showed significant positive effects on continuous use intention. These findings suggest that Korean students perceive AI search assistants as useful tools for academic tasks and as instruments for cultural understanding and multimodal learning. Moreover, it implies that norms within the academic community play a critical role in supporting the sustained use of AI search assistants. For Chinese university students, security motivation was a key driver of continuous use intention, alongside the core motivations of information relevance, efficiency, and noise avoidance. Natural language understanding motivation positively affected both ease of use and continuous use intention, while multimodality motivation significantly influenced usefulness and attitude. Factors such as AI evaluation, competitive pressure, and organizational support primarily exerted indirect effects through the mediation of usefulness and ease of use. Notably, self-efficacy had a direct impact on continuous use intention, whereas perceived vulnerability and perceived response costs negatively affected both ease of use and continuous use intention. Although usefulness influenced attitude, it did not directly predict continuous use intention. Among the variables of the Extended Theory of Planned Behavior (ETPB), only attitude and subjective norms demonstrated significant explanatory power. This indicates that while Chinese students recognize the utility of AI search assistants, their long-term commitment is determined by specific psychological assessments, including self-efficacy, security, trust, and the perception of costs and risks. The overall results demonstrate that AI-based search assistants are establishing themselves as a new information-seeking paradigm in both countries, primarily due to their functional advantages in information relevance, efficiency, and noise reduction. Furthermore, the findings highlight the need for differentiated strategies to promote AI search assistant adoption in each nation. In South Korea, strategies should focus on strengthening features related to cultural understanding, multimodal expression, and noise minimization via domestic platforms. Additionally, it is essential to cultivate positive group norms and improve attitudes through institutional support and faculty engagement. In China, service providers should offer differentiated value propositions centered on security and reliability. Strategies to enhance user self-efficacy, combined with interface designs and educational interventions that mitigate perceived response costs and vulnerability, are of paramount importance. Finally, AI literacy education in both countries must be designed to help students recognize AI-based search assistants as the new standard for information seeking. Educational programs should be structured to empower students to evaluate the balance between information relevance, efficiency, security, and costs, ultimately supporting them in actively and effectively leveraging these tools.

목차 (Table of Contents)

• 제1장 서 론 ········································································································· 1
• 제1절 문제제기 ·································································································· 1
• 제2절 연구의 목적 ···························································································· 7
• 제2장 이론적 배경 ·························································································· 11
• 제1절 AI 기반 검색 어시스턴트의 부상 ··················································· 11
• 1. 정보 검색 패러다임의 발전 과정 ······················································· 11
• 2. 정보 검색 패러다임의 전환과 구조적 한계 ····································· 13
• 3. AI 기반 검색 시스템의 단계적 진화 과정 ······································· 19
• 4. AI 기반 검색 어시스턴트의 등장과 쟁점 ········································· 21
• 제2절 한국과 중국의 AI 기반 검색 어시스턴트 활용 환경 ················· 27
• 1. 한국과 중국의 AI 관련 국가전략과 정책 ········································· 27
• 가. 한국의 AI 관련 국가전략과 정책 ················································· 27
• 나. 중국의 AI 관련 국가전략과 정책 ················································· 30
• 2. AI 기반 검색 어시스턴트 관련 생태 구조 ······································· 34
• 가. 한국의 AI 기반 검색 어시스턴트 관련 생태 구조 ··················· 34
• 나. 중국의 AI 기반 검색 어시스턴트 관련 생태 구조 ··················· 38
• 3. 한국과 중국의 AI 기반 검색 어시스턴트 이용자 특징 ················· 41
• 가. 한국의 AI 기반 검색 어시스턴트 이용자 특징 ························· 41
• 나. 중국의 AI 기반 검색 어시스턴트 이용자 특징 ························· 43
• 제3절 기술기반 행동의도 예측을 위한 통합모델의 활용에 관한 논의·········· 46
• 1. 환경 변인(인지된 트렌드, 경쟁 압력, 조직적 지원, 교수자지원)이 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ···················· 46
• 2. 제1 개인차 변인: AI 리터러시 변인이 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ···· 52
• 3. 제2 개인차 변인: PMT 변인이 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ········· 58
• 4. 제3 개인차 변인: AI 기반 검색 어시스턴트의 이용동기가 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ······················ 6
• 5. 기술수용모델 변인이 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ···· 66
• 6. 확장된 계획행동이론 변인이 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향에 관한 논의 ····· 69
• 7. IMTBPT-REI 모델의 활용 가능성에 관한 논의 ··························· 72
• 가. 통합 모델을 활용한 AI 기반 검색 어시스턴트 지속적 이용의도의 예측 ··········· 72
• 나. IMTBPT 모델을 활용한 AI 기반 검색 어시스턴트 지속적 이용의도의 예측 ········ 75
• 다. IMTBPT-REI 모델을 활용한 AI 기반 검색 어시스턴트 지속적 이용의도의 예측 ········ 79
• 제3장 연구방법 ································································································· 83
