Authors' Values and Attitudes Towards AI-bridged Scalable Personalization of Creative Language Arts

Title of the Paper

Authors' Values and Attitudes Towards Scalable Personalized Creative Language Arts Bridged by AI

Paper Information

• Subject Area: Human-Computer Interaction (HCI), Generative AI, Creative Writing
• Keywords: Author Control, Creative Language Arts, Creative Writing, Generative AI, Large Language Models, Scalable Personalization

Research Background and Issues

• Problems and Challenges:

  1. Generative AI technologies (e.g., Large Language Models, LLMs) can produce creative texts, transforming the form of Creative Language Arts (CLA) and offering the potential to reflect an author's vision through personalization.
  2. There is a lack of deep understanding of how human authors perceive AI's involvement in the creative process, particularly in scalable personalized CLA (AI-bridged CLA).

• Significance:

  • Historically, CLA has evolved with technological advancements (e.g., from oral traditions to print media). AI offers a potential bridge for creating content that is both personalized and scalable.
  • Understanding authors' perspectives on AI's role is critical for designing AI tools that align with ethical and value-driven principles.

• Research Motivation and Related Work:

  • Existing research primarily focuses on using generative AI to support fixed-format writing, with limited exploration of personalized media.
  • Continuous innovation in CLA requires integrating AI with authors' visions and audience needs.

Solution

• Research Methods and Design:

  • Proposed the concept of AI-bridged CLA, defining AI as a "bridge" between authors and audiences to enable context-aware and interactive modifications of creative content.
  • Developed 16 hypothetical scenarios (based on four dimensions: generation and transformation tasks, author control, and audience interaction) and selected five key scenarios to guide interviews with authors.

• Implementation Details:

  1. Interview Study: Conducted semi-structured interviews with 18 CLA authors from diverse fields, covering eight types of creative writing, including poetry, novels, essays, scripts, and popular song lyrics.
  2. Dynamic Analysis: Performed qualitative analysis on translated interview content to compile an author-audience dynamic map.

• Innovations:

  • Examined how AI influences the creative dynamics of CLA from both author and audience perspectives, analyzing authors' "attachments," "goals," and "expectations" during the creative process to provide theoretical foundations for designing AI support tools.

Research Findings

• Discoveries:

  1. Author-Audience Dynamics:
     • A motivational cycle emerges from authors' satisfaction with the creative process, audience resonance with the artwork, and audience feedback to authors (e.g., responses and financial rewards).
  2. Role of AI:
     • AI can enhance audience engagement through interactive features, increasing content personalization and entertainment value.
     • However, some authors expressed concerns about AI altering the core creativity and diminishing the uniqueness of their work.
  3. Specific Examples:
     • For instance, in the "personalization" scenarios of poetry and novels, AI adapted works to suit audience backgrounds. Some authors found this beneficial for broader understanding.

• Advantages Compared to Existing Solutions:

  • Broader coverage of author interviews (encompassing eight types of creative writing).
  • Explored the trade-offs between AI ethics and the dilution of authorial intent in creative processes.

• Experimental or Evaluation Results:

  • Through coding analysis, two dynamic maps were constructed: one representing author-audience interactions without AI, and the other illustrating the impact of AI-bridged CLA on the motivational cycle.

• Limitations and Future Directions:

  1. Limitations:
     • Data is based on hypothetical scenarios, which may not fully reflect real-world situations.
     • The study did not directly investigate audience perspectives, resulting in a lack of a comprehensive view.
  2. Future Directions:
     • Study the implementation effects of AI-bridged CLA in real-world contexts.
     • Expand to other creative forms such as music and video to explore the dynamic impact of AI-bridged art forms more broadly.