Breaking the Bias: Gender Fairness in LLMs Using Prompt Engineering and In-Context Learning

IF 0.2 0 HUMANITIES, MULTIDISCIPLINARY

Rupkatha Journal on Interdisciplinary Studies in Humanities Pub Date : 2023-12-14 DOI:10.21659/rupkatha.v15n4.10

Satyam Dwivedi, Sanjukta Ghosh, Shivam Dwivedi

引用次数: 0

Abstract

Large Language Models (LLMs) have been identified as carriers of societal biases, particularly in gender representation. This study introduces an innovative approach employing prompt engineering and in-context learning to rectify these biases in LLMs. Through our methodology, we effectively guide LLMs to generate more equitable content, emphasizing nuanced prompts and in-context feedback. Experimental results on openly available LLMs such as BARD, ChatGPT, and LLAMA2-Chat indicate a significant reduction in gender bias, particularly in traditionally problematic areas such as ‘Literature’. Our findings underscore the potential of prompt engineering and in-context learning as powerful tools in the quest for unbiased AI language models.

查看原文本刊更多论文

打破偏见：利用提示工程和情境学习实现法律硕士的性别公平

大型语言模型（LLM）被认为是社会偏见的载体，尤其是在性别代表性方面。本研究介绍了一种采用提示工程和语境学习的创新方法，以纠正大型语言模型中的这些偏见。通过我们的方法，我们有效地引导 LLM 生成更公平的内容，强调细微的提示和上下文反馈。在 BARD、ChatGPT 和 LLAMA2-Chat 等公开的 LLM 上进行的实验结果表明，性别偏见显著减少，尤其是在 "文学 "等传统的问题领域。我们的研究结果凸显了提示工程和语境学习作为无偏见人工智能语言模型的强大工具的潜力。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Rupkatha Journal on Interdisciplinary Studies in Humanities HUMANITIES, MULTIDISCIPLINARY-

自引率

0.00%

发文量

129

审稿时长

8 weeks

期刊介绍： “The fundamental idea for interdisciplinarity derives” as our Chief Editor Explains, “from an evolutionary necessity; namely the need to confront and interpret complex systems…An entity that is studied can no longer be analyzed in terms of an object of just single discipline, but as a contending hierarchy of components which could be studied under the rubric of multiple or variable branches of knowledge.” Following this, we encourage authors to engage themselves in interdisciplinary discussion of topics from the broad areas listed below and apply interdsiciplinary perspectives from other areas of the humanities and/or the sciences wherever applicable. We publish peer-reviewed original research papers and reviews in the interdisciplinary fields of humanities. A list, which is not exclusive, is given below for convenience. See Areas of discussion. We have firm conviction in Open Access philosophy and strongly support Open Access Initiatives. Rupkatha has signed on to the Budapest Open Access Initiative. In conformity with this, the principles of publications are primarily guided by the open nature of knowledge.