Your Lawyer May Soon Be an Algorithm
Algorithms can already do menial law tasks as well as a junior lawyer.
A courtroom with computers. Image: Wolf Law Library/Flickr
Lawyers are the latest to find themselves in the firing line of robots—or artificially intelligent algorithms, at any rate.
The robotic revolution has been predicted to spread its techno-tendrils far and wide in the job market, and a recent report by UK consultancy firm Jomati Consultants suggests that they'll be creeping ever further into the legal profession by 2030. The report suggests that the "economic model of law firms is heading for a structural revolution, some might say a structural collapse."
While robots stealing jobs is nothing new, that lawyers could see their jobs be automated might seem surprising on the surface. In a much-publicised 2013 report that predicted 47 percent of US jobs were at risk of automation in the next two decades, authors Carl Frey and Michael Osborne put lawyers at comparative low risk of robotic replacement. In a recent follow-up focusing on the UK job market, they also named law as one of the professions at the least risk of roboticization, alongside sectors including financial services and engineering.
But some jobs in law find themselves at the riskier end of the spectrum. In the 2013 paper, paralegals and legal assistants were ranked around the 100th most at-risk occupation (out of 702) with a "computerisable" probability of 0.94.
That's less surprising; the general trend seems to be that high-paying jobs are much safer from robotisation than low-paying jobs. The hotshot experienced lawyer dealing with highly unpredictable problems is less replaceable than the legal secretary doing his or her research grunt work.
The economic model of law firms is heading for a structural revolution
The kind of roles AI could take over in law are evident, and computers have already carved a niche in some tasks. The most obvious are routine tasks like combing through documents—something a human's eyes and brain find tiresome after a while but a computer has no problem with.
That could be as simple as searching tons of documents for keywords, but more intelligent tools can go further, taking in context and sentiment. "Predictive coding" algorithms are increasingly used by lawyers to help in the discovery process—when they're trying to unearth evidence to support their case. For example, imagine combing through a corporation's massive data stores, most of which is probably irrelevant, in search for signs of financial malfeasance. Algorithms are able to sort through for relevant docs based on initial samples given by a human.
The main attraction for companies is obvious: an algorithm doesn't require a salary and reasonable working conditions, and it can process information a lot quicker. The Jomati report suggests rough figures: a very junior lawyer doing this kind of work might cost $100,000 a year. An AI bot might cost around $500,000, with that price set to fall as the market becomes more competitive. "Even at this price they would be worth the cost as they can work 24 hours a day, 7 days a week with no downtime, thereby eclipsing the chargeable hours of the most workaholic lawyers," the authors write.
And a 2011 paper in the Richmond Journal of Law and Technology found that "e-discovery" was not only more efficient but also gave better results. Of course, the documents an algorithm picks out still have to be reviewed by a human—but it reduces the workload, the man hours, and the cost. And the low-level jobs.
But even before a case gets rolling, AI can play a part in getting the legal cogs turning. Talking to Slate earlier this year, legal technology scholar Daniel Katz proposed that AI may help complainants decide whether it's worth going to the hassle and expense of bringing a lawsuit—by predicting if they'll win.
Katz, who works at Michigan State University, recently published a paper in which he claimed a computer model was able to successfully predict US Supreme Court decisions 70 percent of the time. The model used "only data available prior to the date of decision," effectively comparing similar cases in order to forecast the logical outcome.
The number of associates that firms need to hire will be greatly reduced
And in other initiatives, AI is actually helping to make those decisions. A New Scientist article last year described a German application that makes automatic decisions about child benefit payments.
AI making legal decisions is a compelling idea. Not only is it more efficient, but couldn't it somehow be more just to place decisions in an entity unable to be swayed by bias or emotion?
We're not at that point yet, and the lawyers at that kind of high-level decision-making level are the ones with least to worry about in terms of job security. They're the ones that have the (so-far) uniquely human talent of keeping clients happy and drawing in new business.
But it seems more feasible that the career path of lawyers trying to reach those positions could change, if more of the lower steps on the ladder—the more mundane tasks that conventionally act as a training ground for future big dog lawyers—are taken over by the bots.
"The number of associates that firms need to hire will be greatly reduced, at least if the intention is to use junior lawyers for billable work rather than primarily to educate and train them ready to become business winners," the Jomati consultants predict.
In a recent blog post, lawyer and consultant Kenneth Grady proposed a law-specific Turing Test to see if an AI would be capable of passing for a newly graduated lawyer when questioned only on the subject of the law.
Whether any recent graduates would dare fill in for the human side of the challenge while their law school debt looms over them is another question.