• 제1절 연구문제 ································································································ 83
• 제2절 연구 대상 ······························································································ 88
• 1. 한국 응답자 특성 ··················································································· 89
• 2. 중국 응답자 특성 ··················································································· 90
• 제3절 주요 변인의 측정 ·············································································· 92
• 1. 환경 변인 ································································································· 92
• 2. 개인차 변인 ····························································································· 93
• 3. AI 기반 검색 어시스턴트 이용동기 ··················································· 96
• 4. 기술수용 모델 변인 ··············································································· 99
• 5. 확장된 계획행동이론 변인 ································································· 100
• 6. 종속변인 ································································································· 101
• 제4절 분석방법 ······························································································ 102
• 제4장 연구결과 ······························································································· 104
• 제1절 연구문제 1의 결과 ············································································ 104
• 1. 한국 대학생들의 AI 기반 검색 어시스턴트 이용동기 ················· 104
• 2. 중국 대학생들의 AI 기반 검색 어시스턴트 이용동기 ················· 110
• 제2절 연구문제 2의 결과 ············································································ 117
• 1. 연구문제 2-1의 결과: IMTBPT-REI 구성 변인이 한국 대학생들의 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향 분석··············· 117
• 2. 연구문제 2-2의 결과: IMTBPT-REI 구성 변인이 중국 대학생들의 AI 기반 검색 어시스턴트 지속적 이용의도에 미치는 영향 분석 ········· 120
• 3. 초기 구조모형 도출을 위한 함의 ····················································· 123
• 제3절 연구문제 3의 결과: 한국 대학생의 AI 기반 검색 어시스턴트 지속적 이용의도 결정요인 ········· 127
• 1. 한국 초기 구조모형의 신뢰도 및 타당도 검증 ····························· 127
• 가. 한국 초기 모형의 신뢰도 및 수렴타당성 평가 ························ 127
• 나. 한국 초기 모형의 판별타당성 평가 ············································ 130
• 다. 한국 초기 구조모형의 적합도 및 설명력 평가 ························ 132
• 2. 한국 초기 구조모형의 구조적 관계 분석 결과 ····························· 133
• 3. 한국 수정 구조모형의 신뢰도 및 타당도 검증 ····························· 139
• 가. 한국 수정 모형의 신뢰도 및 수렴타당성 평가 ······················ 139
• 나. 한국 수정 모형의 판별타당성 평가 ·········································· 140
• 다. 한국 수정 구조모형의 적합도 및 설명력 평가 ······················ 140
• 4. 한국 수정 구조모형의 구조적 관계 분석 결과 ··························· 142
• 제4절 연구문제 4의 결과: 중국 대학생의 AI 기반 검색 어시스턴트 지속적 이용의도 결정요인 ····· 146
• 1. 중국 초기 구조모형의 신뢰도 및 타당도 검증 ··························· 146
• 가. 중국 초기 모형의 신뢰도 및 수렴타당성 평가 ······················ 146
• 나. 중국 초기 모형의 판별타당성 평가 ·········································· 149
• 다. 중국 수정 구조모형의 적합도 및 설명력 평가 ······················ 151
• 2. 중국 초기 구조모형의 구조적 관계 분석 결과 ··························· 152
• 3. 중국 수정 구조모형의 신뢰도 및 타당도 검증 ··························· 157
• 가. 중국 수정 모형의 신뢰도 및 수렴타당성 평가 ······················ 157
• 나. 중국 수정 모형의 판별타당성 평가 ·········································· 158
• 다. 중국 수정 구조모형의 적합도 및 설명력 평가 ······················ 158
• 4. 중국 수정 구조모형의 구조적 관계 분석 결과 ··························· 159
• 제5절 연구문제 5의 결과: 한·중 대학생들의 AI 기반 검색 어시스턴트 지속적 이용의도 결정요인의 공통점과 차이점 ········ 164
• 제5장 결 론 ··································································································· 176
• 제1절 연구결과의 요약 및 논의 ······························································ 176
• 1. IMTBPT-REI 모델의 적합성과 적용 가능성에 대한 논의 ······ 176
• 2. 한국과 중국 대학생의 AI 기반 검색 어시스턴트 지속적 이용의도 경로분석 결과의 공통점과 논의 ···· 178
• 3. 한국과 중국 대학생의 AI 기반 검색 어시스턴트 지속적 이용의도 경로분석 결과의 차이점과 논의 ····· 182
• 제2절 연구의 시사점 및 한계점 ······························································ 190
• 1. 연구의 시사점 ····················································································· 190
• 2. 연구의 한계와 전망 ········································································· 197
• 참고문헌 ··········································································································· 200
• [부록] ················································································································ 253
• 부록 1. AI 기반 검색 어시스턴트 이용동기 측정 문항(사전 인터뷰 기반) ····· 253
• 부록 2. 한국 초기 모형의 측정문항별 VIF 값 ···································· 254
• 부록 3. 초기 모형의 판별타당성 검증 결과(한국): HTMT 기준 ···· 256
• 부록 4. 수정 모형의PLS-SEM 기반 신뢰도 및 수렴타당성 분석 결과(한국) ······· 257
• 부록 5. 한국 수정 모형의 측정문항별 VIF 값 ······································ 259
• 부록 6. 수정 모형의 상관관계 분석 결과 및 AVE 제곱근 분석 결과(한국) ······· 261
• 부록 7. 수정 모형의 판별타당성 검증 결과(한국): HTMT 기준 ······ 262
• 부록 8. 중국 초기 모형의 측정문항별 VIF 값 ······································ 263
• 부록 9. 초기 모형의 판별타당성 검증 결과(중국): HTMT 기준 ······ 265
• 부록 10. 수정 모형의PLS-SEM 기반 신뢰도 및 수렴타당성 분석 결과(중국) ······ 266
• 부록 11. 중국 수정 모형의 측정문항별 VIF 값 ···································· 268
• 부록 12. 수정 모형의 상관관계 분석 결과 및 AVE 제곱근 분석 결과(중국) ······· 270
• 부록 13. 수정 모형의 판별타당성 검증 결과(중국): HTMT 기준 ···· 271
• 부록 14. 한국어 설문지 ··············································································· 272
• 부록 15. 중국어 설문지 ··············································································· 283
• Abstract ············································································································ 293