Files and links (1)
Published VersionCC BY V4.0, Open Access
Conference paper
A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writing
Findings of the Association for Computational Linguistics: EMNLP 2023, pp.14504-14528
Conference on Empirical Methods in Natural Language Processing, 2023 (Singapore, 06-Dec-2023–10-Dec-2023)
Association for Computational Linguistics
2023
Related media
Expert Quote
Abstract
We evaluate a range of recent LLMs on En-glish creative writing, a challenging and complex task that requires imagination, coherence, and style. We use a difficult, open-ended scenario chosen to avoid training data reuse: an epic narration of a single combat between Ig-natius J. Reilly, the protagonist of the Pulitzer Prize-winning novel A Confederacy of Dunces (1980), and a pterodactyl, a prehistoric flying reptile. We ask several LLMs and humans to write such a story and conduct a human evalu-tion involving various criteria such as fluency, coherence, originality, humor, and style. Our results show that some state-of-the-art commercial LLMs match or slightly outperform our writers in most dimensions; whereas open-source LLMs lag behind. Humans retain an edge in creativity, while humor shows a binary divide between LLMs that can handle it comparably to humans and those that fail at it. We discuss the implications and limitations of our study and suggest directions for future research.
Details
- Title
- A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writing
- Authors
- Carlos Gómez-Rodríguez (Author) - Universidade da CoruñaPaul Williams (Author) - University of the Sunshine Coast, Queensland, School of Business and Creative IndustriesBree Glasbergen (Contributor) - University of the Sunshine Coast, Queensland, Student Services and Engagement
- Publication details
- Findings of the Association for Computational Linguistics: EMNLP 2023, pp.14504-14528
- Conference details
- Conference on Empirical Methods in Natural Language Processing, 2023 (Singapore, 06-Dec-2023–10-Dec-2023)
- Publisher
- Association for Computational Linguistics
- Date published
- 2023
- DOI
- 10.18653/v1/2023.findings-emnlp.966
- Copyright note
- Creative Commons Attribution 4.0 International License
- Organisation Unit
- School of Business and Creative Industries; Indigenous and Transcultural Research Centre; Student Services and Engagement; School of Law and Society; Sustainability Research Cluster
- Language
- English
- Record Identifier
- 991091296402621
- Output Type
- Conference paper
Metrics
1 File views/ downloads
19 Record Views