In what should primarily be seen as a bit of fun, legal AI pioneer, LawGeex, has again held a ‘man vs machine’ competition, pitting its NLP algorithms against a human lawyer. The competition was part of a special report broadcast with HBO in conjunction with Vice News about ‘The Future of Work’.
And, as one would expect……the AI + human side beat the human alone side.
The report looked at a wide range of AI and automation areas, but included law as part of the segment. You can watch the short version of the LawGeex challenge here.
Vice News Correspondent and Producer Gianna Toboni came to LawGeex’s New York City office to put its AI to the test. Vice News presented both LawGeex and a human lawyer they selected with two NDAs, one which was four-pages long and the other two-pages long.
By chance, the human lawyer was none other than Tunji Williams, the co-founder of legal tech startup DealWIP, who is also based in New York. DealWIP is a deal platform that helps lawyers close transactions more efficiently.
‘LawGeex bested the human lawyer. Tunji Williams, graduate of the Top 10 Law School at the University of Virginia, took over an hour to review the two documents which he did with 85% and 83% accuracy, respectively. LawGeex, operated by LawGeex CEO, Noory Bechor, spent just 18 minutes reviewing and achieved 95% accuracy on both contracts.’
Toboni asked why Williams didn’t seem disappointed with the result. Williams answered: ‘I wasn’t disappointed when the iPhone came out and I could do more things with this piece of technology, so this is exciting to me.’
While Bechor added with regard to the need for lawyers, ‘Absolutely [we need lawyers]. Defining the policy, serving as the escalation point, handling negotiation, and handling more complex contract categories. These are and should be the role of lawyers.’
‘You have a new generation of lawyers that are much more tech savvy. The ones that can actually leverage technology are the ones that manage to prosper,’ he added.
This is the second time LawGeex has run a ‘man vs machine’ challenge. This time was obviously more for fun than a serious scientific study. But, an earlier challenge they held was far broader and detailed. It can be seen here.