In an era where language models are increasingly integrated into
decision-making and communication, understanding the biases within Large
Language Models (LLMs) becomes imperative, especially when these models are
applied in the economic and political domains. This work investigates the
impact of fine-tuning and data selection on economic and political biases in
LLM. We explore the methodological aspects of biasing LLMs towards specific
ideologies, mindful of the biases that arise from their extensive training on
diverse datasets. Our approach, distinct from earlier efforts that either focus
on smaller models or entail resource-intensive pre-training, employs
Parameter-Efficient Fine-Tuning (PEFT) techniques. These techniques allow for
the alignment of LLMs with targeted ideologies by modifying a small subset of
parameters. We introduce a systematic method for dataset selection, annotation,
and instruction tuning, and we assess its effectiveness through both
quantitative and qualitative evaluations. Our work analyzes the potential of
embedding specific biases into LLMs and contributes to the dialogue on the
ethical application of AI, highlighting the importance of deploying AI in a
manner that aligns with societal values.

本研究探究了大型语言模型中的经济和政治偏见以及调整细节对其产生的影响，通过使用参数效率微调技术，我们介绍了一种系统的数据集选择、注释和指令调整方法，并通过定量和定性评估评估其有效性，旨在讨论 AI 的道德应用，强调将 AI 部署在符合社会价值观的方式上的重要性。

分析对数据选择和微调对 LLMs 中的经济和政治偏见的影响

Analyzing the Impact of Data Selection and Fine-Tuning on Economic and  Political Biases in LLMs

A constitution is the highest legal document of a country and serves as a
guide for the establishment of other laws. The constitution defines the
political principles, structure, hierarchy, position, and limits of the
political power of a country's government. It determines and guarantees the
rights of citizens. This study aimed at topic modeling of Iranian laws. As part
of this research, 11760 laws were collected from the Dotic website. Then, topic
modeling was conducted on the title and content of the regularizations using
LDA. Data analysis with topic modeling led to the identification of 10 topics
including Economic, Customs, Housing and Urban Development, Agriculture,
Insurance, Legal and judicial, Cultural, Information Technology, Political, and
Government. The largest topic, Economic, accounts for 29% of regulations, while
the smallest are Political and Government, accounting for 2%. This research
utilizes a topic modeling method in exploring law texts and identifying trends
in regularizations from 2016-2023. In this study, it was found that
regularizations constitute a significant percentage of law, most of which are
related to economics and customs. Cultural regularizations have increased in
2023. It can be concluded any law enacted each year can reflect society's
conditions and legislators' top concerns.

该研究使用主题建模方法对伊朗法律进行了探索，识别了经济、法规、文化、信息技术和政治等主题，并发现法规主要涉及经济和海关，而文化规范在 2023 年有所增加。每年制定的法律能反映社会状况和立法者的关注重点。