Evaluating Large Language Models as Expert Annotators

TL;DR

This study evaluates whether large language models can replace human experts in specialized domain annotation tasks, finding limited benefits from reasoning techniques and revealing certain model behaviors in multi-agent discussions.

Contribution

It introduces a multi-agent discussion framework for LLM-based annotation and assesses their effectiveness in finance, biomedicine, and law domains, highlighting the limited gains from reasoning models.

Findings

Reasoning techniques offer minimal performance improvements.

Multi-agent discussions reveal persistent model behaviors.

LLMs show limited ability to adapt annotations in expert domains.

Abstract

Textual data annotation, the process of labeling or tagging text with relevant information, is typically costly, time-consuming, and labor-intensive. While large language models (LLMs) have demonstrated their potential as direct alternatives to human annotators for general domains natural language processing (NLP) tasks, their effectiveness on annotation tasks in domains requiring expert knowledge remains underexplored. In this paper, we investigate: whether top-performing LLMs, which might be perceived as having expert-level proficiency in academic and professional benchmarks, can serve as direct alternatives to human expert annotators? To this end, we evaluate both individual LLMs and multi-agent approaches across three highly specialized domains: finance, biomedicine, and law. Specifically, we propose a multi-agent discussion framework to simulate a group of human annotators, where LLMs are tasked to engage in discussions by considering others' annotations and justifications before finalizing their labels. Additionally, we incorporate reasoning models (e.g., o3-mini) to enable a more comprehensive comparison. Our empirical results reveal that: (1) Individual LLMs equipped with inference-time techniques (e.g., chain-of-thought (CoT), self-consistency) show only marginal or even negative performance gains, contrary to prior literature suggesting their broad effectiveness. (2) Overall, reasoning models do not demonstrate statistically significant improvements over non-reasoning models in most settings. This suggests that extended long CoT provides relatively limited benefits for data annotation in specialized domains. (3) Certain model behaviors emerge in the multi-agent discussion environment. For instance, Claude 3.7 Sonnet with thinking rarely changes its initial annotations, even when other agents provide correct annotations or valid reasoning.