Subscribe to LexTalk to stay on top of today’s legal issue and trends.
Catapult Your Career |
Industry Insights & Trends |
Product Training & Tips
The article below has been republished in full courtesy of Law360, written by Eric Kroh.
At the time Ken Jennings stepped up to the lectern to compete against “Jeopardy’s” first robot competitor, he had already secured his spot in history with a record 74-game winning streak. His other (human) opponent, Brad Rutter, was the biggest all-time money winner, having taken home a record $3.2 million.
The two trivia superstars were there to represent the best of human intellect in the popular quiz show’s first contest pitting man against machine. But they were no match for IBM Corp.’s Watson, which, as show host Alex Trebek explained, had the informational firepower of 2,800 computers tied together in a super-high-speed network with a memory capacity of more than 15 trillion bytes.
The first question of the two-day competition in early 2011 was answered correctly by Rutter, who at that point had been undefeated after appearing on the show 19 times. Watson fell behind the others after a series of missed clues, but it soon made up ground, and by the midpoint of the first day it was tied for the lead with Rutter.
The tight race soon turned into a rout, however, with Watson running the board as Rutter and Jennings could only watch resignedly.
“Who is Franz Liszt? Who is the Church Lady? What is narcolepsy?” Watson droned in its synthesized voice.
MAN VERSUS MACHINE: 74-game winner Ken Jennings gives it up for “Watson,” the IBM supercomputer that went on to beat Jennings and fellow “Jeopardy” champ Brad Rutter in 2011. (Credit: AP)
The second day of the competition went much the same. A pivotal moment came in the second round when Watson scored a “Daily Double” in a category called “Legal ‘E’s.” By the end of the day, Watson had racked up a two-day total of $75,000. The other two didn’t break $25,000.
“I, for one, welcome our new computer overlords,” Jennings said at the time, referencing a famous line from “The Simpsons” that had spawned repeated use of a similar phrase. He later quipped in a piece for Slate that “‘Quiz show contestant’ may be the first job made redundant by Watson, but I'm sure it won't be the last.”
Since that initial public demonstration of the power of artificial intelligence, the technology has been heralded as a game-changer for a wide variety of industries. IBM has licensed the Watson platform to software developers to create tools to give financial advice, plan vacations and help doctors diagnose diseases. The company’s then-general counsel, Robert Weber, predicted in a 2011 column for The National Law Journal that the technology would be a “boon” to the legal profession, “a digital associate, if you will.”
It was only a matter of time before Watson entered the legal industry, and in 2014 a lawyer named Andrew Arruda founded a company called Ross Intelligence to develop a virtual legal assistant powered by the platform.
Companies like Ross are starting to gain traction as technological advancements make virtual legal assistants a reality and firms look for ever more ways to be competitive in a cutthroat legal market. Now, artificial intelligence, along with a host of other technologies, threatens to upend the traditional BigLaw model and could force more associates — already pressed for jobs — out of work.
Automation is not new in the legal industry, but artificial intelligence represents the next step in software taking more and more duties off the hands of human lawyers. The collection of technologies that fall under the rubric of “artificial intelligence” — machine learning, natural language processing, expert systems and others — enable computers to handle legal tasks they were previously incapable of performing.
The founding team of Ross Intelligence includes (from left) CEO Andrew Arruda, CTO Jimoh Ovbiagele and Pargles Dall’Oglo. (Courtesy of Ross Intelligence)
The assistant pioneered by Arruda’s company, nicknamed Ross, uses natural language processing to understand questions posed to it by lawyers and to return information on relevant court cases and legislation in a matter of seconds.
So far, Ross, which has undergone testing by some 20 law firms, has learned bankruptcy law, and it will soon branch out into other practice areas including labor and employment, intellectual property, tax and personal injury. Arruda’s goal is to have every lawyer have access to Ross.
In early May, Ross Intelligence said BakerHostetler had agreed to license Ross in its bankruptcy practice, one of the company’s first commercial BigLaw clients. A couple of weeks later, it announced that Latham & Watkins LLP and von Briesen & Roper SC were licensing Ross.
Ross Intelligence is just one of many recently launched companies that are harnessing bleeding-edge technologies to help lawyers do their work faster, better and cheaper. Companies have built software to scour documents for information, analyze data, predict the outcome of cases, and perform other legal functions. Traditionally slow to innovate, the legal profession seems on the verge of being swept up in a wave of new technologies that will reshape the industry.
