Contreras Irene, Hossfeld Samia, de Boer Katharine, Wiedler Jane Thorley, Ghidinelli Monica
AO Education Institute, AO Foundation, Davos, Switzerland.
J CME. 2024 Dec 6;13(1):2434322. doi: 10.1080/28338073.2024.2434322. eCollection 2024.
Producing high-quality and engaging educational videos for continuing medical education (CME) is traditionally time-consuming and costly. Generative AI tools have shown promise in creating synthetic videos that mimic traditional lecture videos. We conducted a comparative analysis of four AI video generation platforms HeyGen, Synthesia, Colossyan, and HourOne using the Kano model. Our analysis revealed that HeyGen met most of our requirements. We created two videos and collected feedback from 25 learners. The feedback indicated that the videos were of good quality, engaging, and well-paced for learning. Only 32% recognised the videos as AI-generated, citing limited facial expressions, hand gestures and monotone vocal expression. Importantly, only 24% considered disclosure of AI-generated content necessary. This research indicates that AI-generated videos can be a viable alternative to traditionally produced educational videos. It offers an efficient, cost-effective solution for producing educational content. Ethical considerations regarding AI content disclosure should be addressed to maintain transparency.
传统上，为继续医学教育（CME）制作高质量且引人入胜的教育视频既耗时又昂贵。生成式人工智能工具在创建模仿传统讲座视频的合成视频方面已展现出潜力。我们使用卡诺模型对四个人工智能视频生成平台HeyGen、Synthesia、Colossyan和HourOne进行了比较分析。我们的分析表明，HeyGen满足了我们的大部分要求。我们制作了两个视频，并收集了25名学习者的反馈。反馈表明，这些视频质量良好、引人入胜且学习节奏把握得很好。只有32%的人将这些视频识别为人工智能生成的，理由是面部表情有限、手势和单调的语音表达。重要的是，只有24%的人认为有必要披露人工智能生成的内容。这项研究表明，人工智能生成的视频可以成为传统制作的教育视频的可行替代方案。它为制作教育内容提供了一种高效、经济有效的解决方案。应解决有关人工智能内容披露的伦理考量，以保持透明度。
