Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey on Large Language Model Hallucination via a Creativity Perspective

Xuhui Jiang, Yuxing Tian, Fengrui Hua, Chengjin Xu, Yuanzhuo Wang, Jian Guo

TL;DR

This survey reframes LLM hallucinations from solely a reliability problem to a potential engine for creativity. It systematically reviews hallucination taxonomy, detection, and reduction, then links creativity definitions and assessment to LLM outputs, proposing a divergent-convergent framework to harness hallucinations for creative ends. By integrating historical parallels, cognitive science, and recent LLM research, the paper outlines methods and metrics for evaluating and refining creative outputs while managing risks. The work lays a roadmap for theory, datasets, benchmarks, and multimodal extensions to expand creative applications of LLM hallucinations.

Abstract

Hallucinations in large language models (LLMs) are always seen as limitations. However, could they also be a source of creativity? This survey explores this possibility, suggesting that hallucinations may contribute to LLM application by fostering creativity. This survey begins with a review of the taxonomy of hallucinations and their negative impact on LLM reliability in critical applications. Then, through historical examples and recent relevant theories, the survey explores the potential creative benefits of hallucinations in LLMs. To elucidate the value and evaluation criteria of this connection, we delve into the definitions and assessment methods of creativity. Following the framework of divergent and convergent thinking phases, the survey systematically reviews the literature on transforming and harnessing hallucinations for creativity in LLMs. Finally, the survey discusses future research directions, emphasizing the need to further explore and refine the application of hallucinations in creative processes within LLMs.

A Survey on Large Language Model Hallucination via a Creativity Perspective

TL;DR

This survey reframes LLM hallucinations from solely a reliability problem to a potential engine for creativity. It systematically reviews hallucination taxonomy, detection, and reduction, then links creativity definitions and assessment to LLM outputs, proposing a divergent-convergent framework to harness hallucinations for creative ends. By integrating historical parallels, cognitive science, and recent LLM research, the paper outlines methods and metrics for evaluating and refining creative outputs while managing risks. The work lays a roadmap for theory, datasets, benchmarks, and multimodal extensions to expand creative applications of LLM hallucinations.

Abstract

Hallucinations in large language models (LLMs) are always seen as limitations. However, could they also be a source of creativity? This survey explores this possibility, suggesting that hallucinations may contribute to LLM application by fostering creativity. This survey begins with a review of the taxonomy of hallucinations and their negative impact on LLM reliability in critical applications. Then, through historical examples and recent relevant theories, the survey explores the potential creative benefits of hallucinations in LLMs. To elucidate the value and evaluation criteria of this connection, we delve into the definitions and assessment methods of creativity. Following the framework of divergent and convergent thinking phases, the survey systematically reviews the literature on transforming and harnessing hallucinations for creativity in LLMs. Finally, the survey discusses future research directions, emphasizing the need to further explore and refine the application of hallucinations in creative processes within LLMs.
Paper Structure (16 sections, 4 figures)

This paper contains 16 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Hallucination in the Era of LLM
  3. Hallucination Taxonomy
  4. Hallucination Detection
  5. Hallucination Reduction
  6. The Creativity Hidden in Hallucination
  7. The Creativity of Large Language Models
  8. Definition of Creativity
  9. Approaches for measuring creativity
  10. Recent works on evaluating LLM's creativity
  11. Harness LLM Hallucination For Creativity
  12. Divergent Phase of LLM
  13. Convergent Phase of LLM
  14. Conclusion and Future Work
  15. Conclusion
  16. ...and 1 more sections

Figures (4)

  • Figure 1: General framework of the survey.
  • Figure 2: Cases of relations between hallucination and creativity.
  • Figure 3: The divergent phase stimulates the hallucination of LLMs, fostering creative thought.
  • Figure 4: The convergent phase refines hallucinations into valuable creative contributions.