Self-Reflection Outcome is Sensitive to Prompt Construction

Large language models (LLMs) demonstrate impressive zero-shot and few-shot reasoning capabilities. Some propose that such capabilities can be improved through self-reflection, i.e., letting LLMs reflect on their own output to identify and correct mistakes in the initial responses. However, despite s...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Liu, Fengyuan, AlDahoul, Nouar, Eady, Gregory, Zaki, Yasir, AlShebli, Bedoor, Rahwan, Talal
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 14.06.2024
Subjects
Errors
Large language models
Prompt engineering
Online AccessGet full text

Cover

More Information
Summary:Large language models (LLMs) demonstrate impressive zero-shot and few-shot reasoning capabilities. Some propose that such capabilities can be improved through self-reflection, i.e., letting LLMs reflect on their own output to identify and correct mistakes in the initial responses. However, despite some evidence showing the benefits of self-reflection, recent studies offer mixed results. Here, we aim to reconcile these conflicting findings by first demonstrating that the outcome of self-reflection is sensitive to prompt wording; e.g., LLMs are more likely to conclude that it has made a mistake when explicitly prompted to find mistakes. Consequently, idiosyncrasies in reflection prompts may lead LLMs to change correct responses unnecessarily. We show that most prompts used in the self-reflection literature are prone to this bias. We then propose different ways of constructing prompts that are conservative in identifying mistakes and show that self-reflection using such prompts results in higher accuracy. Our findings highlight the importance of prompt engineering in self-reflection tasks. We release our code at https://github.com/Michael98Liu/mixture-of-prompts.
ISSN:2331-8422