Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
119 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
43 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

SARD: A Human-AI Collaborative Story Generation (2403.01575v1)

Published 3 Mar 2024 in cs.HC and cs.AI

Abstract: Generative artificial intelligence (GenAI) has ushered in a new era for storytellers, providing a powerful tool to ignite creativity and explore uncharted narrative territories. As technology continues to advance, the synergy between human creativity and AI-generated content holds the potential to redefine the landscape of storytelling. In this work, we propose SARD, a drag-and-drop visual interface for generating a multi-chapter story using LLMs. Our evaluation of the usability of SARD and its creativity support shows that while node-based visualization of the narrative may help writers build a mental model, it exerts unnecessary mental overhead to the writer and becomes a source of distraction as the story becomes more elaborated. We also found that AI generates stories that are less lexically diverse, irrespective of the complexity of the story. We identified some patterns and limitations of our tool that can guide the development of future human-AI co-writing tools.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (25)
  1. A survey on story generation techniques for authoring computational narratives. IEEE Transactions on Computational Intelligence and AI in Games, 9(3):239–253, 2016.
  2. Design space for intelligent and interactive writing assistants. In Proceedings of the ACM CHI Conference on Human Factors in Computing Systems. ACM, 2024.
  3. Help me write a poem: Instruction tuning as a vehicle for collaborative poetry writing. arXiv preprint arXiv:2210.13669, 2022.
  4. Automatic novel writing: A status report. Technical report, University of Wisconsin-Madison Department of Computer Sciences, 1973.
  5. Natlie Dehn. Story generation after tale-spin. In IJCAI, volume 81, pages 16–18, 1981.
  6. Hierarchical neural story generation. arXiv preprint arXiv:1805.04833, 2018.
  7. Story centaur: Large language model few shot learning as a creative writing tool. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 244–256, 2021.
  8. Wordcraft: story writing with large language models. In 27th International Conference on Intelligent User Interfaces, pages 841–852, 2022.
  9. Ai as an active writer: Interaction strategies with generated text in human-ai collaborative fiction writing. In Joint Proceedings of the ACM IUI Workshops, volume 10. CEUR-WS Team, 2022a.
  10. How novelists use generative language models: An exploratory user study. In HAI-GEN+ user2agent@ IUI, 2020.
  11. Buncho: ai supported story co-creation via unsupervised multitask learning to increase writers’ creativity in japanese. In Extended Abstracts of the 2021 CHI Conference on Human Factors in Computing Systems, pages 1–10, 2021.
  12. Vishakh Padmakumar and He He. Does writing with language models reduce content diversity? arXiv preprint arXiv:2309.05196, 2023.
  13. Re3: Generating longer stories with recursive reprompting and revision. arXiv preprint arXiv:2210.06774, 2022b.
  14. Co-writing screenplays and theatre scripts with language models: Evaluation by industry professionals. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pages 1–34, 2023.
  15. Creative writing with a machine in the loop: Case studies on slogans and stories. In 23rd International Conference on Intelligent User Interfaces, pages 329–340, 2018.
  16. Stylized story generation with style-guided planning. arXiv preprint arXiv:2105.08625, 2021.
  17. Learning to generate text in arbitrary writing styles. arXiv preprint arXiv:2312.17242, 2023.
  18. Why johnny can’t prompt: how non-ai experts try (and fail) to design llm prompts. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pages 1–21, 2023.
  19. John Brooke. Sus: a “quick and dirty’usability. Usability evaluation in industry, 189(3):189–194, 1996.
  20. Drawing with reframer: Emergence and control in co-creative ai. In Proceedings of the 28th International Conference on Intelligent User Interfaces, pages 264–277, 2023.
  21. Determining what individual sus scores mean: Adding an adjective rating scale. Journal of usability studies, 4(3):114–123, 2009.
  22. Sparks: Inspiration for science writing using language models. In Designing interactive systems conference, pages 1002–1019, 2022.
  23. Where to hide a stolen elephant: Leaps in creative writing with multimodal machine intelligence. ACM Transactions on Computer-Human Interaction, 30(5):1–57, 2023.
  24. Art or artifice? large language models and the false promise of creativity. arXiv preprint arXiv:2309.14556, 2023.
  25. On the creativity of large language models. arXiv preprint arXiv:2304.00008, 2023.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (4)
  1. Ahmed Y. Radwan (3 papers)
  2. Khaled M. Alasmari (1 paper)
  3. Omar A. Abdulbagi (1 paper)
  4. Emad A. Alghamdi (9 papers)
Citations (3)
View on Semantic Scholar

