Making AI-Enhanced Videos: Analyzing Generative AI Use Cases in YouTube Content Creation

Source: Making AI-Enhanced Videos: Analyzing Generative AI Use Cases in YouTube Content Creation · arXiv Authors: Torin Anderson, Shuo Niu Licence: CC-BY-NC-SA-4.0 — http://creativecommons.org/licenses/by-nc-sa/4.0/

Conference: Extended Abstracts of the CHI Conference on Human Factors in Computing Systems; April 26-May 1, 2025; Yokohama, JapanExtended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25), April 26-May 1, 2025, Yokohama, JapanDOI: 10.1145/3706599.3719991ISBN: 979-8-4007-1395-8/2025/04CCS: Human-centered computing Empirical studies in collaborative and social computingCCS: Human-centered computing Collaborative and social computing theory, concepts and paradigms

Abstract

Generative AI (GenAI) tools enhance social media video creation by streamlining tasks such as scriptwriting, visual and audio generation, and editing. These tools enable the creation of new content, including text, images, audio, and video, with platforms like ChatGPT and MidJourney becoming increasingly popular among YouTube creators. Despite their growing adoption, knowledge of their specific use cases across the video production process remains limited. This study analyzes 274 YouTube how-to videos to explore GenAI’s role in planning, production, editing, and uploading. The findings reveal that YouTubers use GenAI to identify topics, generate scripts, create prompts, and produce visual and audio materials. Additionally, GenAI supports editing tasks like upscaling visuals and reformatting content while also suggesting titles and subtitles. Based on these findings, we discuss future directions for incorporating GenAI to support various video creation tasks.

Keywords:

Generative AI; YouTube; video; content creator

1. Introduction

Generative AI (GenAI) is “artificial intelligence systems that can create new content, such as text, images, audio, or video, rather than just analyzing or acting on existing data” (Gozalo-Brizuela and Garrido-Merchan, 2023). Over the years, the applications of these tools have significantly expanded, including crafting digital content such as creative writing (Lee et al., 2024), short videos (Wang et al., 2024), AI-generated images (Mahdavi Goloujeh et al., 2024), and music composition (Louie et al., 2020). With the emergence of tools like ChatGPT, Sora, MidJourney, and Leonardo.AI, many YouTube content creators have begun incorporating these tools into their video production. On video-sharing platforms (Bartolome and Niu, 2023), creators are leveraging GenAI tools to streamline traditionally time-consuming steps, such as writing scripts or creating visual and audio materials (Hoose and Rosenbohm, 2024; Fancourt et al., 2020). For instance, researchers have observed that YouTubers utilize GenAI tools to create videos on artistic, marketing, knowledge-sharing, and entertainment topics (Lyu et al., 2024). Despite the rapid adoption of GenAI tools, there remains limited understanding of the specific use cases of these tools in video creation.

Recent research on video creation has explored the challenges inherent in common creative practices and the potential solutions offered by AI-driven approaches (Choi et al., 2023; Kim and Kim, 2024). During the Planning Phase, YouTubers often struggle with finding inspiration and selecting relevant topics. In the Performance Phase, creators must dedicate time to recording videos and narrations. Subsequently, the Editing Phase requires them to refine their video materials to enhance the overall viewing experience. Finally, during the Uploading Phase, creators craft video titles and descriptions aimed at optimizing viewer engagement. These tasks are frequently accompanied by challenges such as generating new ideas, fine-tuning video content, and managing the pressure to produce high-quality, engaging material that attracts audiences and garners popularity (Kim and Kim, 2024; Choi et al., 2023). While GenAI has demonstrated potential in addressing these challenges (Lyu et al., 2024), a comprehensive understanding of its use cases within this context remains limited.

To bridge this gap, we conduct a preliminary analysis of YouTube how-to videos. How-to videos feature amateur narrators explaining knowledge or teaching skills (Utz and Wolfers, 2022). These videos exemplify YouTube’s participatory culture, where peers share skills and enable others to learn, thereby advancing collective knowledge within the community (Chau, 2010). Analyzing how-to videos allows us to uncover methods, tips, explanations, descriptions, and conclusions regarding YouTubers’ experiences and knowledge of using GenAI tools (Yang et al., 2023). Our analysis includes the development of a conceptual framework (Figure 1) that illustrates the applications of GenAI in the video creation process. Specifically, we propose a categorization of key GenAI use cases to highlight its roles across various stages of video production, thereby inspiring future designs and the development of GenAI-powered tools to support content creators.

We collected and analyzed 274 videos where creators demonstrated the use cases of GenAI tools in video production. Through thematic analysis, we annotated the use cases of GenAI across the planning, production, editing, and uploading phases. As illustrated in Figure 1, our findings reveal that during the planning phase, YouTubers use GenAI to identify topics and generate video scripts. In the production phase, GenAI is employed to create prompts and produce visual and audio materials. During editing, GenAI tools are used to upscale visuals, resolve video issues, and reformat content. Finally, in the uploading phase, creators rely on GenAI to suggest titles and add subtitles. These applications illustrate how the creator community is leveraging GenAI in practice to streamline video production workflows and reduce the creative labor (Hoose and Rosenbohm, 2024).

Figure 1. A conceptual framework illustrating GenAI use at different phases of video creation.

2. Method

The YouTube videos included in our study were retrieved using the YouTube Data API^[1] on March 26, 2024. The search uses the query “How to (edit OR generate OR create OR make OR use) AI (text OR image OR audio OR video OR animation OR content)”, which incorporates multiple operations referencing the use of AI and various types of content. This query was set to search videos posted between September 1, 2023, and February 29, 2024, yielding six months of data. The YouTube search process resulted in a total of 3,854 videos. Subsequently, videos were programmatically excluded if they did not have closed captions, were not in English, or had a duration shorter than 470 seconds or longer than 1,372 seconds (considered outliers in duration). After applying these filters, 1,046 videos remained.

To identify GenAI use cases, we randomly selected 200 videos to analyze overall themes. However, we observed that some videos were not how-to videos (e.g., advertisement or news about GenAI), did not involve GenAI, or were unrelated to GenAI in video creation. To ensure relevance, three researchers manually reviewed all videos using three criteria: (1) the video must demonstrate clear steps for using GenAI tools, (2) the tools featured must have generative AI capabilities (e.g., ChatGPT, MidJourney), excluding those focused solely on clipping or editing, and (3) the final output must be a video. A video was kept if at least two of the three annotators agreed on its inclusion. This process resulted in a final dataset of 274 videos for analysis.

To facilitate annotation, we used ChatGPT to segment video transcripts into small clips. The clip is examined individually to mark GenAI use cases. The prompt to segment videos was: “Split the closed captions into video segments based on coherence between consecutive texts in the closed captions. Each YouTube video segment should include timestamps in the format of ‘HH:MM:SS