One of Ross Intelligence’s investors is NextLaw Labs, a proving ground for legal tech startups based in Palo Alto, California, and supported by Dentons. CEO Dan Jansen came to NextLaw Labs in 2015 from a varied background that includes launching an online advertising business, leading the media and entertainment practice at a global consulting firm and founding a credit union.
Dan Jansen (left), CEO of NextLaw Labs, has a background in media and advertising, while NextLaw Labs global chair John Fernandez is also U.S. chief innovation officer at Dentons.
From the perspective of an outsider, the profession of law is lagging behind on the adoption of technology, Jansen said. The global legal services industry, which generates annual revenues of between $600 billion and $1 trillion, depending on how you measure it, is overdue for disruption, he said.
For quite some time, offering bespoke legal advice and charging a premium for it has worked well for law firms, and without pressure from the outside there has been no need for attorneys to change the way they do business, Jansen said. Under the billable hour compensation model, firms even have a disincentive to become more streamlined and efficient, because they cannot charge for time saved, he said.
However, in a shift that began before the financial crisis of 2007 and 2008, the marketplace for legal services has become more competitive, according to Jansen’s colleague, John Fernandez, global chair of NextLaw Labs and U.S. chief innovation officer at Dentons. Clients, who have gone through their own rounds of cost-cutting, have noticed that their legal fees continue to grow and are demanding more streamlining from their attorneys, he said.
“Changes have been building for quite a while now,” Fernandez said. “There’s more intense pressure on the client’s side in terms of cost and consolidations, and a general awareness about how technology can be deployed today,” he said. “There’s a real tipping point in the industry.”
Do It Better
Acouple of years ago, Jonathan Henderson, chair of the corporate and transactional practice group at Polsinelli PC, started thinking about how his firm conducted due diligence on mergers and other transactions. Due diligence is typically a time- and labor-intensive process, with attorneys camped out in a room searching through stacks of contracts to identify relevant clauses.
The contract review process takes up so much time that it can eat up a large percentage of the costs of a corporate transaction. Though tedious, contract review is complex and important work, and mistakes can be disastrous.
Henderson, who has spent 25 years as an M&A lawyer, thought the work might be better suited for computers, theorizing they would be less prone to error searching for clauses at 3 o’clock in the morning than a first-year associate. He started looking for a technology tool to do the job but was not impressed with what he found. Then, in April of last year, he came across software developed by a company called Kira Systems that seemed to be exactly what he was looking for.
Kira was developed by Noah Waisberg, who as a former junior M&A attorney at Weil Gotshal & Manges LLP in New York spent countless hours and late nights of his own reviewing contracts to perform due diligence for major corporate transactions. Waisberg founded Kira Systems in 2011 along with Alexander Hudek, who holds a doctorate in computer science from the University of Waterloo.
“Our idea was, since people are looking through these agreements for the same questions, maybe we could build software to help find that information,” Waisberg said.
The result is a series of products launched in early 2014 that use machine learning to scan contracts and other documents for information specified by the user. Lawyers can train Kira to recognize specific kinds of information in documents by feeding it a test sample, which the software uses to find similar information in other documents. For example, if a contract reviewer looking for change-of-control provisions shows Kira some examples of the provisions, the software learns what the provisions look like and can immediately highlight them in other contracts for the reviewer to analyze.
Kira Systems was founded in 2011 by former Weil Gotshal attorney Noah Waisberg (left) and Alexander Hudek.
Kira’s clients, which include BigLaw firms and Big Four accounting firms, report that the software enables them to do the work in 20 percent to 90 percent less time, Waisberg said. Kira is also more accurate than humans. In one test, a law firm took two junior attorneys — one using Kira software and one not — and set them to work searching for assignment clauses across 50 agreements, Waisberg said. The attorney who didn’t use Kira took six hours to find the clauses, while the Kira-enabled attorney took two-and-a-half hours and found some clauses the other had missed, he said.
“We’re helping lawyers do what they currently do and do it better in a different way,” Waisberg said.
Henderson said the value he sees in Kira is in enabling attorneys to get through the tedious contract review process of a transaction more quickly so they can focus on more important analytical work.
“In the deal world, speed is a competitive advantage,” Henderson said. “If I can get to those answers quicker, then processes can accelerate ... acquisitions occur on time, revenue is produced, and lots of good things happen.”
