Investigating Annotator Bias in Large Language Models for Hate Speech Detection

TL;DR

This paper investigates biases in large language models when annotating hate speech, focusing on gender, race, religion, and disability, and introduces a new dataset to analyze these biases comprehensively.

Contribution

It provides a detailed analysis of LLM biases in hate speech annotation across four categories and introduces HateBiasNet, a new dataset for bias research.

Findings

LLMs exhibit significant biases in hate speech annotation.

Bias varies across different LLMs and demographic categories.

The study offers insights for improving LLM-based annotation methods.

Abstract

Data annotation, the practice of assigning descriptive labels to raw data, is pivotal in optimizing the performance of machine learning models. However, it is a resource-intensive process susceptible to biases introduced by annotators. The emergence of sophisticated Large Language Models (LLMs) presents a unique opportunity to modernize and streamline this complex procedure. While existing research extensively evaluates the efficacy of LLMs, as annotators, this paper delves into the biases present in LLMs when annotating hate speech data. Our research contributes to understanding biases in four key categories: gender, race, religion, and disability with four LLMs: GPT-3.5, GPT-4o, Llama-3.1 and Gemma-2. Specifically targeting highly vulnerable groups within these categories, we analyze annotator biases. Furthermore, we conduct a comprehensive examination of potential factors contributing to these biases by scrutinizing the annotated data. We introduce our custom hate speech detection dataset, HateBiasNet, to conduct this research. Additionally, we perform the same experiments on the ETHOS (Mollas et al. 2022) dataset also for comparative analysis. This paper serves as a crucial resource, guiding researchers and practitioners in harnessing the potential of LLMs for data annotation, thereby fostering advancements in this critical field.