When Google used its DeepMind artificial intelligence to defeat the world’s greatest Go player in 2016, it gave us a glimpse at the potential for AI to outsmart humans. Critics feared more demonstrations of human’s powerlessness against robot competitors would follow.
It turns out, they were right. A new study by legal AI company LawGeex put 20 experienced lawyers up against trained AI to determine who could spot contract errors with greater speed and accuracy. Once again, the robots won.
Contestants were given four hours to review five non-disclosure agreements looking for 30 different legal issues found in everyday business contracts. After sifting through 11 pages containing 152 paragraph and more than 3,000 clauses, each contestant submitted their work for scoring.
Sadly, for Team Humans, it was a blowout. The lawyers put up a good fight, finding a respectable 85 percent of mistakes, but were comprehensively outdone by AI, which had a 94 percent accuracy rate.
That may not seem like a huge margin, but consider this: the fastest human lawyer completed the task in 51 minutes and the average time was 92 minutes. The AI? It took just 26 seconds. Additionally, the AI had a 100 percent accuracy rate for one document where the best result from a human was just 96 percent.
Obviously, the results raise concerns about the job security of lawyers. Will AI be defending us in court within the next decade? Probably not, Erika Buell, clinical professor at Duke University School of Law, tells Mashable. In fact, AI could end up being a lawyer’s best friend.
“Having the AI do a first review of an NDA, much like having a paralegal issue spot, would free up valuable time for lawyers to focus on client counseling and other higher-value work,” Buell told Mashable.
Earlier this year, Microsoft and Alibaba AI defeated humans in a global reading test conducted by Stanford. Like the AI used in LawGeex’s test, Microsoft insists AI will only reach its full potential when used in harmony with humans.