Comparing User Perceptions of AI-Generated Short Video Clips and Text Summaries of Academic Papers: Toward the Development of the “Paper 2 Clip” System
We prototyped “Paper 2 Clip,” a generative AI system converting academic papers into short videos. ChatGPT generated narration and Sora prompts, Sora created silent scene clips, then AI narration and subtitles were combined. An online survey of 517 Japanese participants compared 1-minute short videos and text summaries across four papers. While information accuracy was comparable between formats, short videos were rated more favorably on most other items. The mixed-design ANOVA showed significant main effects for format, with no age or media habit interactions. This study contributed to positioning AI-generated academic short videos as innovative science communication methods in higher education.
Authors
- Kayoko Kurita (ORCID: https://orcid.org/0000-0002-6737-2146)
- Hayato Sezaki
- Takashi Goto (ORCID: https://orcid.org/0000-0001-9632-0540)
- Ayako Kurono
- Hideo Kawamata
Institutions
- Kobe International University (JP)
- Iwate University (JP)
- IBM Research - Tokyo (JP)
- Toranomon Hospital (JP)
- Kobe University (JP)
- The University of Tokyo (JP)
Publication Details
- Journal
- Procedia Computer Science
- Published
- 2025-01-01
- DOI
- https://doi.org/10.1016/j.procs.2025.10.006
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- article
- Field-Weighted Citation Impact
- 0.00