정보격차가 인공지능 서비스 이용 경험에 미치는 영향 : 접근ㆍ역량ㆍ활용 격차의 매개 효과와 취약계층 간 이질성 분석
The Effect of Digital Divide on AI Service Usage Experience : Mediating Role of Access, Skill, and Usage Gaps with Heterogeneity Analysis across Vulnerable Groups
This study draws on microdata from the 2023 NIA Digital Information Gap Survey (N = 17,300) to examine what really drives the gap in AI service adoption. Hierarchical binary logistic regression was used alongside bootstrap- based serial mediation analysis. Among the three dimensions of the digital divide, digital skill turned out to be by far the strongest predictor of AI usage (OR = 1.887 in Block 3), dwarfing the effects of access (OR = 1.143). When usage diversity was added in Block 4, skill remained dominant (OR = 1.708) while usage also showed a significant positive effect (OR = 1.418). The serial mediation path running from access through skill to AI usage was statistically significant (β = 0.02988, 95% CI [0.02522, 0.03466]), which points to skill as the primary bottleneck in the pathway from access to AI adoption. Splitting the sample by population group revealed pronounced inter-group variation : the AI adoption gap was widest among agricultural workers and older adults, while marriage immigrants recorded AI usage rates that, contrary to initial expectations, exceeded those of the general population. On balance, these results suggest that investing in digital competency education is likely to yield greater returns for closing the nascent AI divide than continuing to channel resources primarily into device-distribution programmes. However, the cross-sectional design precludes strict causal inference, the skill measure does not capture AI-specific competencies, and the OLS-based mediation estimates may involve approximation error given the binary outcome. Future research should address these limitations through longitudinal data and AI-tailored literacy instruments.
목차
Abstract 1. 서론 1.1 연구배경 1.2 연구 목적 및 필요성 2. 이론적 배경 2.1 디지털 정보격차의 다차원 위계 모형 2.2 AI 격차 : 디지털 격차의 새로운 층위 2.3 취약계층의 디지털 격차와 이질성 2.4 연구 모형 및 가설 3. 연구 방법 3.1 데이터 및 분석 대상 3.2 변수 구성 3.3 분석 방법 4. 분석 결과 4.1 집단별 AI 서비스 이용 경험률 및 격차 지수 4.2 위계적 로지스틱 회귀분석 결과 (H1 검증) 4.3 연쇄 매개 효과 분석 결과 (H2 검증) 4.4 집단별 로지스틱 회귀 비교 (H3 검증) 4.5 집단별 AI 인식 비교 5. 결론 및 정책 제언 5.1 연구 결과 요약 5.2 정책적 제언 5.3 연구의 한계 및 향후 과제 References
키워드
Digital DivideAI Service UsageDigital SkillVulnerable GroupsSerial MediationAI Divide
저자
이종서 [ Jong-Seo Lee | Ph.D. candidate, Seoul National University of Science and Technology, Graduate School of IT Policy ]