AI 추천 설명 유형에 따른 OTT 이용자의 추천 수용 과정 분석: 지각된 투명성, 알고리즘 신뢰 및 정보 만족도의 직렬 매개효과

초록

This study examined how AI recommendation explanation types (no explanation, quantitative, qualitative, and contextual) affect recommendation acceptance intention through a serial mediation path (perceived transparency → algorithm trust → information satisfaction → acceptance intention), as well as the moderating effect of algorithmic literacy, using a four-group between-subjects randomized controlled experiment with 459 OTT users. The explanation groups reported significantly higher perceived transparency and information satisfaction than the control group. With no significant direct effects and only the indirect paths originating from transparency reaching significance, full mediation with transparency as the entry point was confirmed, whereas algorithmic literacy exhibited no moderating effect. This study contributes by extending the Expectation-Confirmation Model (ECM-IT) to the context of OTT recommendation explanations, identifying perceived transparency as the entry point of the serial mediation, and proposing transparency-centered UX design principles.

키워드