Experimenting with Legal AI Solutions: The Case of Question-Answering for Access to Justice

TL;DR

This paper presents a human-centric legal NLP pipeline, introduces the LegalQA dataset with expert-annotated legal questions and answers, and demonstrates that retrieval-augmented generation with limited data can match or outperform broader internet retrieval.

Contribution

It introduces the LegalQA dataset, develops an evaluation protocol, and shows that limited citation-based retrieval can effectively answer legal questions, highlighting potential and limitations of open-source legal AI.

Findings

Retrieval-augmented generation with 850 citations matches or outperforms internet-wide retrieval.

LegalQA dataset covers diverse legal questions with expert answers and citations.

Open-source models currently lag behind closed-sourced solutions in legal AI.

Abstract

Generative AI models, such as the GPT and Llama series, have significant potential to assist laypeople in answering legal questions. However, little prior work focuses on the data sourcing, inference, and evaluation of these models in the context of laypersons. To this end, we propose a human-centric legal NLP pipeline, covering data sourcing, inference, and evaluation. We introduce and release a dataset, LegalQA, with real and specific legal questions spanning from employment law to criminal law, corresponding answers written by legal experts, and citations for each answer. We develop an automatic evaluation protocol for this dataset, then show that retrieval-augmented generation from only 850 citations in the train set can match or outperform internet-wide retrieval, despite containing 9 orders of magnitude less data. Finally, we propose future directions for open-sourced efforts, which fall behind closed-sourced models.