Legal Data Firm Makes Push to Introduce AI to Lawyers' Work

LexisNexis on Thursday announced the release of Lexis AI, a platform that will use artificial intelligence to tap into the company’s database of documents and records to draft documents and research case law.

The announcement comes as AI continues to disrupt industries from education to entertainment and amid growing calls to keep artificial intelligence in check. (Pexels)

A leading legal research database company is making an early push to introduce generative artificial intelligence to tasks typically performed by paralegals or lawyers.

LexisNexis on Thursday announced the release of Lexis AI, a platform that will use artificial intelligence to tap into the company's database of documents and records to draft documents and research case law. Some large US law firms, including Baker McKenzie, Reed Smith LLP and Foley & Lardner LLP are working with the company to provide early support for the product, the company said.

The announcement comes as AI continues to disrupt industries from education to entertainment and amid growing calls to keep artificial intelligence in check. Geoffrey Hinton — the so-called godfather of AI — recently resigned from Google over his concerns about the impact of AI on society, including on employment.

LexisNexis operates one of the major legal research databases that lawyers use to fine cases to bolster their legal arguments. That research has historically been performed by junior lawyers or paralegals. Bloomberg LP, the parent of Bloomberg News, also owns Bloomberg Law, which directly competes with LexisNexis in the research database business.

Jeff Pfeifer, LexisNexis's chief product officer for the UK and North America, cited a March survey that found a majority of lawyers and law students believe generative AI platforms like ChatGPT will “significantly transform the practice of law.”

But he sought to distinguish the company's product from the more mainstream examples, which have recently attracted criticism for incorporating false or misleading information in their responses.

Traditional large-language models have struggled, as the content supporting them can be dated, lacks citation authority and is prone to “conceptual and factual hallucinations,” Pfeifer said. Lexis AI will tap into the company's database to provide more verifiable content, he said.

“I don't think that it's possible to warrant at this stage of AI development that any model can be hallucination-free and and we've spoken with many of our customers about that,” Pfeifer said. “We tried to minimize it to the greatest degree possible by interconnecting the model and its answer to content that it's exposed to from LexisNexis.”