Artificial Intelligence is now encroaching on the legal profession, according to a new study released by the A.I. company LawGeex this month.

The study detailed a competition between twenty experienced contract lawyers and an A.I. built to examine non-disclosure agreements (NDAs). The participating lawyers came from a wide variety of legal backgrounds; some worked in prestigious law firms, like Alston & Bird, and others worked as in-house lawyers for companies including Goldman Sachs and Cisco. They were all thoroughly vetted to ensure they had ample experience in reviewing NDAs.

Participants were given five different NDAs to evaluate, and were awarded points for flagging potential legal issues in the contracts. The algorithm outperformed the average human score on all of the NDAs — and it wasn’t even particularly close.

law study
Only "Lawyer 19" consistently outperformed the algorithm.

“The test gave me a practical glimpse into how technology can automate a staple of the legal profession — reviewing NDAs,” said Harvard-educated lawyer and study participant Shena Shenoi. “The type of issue-spotting carried out is credible and quite similar to how we (manually) do this type of work and have for decades.”

This is an important innovation because reviewing contracts requires ample time and attention from highly trained legal experts. The work is both ubiquitous and tedious; if computers could review the documents instead, it would free up lawyers to spend their time on other tasks. To that end, LawGeex developers worked on the algorithm for over three years, training it on tens of thousands of contracts.

“Having a tool that could automate this process would free up skilled attorneys to spend their time on higher-level tasks without having to hire paralegal support,” said participating attorney Grant Gulveson.

It’s not the first time A.I. companies have attempted to disrupt the legal world. Last November in England, a startup called Case Cruncher demonstrated that their algorithm could guess whether or not insurance claims merited legal attention at a better rate than a cohort of human lawyers.

But LawGeex’s A.I. is different from earlier efforts in one important respect. While the Case Cruncher program is a predictive system, the LawGeex algorithm actually does the work of a lawyer by pointing out legal flaws in a document.

That doesn’t necessarily mean that robots are coming to take lawyers’ jobs — although paralegals might be in some trouble. Advances in A.I. offer a future where attorneys work with computers in concert, like a pilot using an autopilot system. “The reality of this powerful technology is that it is not meant to be, nor indeed is it currently capable of being, used as a standalone tool,” the study said.

If LawGeex can successfully bring their algorithm to practicing lawyers, they may drastically transform the legal system. Aside from freeing lawyers from mountains of paperwork, widespread use of their algorithm could decrease the cost of legal assistance, because so much of the attorney’s job would be taken on by a computer.

“This study represents a major landmark in the history of legal technology,” the study said.