Abstract
The availability of software which can produce convincing yet synthetic media poses both threats and benefits to tertiary education globally. While other forms of synthetic media exist, this study focuses on deepfakes, which are advanced Generative AI (GenAI) fakes of real people. This conceptual paper assesses the current literature on deepfakes across multiple disciplines by conducting an initial scoping review of 182 peer-reviewed publications. The review reveals three major trends: detection methods, malicious applications, and potential benefits, although no specific studies on deepfakes in the tertiary educational context were found. Following a discussion of these trends, this study applies the findings to postulate the major risks and potential mitigation strategies of deepfake technologies in higher education, as well as potential beneficial uses to aid the teaching and learning of both deepfakes and synthetic media. This culminates in the proposal of a research agenda to build a comprehensive, cross-cultural approach to investigate deepfakes in higher education.
Abstract (translated)
软件生产可信且合成媒体的存在既带来了威胁又带来了好处,对全球高等教育。虽然存在其他形式的合成媒体,但本研究重点关注深度伪造(Deepfakes),这是真实人的高级生成人工智能(GenAI)伪造。本文通过对182篇经同行评审的出版物进行初步范围审查,评估了当前关于深度伪造的文献。审查揭示了三个主要趋势:检测方法、恶意应用和潜在好处,尽管在高等教育背景下没有发现关于深度伪造的具体研究。在讨论这些趋势之后,本研究将研究结果应用于提出高等教育中深度伪造技术的主要风险和潜在缓解策略,以及潜在有益于帮助深度伪造和合成媒体的教学和学习。这导致了对在高等教育中研究深度伪造的研究议程的提议,以建立一种全面、跨文化的方法来研究深度伪造在高等教育中的影响。
URL
https://arxiv.org/abs/2404.15601
