Trending In Legal Transformation

Your #1 source of legal transformation knowledge.

Lexion adopts GPT-3 in its new AI Contract Assist product

Generative Pre-trained Transformer 3 (GPT-3) is a hot topic these days. Introduced in May 2020, GPT-3 is the latest iteration in the GPT-n series of language prediction models created by San Francisco-based artificial intelligence research lab OpenAI.

GPT-3 seems to have all the answers. Given an initial text as a prompt, it continues producing human-like text. Budding artists can use a recent version of GPT-3 to create images. It can even help you produce new recipes in your kitchen. And it can also have specific legal applications. 

TILT tested GPT-3 and the results are impressive, although it sometimes provides semi-accurate or incomplete answers. 

Regardless, the tool can obviously benefit a wide range of industries, including the legal industry. Lawyers can use it off-the-shelf to look for legal information, or even ask GPT-3 to draft legal clauses, based on specific inputs. Legal tech providers, meanwhile, can integrate GPT-3 into their products. Seattle-headquartered legal tech startup Lexion, which offers tools to simplify contract management, has recently done just this. 

Lexion’s latest product — AI Contract Assist — uses GPT-3 to further help lawyers in the contract review process. The tool is integrated with Microsoft Word, where it can suggest specific language edits to documents or draft entire clauses. Lawyers can use it to search for alternative ways of phrasing, or to make language changes at scale. AI Contract Assist can also summarize specific parts of a given document — which seems particularly useful for junior team members. The tool is currently in the beta-testing phase

Lexion and other legal tech providers who are testing GPT-3 should be praised for their pioneering efforts. Although GPT-3 is an open platform, users need to pay to integrate it into their products. Their investment should pay off once GPT-3 becomes fully reliable and free from the unexpected answers it still occasionally provides. For now, human supervision remains necessary.  

Facebook
Twitter
LinkedIn