Summary

Human-AI Collaborative Story Generation through SARD

Introduction

In the domain of storytelling, where human creativity forms the crux, the advent and integration of generative artificial intelligence (GenAI) technologies like LLMs have signaled a new age of narrative innovation. The paper under review introduces SARD, a visual drag-and-drop interface designed to harness the capabilities of LLMs for the generation of multi-chapter stories, examining its efficacy in bridging human creativity with AI's computational prowess. While SARD innovates on process facilitation, it surfaces challenges pertinent to node-based narrative visualization, lexical diversity of AI-generated content, and the cognitive load on writers. These findings underscore the complexities of integrating AI in creative processes and set a foundational path for future explorations in human-AI co-creative writing tools.

Methodology and System Design

SARD's architecture is predicated on a user-friendly, storyboard-based authoring tool, pivotally integrated with generative AI models through REST API and WebSocket communications. The system allows users to initiate stories by selecting genres and structures, then populate their narratives with characters, events, and relationships through a drag-and-drop interface. Descriptive content generation from images and narrative coherence across chapters are leveraged through well-crafted prompts to GPT-4, demonstrating a deep integration of AI in the creative workflow.

Key aspects of SARD include:

  • A canvas that serves as a visual storyboard for adding and linking narrative elements.
  • Options for setting the narrative genre and structure, defining the boundaries within which AI generates content.
  • The allowance for adding characters, events, and actions through nodes, providing a structured approach to story development.
  • Event ordering functionality, ensuring logical narrative progression.

Evaluation and Findings

Two distinct studies were conducted to evaluate SARD—a usability and collaboration examination followed by an assessment of story quality. Insights reveal a nuanced relationship between the tool's usability and the creative support it offers. Despite moderate sentiments on system usability, findings highlight a disparity between the intended and actual support of creative expression and collaboration with AI.

Key findings include:

  • Node-based visualization aids in building a mental model of the narrative, though it may contribute to cognitive overload as narratives complexify.
  • SARD-generated stories exhibit reduced lexical diversity, prompting a reconsideration of AI's role in enhancing narrative creativity.
  • User feedback suggests a desire for greater control over the narrative generation process, pointing towards necessary refinements in the interactivity and transparency of AI-driven prompts and outputs.

Implications and Future Directions

The exploration into SARD's capabilities and limitations yields rich insights into the evolving landscape of AI-assisted storytelling. This research opens several avenues for future investigations, particularly in enhancing user control and contribution in AI-collaborative processes, refining the cognitive ergonomics of authoring interfaces, and addressing the qualitative aspects of AI-generated content—specifically in terms of creativity, diversity, and alignment with user intentions.

Anticipated future work includes:

  • Enabling direct user interaction with AI prompts, facilitating a more intimate co-creative process.
  • Refinement of the node-based system to reduce cognitive load and improve narrative visualization.
  • Leveraging LLMs for more versatile and responsive story element generation, potentially through natural text input describing story plots.

Conclusion

The paper of SARD as a human-AI collaborative tool for story generation embarks on a critical inquiry into the feasibilities and challenges of integrating AI into creative writing. Through meticulous design and evaluative rigor, the research provides a foundational understanding of current limitations while positing a future where AI not only augments human creativity but does so in a manner that resonates with the user's cognitive and creative expectations.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown
X Twitter Logo Streamline Icon: https://streamlinehq.com

Tweets