Large language models (LLMs) are AI systems capable of generating and analysing natural language texts. They have been utilized across various fields, including, among others, legal text generation, where they can aid lawyers in drafting litigation-focused documents such as contracts, pleadings, and motions, or even assist them with ideas and legal research for preparing legal reports or memos. However, the use of LLMs in legal text generation raises ethical and legal concerns, such as responsibility, accountability, and liability for the generated texts, as well as potential algorithmic discrimination and gender biases amongst others. In this article, we review the current state of LLMs in legal text generation, focusing on the strengths and limitations of AI-powered legal research software used by law firms based in the EU. We also discuss gender implications of using LLMs for legal text generation and offer some recommendations and best practices to ensure the quality, reliability and fairness of the generated texts.
Casagran et al. (Fri,) studied this question.