플랫폼 서비스에서의 검색에 대한 이용자 신뢰도 및 만족도에 영향을 미치는 요인 분석 (An Empirical Analysis of User Trust and Satisfaction in AI-Driven Platform Search Services)

서지정보

· 발행기관 : 한국정보사회학회
· 수록지 정보 : 정보사회와 미디어 / 26권 / 1호 / 231 ~ 267페이지
· 저자명 : 임세실, 강승규, 박유리

초록

본 연구는 AI 알고리즘 기반 검색 서비스가 제공되는 다양한 플랫폼 환경에서 이용자의 검색 결과 신뢰도와 검색 기능 만족도에 영향을 미치는 요인을 실증적으로 분석하였다. 분석대상 플랫폼 유형에는 검색엔진, 전자상거래, 소셜네트워크서비스, 개인간 중고거래, 여행 및 숙박 예약, 동영상 공유 및 시청, 음식 주문 및 배달, 그리고 생성형 AI 챗봇 등 여덟 가지 서비스를 포함하였고, 2024년 12월에 20~69세 전국 성인 1,035명을 대상으로 시행한 온라인 설문조사 데이터를 다변량 다중회귀분석모델을 통해 분석하였다. 분석 결과, 이용자가 검색에서 중시하는 가치와 알고리즘에 대한 이해도와 인식 수준이 플랫폼 서비스 전반에서 검색 결과에 대한 신뢰도와 검색 기능의 만족도에 유의한 영향을 미치는 것으로 나타났다. 특히 검색 결과의 품질뿐 아니라 결과 생성 과정의 공정성과 투명성에 대한 기대가 충족될 때 검색 결과 신뢰도와 검색 기능 만족도가 높아지는 반면, 알고리즘 변화에 대한 사전고지 부족이나 검색 결과의 상업적 영향에 대한 우려가 있을 경우 신뢰도와 만족도가 저해되는 것으로 나타났다. 이러한 결과는 플랫폼 서비스에서 검색에 대한 신뢰도와 만족도를 높이기 위해서는 플랫폼 서비스에서의 검색 투명성 제고와 검색·추천 시스템의 작동원리 등에 대한 디지털 리터러시 교육이 필요함을 시사한다.

영어초록

Online information has grown exponentially in recent years, making search functions a critical gateway for accessing information and services (Carroll, 2014). Meanwhile, platform-based business models have expanded across diverse domains and now embed search features into a wide range of online services, altering how users seek information and make decisions. Whereas dedicated search engines (e.g., Google, Naver) once dominated information retrieval, users today often search within various platform environments – including e-commerce sites, social media, and video-sharing platforms – rather than relying solely on traditional search engines (Park et al., 2024). Search functionality has thus become integral to services from online shopping and travel booking to food delivery, underscoring its central role in the platform ecosystem. In parallel, the rise of artificial intelligence has fueled highly personalized recommendation systems. Unlike search, which requires active user queries, recommendation algorithms proactively present content or products to users, potentially reducing effort but raising concerns if recommendations are misaligned with user preferences. Both search and recommendation features now shape user choices on platforms. Despite their growing importance, however, most previous research on platform search and recommendation services has focused on technical aspects of algorithms, with relatively little attention to user-centric outcomes like perceived credibility or satisfaction (Park et al., 2008).
To address this gap, the present study empirically examines user trust and satisfaction with AI-driven search and recommendation services across eight major types of online platforms. It focuses on two key user-centric outcomes – search credibility (trust in the search results returned) and satisfaction with the platform’s search functionality – across environments ranging from search engines and e-commerce sites to social media, peer-to-peer marketplaces, travel and lodging portals, video-sharing services, food delivery apps, and generative AI chatbot services. This broad scope captures a diverse range of platform contexts where algorithmic search and recommendation features are integral.
Key predictors of trust and satisfaction include user characteristics (e.g., age, education, frequency of search use, and whether default search settings are adjusted), the primary aspect of search the user prioritizes (such as accuracy, speed, personalization, or transparency of result ranking), user understanding of AI recommendation algorithms (algorithmic literacy), and user perceptions of algorithmic transparency and fairness. The latter encompasses the importance of measures such as platforms explaining their ranking criteria, providing advance notice of algorithm changes, clearly labeling sponsored content, and explaining how user data are used – reflecting concerns about algorithmic transparency and accountability (Shin et al., 2022).
Data for the study were collected through an online survey conducted in December 2024 with a sample of 1,035 adults in South Korea aged 20–69. Respondents reported their usage of the aforementioned platform services and evaluated both the trustworthiness of search results and their satisfaction with the search function for each platform they used. These two outcome measures (perceived search trust and search satisfaction) were captured on five-point Likert scales for each platform type.
To ensure consistent comparison across platform types, the analysis focused on the 682 respondents who had experience with all eight services. Trust and satisfaction ratings differed by platform: participants reported the highest search trust and satisfaction for search engines and online shopping services, and the lowest for peer-to-peer marketplaces. A multivariate multiple regression was then applied to assess the combined effects of all predictors on search result trust and search satisfaction across platforms.
The regression results identified several significant predictors of users’ trust in search results. One key finding was that a user’s top priority in using search influenced their trust: users who placed the highest importance on convenience-oriented features (e.g., quick results or helpful suggestions) tended to report higher trust in the search results, while those who prioritized transparency were less trusting. This suggests that when transparency expectations are unmet, trust in search results declines. Moreover, greater knowledge of how algorithms work was associated with higher perceived credibility of search results. Users who strongly valued algorithmic transparency measures – wanting clear explanations of ranking criteria or advance notice of changes – generally had higher trust, whereas strong concern about advertising bias corresponded with lower trust. Trust in a platform’s search service is thus bolstered when the process is perceived as transparent and fair, and undermined by suspicions of opaque or biased algorithms.
The drivers of user satisfaction largely mirrored those for trust. Platforms that delivered well on the attribute a user prioritized (for example, speed or personalization) yielded higher satisfaction for that user, whereas users who emphasized transparency were less satisfied with platforms perceived to lack transparency. General algorithmic literacy did not significantly affect overall satisfaction. Transparency perceptions similarly influenced satisfaction: users who attached high importance to clear explanations of ranking criteria or to advance notice of algorithm changes reported higher satisfaction, whereas those highly sensitive to the presence of sponsored content reported lower satisfaction.
Overall, both trust and satisfaction depended not only on the perceived quality of search results but also on perceptions of fairness and transparency in how results are generated. Trust was more sensitive to transparency-related factors and user knowledge, whereas satisfaction was more closely tied to each platform’s fulfillment of practical user needs. These findings highlight the importance of both algorithmic transparency and user literacy. Platform providers should strive to improve transparency and perceived fairness (e.g., by clearly labeling sponsored results, explaining key ranking factors, and informing users of major algorithm changes), while parallel efforts to enhance users’ algorithm literacy would enable individuals to better understand and trust these systems. Such efforts can reinforce the trust relationship between users and platforms and foster more positive user experiences. These implications also align with emerging policy frameworks that seek to mandate greater transparency and accountability in platform algorithms.

참고자료

· 없음