Understanding the Dynamics in Deploying AI-Based Content Creation Support Tools in Broadcasting Systems - Benefits, Challenges, and Directions

Research Background and Issues

• Identified Problems or Challenges
  The paper highlights the potential of generative AI in media content creation and the existing challenges, including:

  1. Insufficient integration of generative AI into actual workflows in the broadcasting industry.
  2. Tool development is often technology-driven rather than user-centered.
  3. Low trust in generative AI among broadcasting professionals, coupled with the high complexity of broadcasting workflows, limits the widespread adoption of AI tools.
  4. Legal and ethical issues restrict the scope of AI technology applications.

• Significance of the Problem
  With the proliferation of generative AI, it offers tremendous potential to enhance efficiency and creativity in the content production industry. However, if user needs are not accurately addressed and industry-specific obstacles and challenges are not resolved, generative AI may fail to be effectively adopted and could even negatively impact the industry ecosystem.

• Research Motivation and Related Work
  The broadcasting industry has always been a pioneer in technology adoption. However, unlike previous technologies (e.g., non-linear editing systems), generative AI not only improves efficiency but also automates content creation and supports creativity. Therefore, exploring broadcasting professionals' actual use, perceptions, and potential issues with generative AI tools becomes a highly valuable research direction.

Solution

• Proposed Solution or Method
  The authors developed and studied an AI-based post-production support tool—“AI Editing Assistant”—and conducted in-depth interviews with 37 professionals from the broadcasting industry to investigate the user experience and challenges of generative AI.

• Innovative Aspects

  1. This study comprehensively explores the application of generative AI tools in real-world scenarios specific to broadcasting professionals.
  2. By combining user interviews with practical evaluations of generative AI tools, the study proposes design guidelines and strategies for improving design and user experience.
  3. Special attention is given to collaboration and communication issues across roles and multiple stakeholders.

• Implementation Steps and Key Technologies

  1. Using the “AI Editing Assistant” as a research probe, the study analyzed its five core functions (scene description, audio-to-text search, facial recognition, screen text search, multi-video synchronization).
  2. Conducted 37 in-depth interviews with professionals and extracted key findings through thematic analysis.
  3. Based on the interview results, proposed two sets of design guidelines:
     • User interface and interaction design guidelines for AI tools.
     • Collaboration guidelines for implementing AI tools across multiple stakeholders in broadcasting systems.

Research Outcomes

• Specific Outcomes

  1. Identified significant advantages of AI tools, such as the potential of scene search and video synchronization functions to improve production efficiency.
  2. Proposed a four-stage guideline for AI tool design, including tool introduction, workflow adaptation, mid-term responsibility and compliance, and long-term development goals.
  3. Identified key barriers to the widespread adoption of AI tools, including mismatches between user needs and technical implementation, insufficient communication among multiple stakeholders, and the inertia of existing workflows in the broadcasting industry.

• Advantages Compared to Existing Solutions
  This study focuses specifically on the vertical application of generative AI technology in the broadcasting industry, avoiding the underperformance of general-purpose AI tools in professional environments. Additionally, the research provides highly actionable and targeted design and implementation guidelines.

• Experimental or Evaluation Results
  Through user surveys, differences in user evaluations of various tool functions were identified:

  1. The audio-to-text search function was considered the most useful and user-friendly, but there is room for improvement in its predictive and reliability features.
  2. The multi-video synchronization function performed well in multi-camera environments but was constrained by the complexity of workflows.
  3. The facial recognition function was deemed less practical, failing to meet user needs for extracting contextual information.
  4. The scene description function performed well in terms of speed and efficiency, but its capabilities for recognizing emotions and complex behaviors need further enhancement.

• Limitations and Future Directions

  1. Limitations: The study primarily focused on a single broadcasting company in South Korea, which may limit its generalizability.
  2. Future Directions:
     • Expand the research to include broadcasting systems in diverse cultural and linguistic contexts.
     • Focus on text-to-video generative AI based on large language models (LLMs).
     • Validate and optimize the proposed design guidelines in real production environments.

Conclusion

This study analyzes the application prospects and real-world challenges of generative AI technology in the broadcasting industry, proposing user-centered tool design concepts and multi-stakeholder collaboration implementation guidelines. It provides a clear framework and direction for the future development and application of AI tools in the content production industry. The research emphasizes that technological innovation should not only focus on algorithmic advancements but also prioritize real user needs and usage contexts, establishing a model of innovation that balances practicality and ethical considerations.