PaperTok: Exploring the Use of Generative AI for Creating Short-form Videos for Research Communication

TL;DR

This work tackles the gap in researchers' ability to disseminate scholarly findings via short-form video by introducing PaperTok, an end-to-end AI-assisted system that converts papers into engaging videos through a human-in-the-loop workflow. Using a formative study with science communicators and a mixed-methods evaluation (N=18 researchers; N=100 audience), the authors demonstrate that PaperTok can produce videos that are more engaging while preserving informational value, compared to existing PDF-to-video tools. Key insights reveal a need for fine-grained control, credible signaling, and explicit authorial attribution to sustain trust in AI-assisted science communication. The study discusses design implications for future generative tools that balance automation with expert oversight to support responsible and effective science outreach.

Abstract

The dissemination of scholarly research is critical, yet researchers often lack the time and skills to create engaging content for popular media such as short-form videos. To address this gap, we explore the use of generative AI to help researchers transform their academic papers into accessible video content. Informed by a formative study with science communicators and content creators (N=8), we designed PaperTok, an end-to-end system that automates the initial creative labor by generating script options and corresponding audiovisual content from a source paper. Researchers can then refine based on their preferences with further prompting. A mixed-methods user study (N=18) and crowdsourced evaluation (N=100) demonstrate that PaperTok's workflow can help researchers create engaging and informative short-form videos. We also identified the need for more fine-grained controls in the creation process. To this end, we offer implications for future generative tools that support science outreach.

Figures (7)

• Figure 1: Key screens of PaperTok
• Figure 2: Overview of the PaperTok system's technical pipeline and the user interaction
• Figure 3: Screenshots from videos made using PDFtoBrainrot, SciSpace, and PaperTok based on the artifact contribution paper kim2025amuse
• Figure 4: All participants' ($N = 118$) average ratings across 11 evaluation dimensions for each video type (PDFtoBrainrot, SciSpace, and PaperTok) on our 5-point Likert scale in vertical three-column bar chart format. Significance levels from our mixed-effects model estimated means analysis are denoted above between each video type. Bars indicate standard error.
• Figure 5: Audience members' and researchers' responses to "On a scale of 1 to 5, how much do you agree with this statement: I trust the use of AI for science communication." Audience members were significantly more trusting of AI use in science communication ($p < .05$) compared to researchers. Bars indicate standard error.