AUTHOR=Yang Jinhua , Liu Ting , Luo Yiming Taclis , Pang Patrick Cheong-Iao 

TITLE=Deep learning in cultural imagery dissemination: a systematic scoping review of AI-driven visual transmission mechanisms

JOURNAL=Frontiers in Communication

VOLUME=Volume 10 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/communication/articles/10.3389/fcomm.2025.1645168

DOI=10.3389/fcomm.2025.1645168

ISSN=2297-900X

ABSTRACT=BackgroundIn an era of rapid media technology and AI advancement, deep learning (DL)-driven visual images (VI) is emerging as a critical mode of cultural transmission (CT). Despite the growing application of DL in the VI domain, there is a lack of a systematic review that comprehensively explores its transmission pathways, mechanisms of influence, and associated challenges. This study aims to systematically explore the pathways and impacts of DL-driven VI in CT and identify key trends and issues in the field through a systematic scoping review of existing literature.MethodsThis review analyzes 18 studies published between 2015 and 2024. The literature search was conducted across five databases: WOS, ScienceDirect, Scopus, ACM, and A&HCI. The research was undertaken rigorously following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines, ensuring systematic selection, extraction, and analysis of the identified studies.ResultsThe study analyzed the literature from four aspects: transmission pathways, content, technology, and cultural context, identifying three main research areas: (1) the influence mechanisms of AI and social media on cultural transmission; (2) the role of VI in cross-cultural communication; and (3) the application of AI and digital technology in the conservation of Cultural Ecosystem Services (CES). The study finds that AI-driven visual technologies significantly enhance the breadth and impact of CT, particularly through DL algorithms. However, the field faces critical challenges such as algorithmic bias, cultural homogenization, and the reliability of user-generated content.ConclusionBy systematically synthesizing the existing literature, this study provides a theoretical foundation for future research and points to emerging research directions, such as how to use DL to address ethical challenges in cultural communication and explore the differences in the application of DL and VI in different cultural contexts.