Elizabeth Ellis, director of knowledge management at Torys LLP, which also uses Kira to perform due diligence, said clients don’t particularly care if a firm uses such advanced technologies or not, but the software provides an avenue for the firm to keep up with clients’ expectations for higher-quality legal services, however they are delivered.
“Why shouldn’t clients expect that from their lawyers?” Ellis said. “Why wouldn’t a client be interested in law firms that figure out a faster and better way to deliver legal services?”
Collapse of the Pyramid
Law firm clients are attracted to artificial intelligence not because of the technology so much as how it can bend the cost curve, according to Ken Grady, who goes by the title of “lean law evangelist” at Seyfarth Shaw LLP and spent 20 years in-house at companies such as Wolverine World Wide Inc., K B Toys Inc. and Payless ShoeSource Inc.
Clients primarily care about the accuracy, speed and volume of legal services, and in the past to improve those factors they have had to pay for the necessary labor costs of throwing more lawyers at a problem, Grady said. The outsourcing of labor by managed service providers can potentially reduce labor costs somewhat, but artificial intelligence promises to drive them down significantly without sacrificing accuracy, speed and volume, he said.
“Once you’ve gotten labor to a certain level it’s hard to push it much lower and the next step is this technology,” Grady said.
If law firms adopt artificial intelligence to cut back on labor costs, though, that could also mean they will not need to hire as many attorneys, according to legal industry consultant Jordan Furlong.
“In the short term, with law firms focused on reducing their costs wherever they can — and they’re rapidly running out of places they can cut — then yeah, I think the first approach might be to say, ‘Let’s get this machinery in and we can do without lawyers,’” Furlong said.
“One way or another, it’s going to reduce the total amount of workload that traditionally, especially newer lawyers and younger lawyers have been assigned to do,” he said. “The bottom line is, what this really augurs is the re-engineering and the reinvention of how legal work gets done.”
Technology and automation have already taken a toll on a wide swath of jobs. A study by the Associated Press found that from 2000 to 2010, there was a significant reduction in employment for travel agents, telephone operators, typists, data entry workers, accounting clerks and secretaries, with computers taking over their work. According to the Bureau of Labor Statistics, employment of legal secretaries in the U.S. fell from 221,000 in 2011 to 203,000 in 2015, a drop of some 8 percent. At the same time, the number of paralegals rose to 272,000 from 252,000 as paralegals took on more tasks previously assigned to legal secretaries and other legal support workers, the BLS said.The BLS also projects employment for lawyers to grow at a rate of 6 percent from 2014 to 2024, with competition for jobs remaining strong as increasing ranks of law schools graduates outnumber available jobs each year.But some predict that artificial intelligence threatens to encroach on higher-skilled careers. A November study by consulting firm McKinsey & Co. found that 23 percent of the work performed by lawyers and 69 percent of paralegals’ work had the potential to be automated by adapting currently demonstrated technologies.
According to a December paper titled “Can Robots Be Lawyers?” by Dana Remus of the University of North Carolina School of Law and Frank Levy, an economist at the Massachusetts Institute of Technology, artificially intelligent machines will not completely replace lawyers anytime soon, but they will automate some of the tasks traditionally performed by human attorneys.
Remus and Levy estimated how much of lawyers’ work could actually be automated by technology using data from Sky Analytics Inc. on $2.3 billion in invoices billed by law firms in 2014 that recorded how lawyers spent their time and grouped the tasks into 13 different categories, such as document review, due diligence and legal writing. The pair then gauged how well those tasks could be automated based on interviews and studies on the effects of automation on other professions. The data did not include information on solo practitioners, who make up about 40 percent of practicing lawyers, which they hope to include in a later update, Levy said.
What they found was that at large law firms, most of lawyers’ time, about 56 percent, was spent on tasks like advising clients and courtroom appearances that could not easily be automated. About 40 percent of their time was spent on tasks such as due diligence and legal research that could be moderately displaced by automation and only about 4 percent was spent doing tasks that are easily automated, such as document review.In all, Remus and Levy estimated that under a worst-case scenario there could be a 13 percent reduction in lawyer employment due to law firms’ adoption of artificial intelligence and other technologies. If the adoption of technology were phased in over five years the volume of legal services would need to increase by at least 2.5 percent per year to offset the reduced demand for lawyers’ services, Remus and Levy said.
“The main thing is that the legal market is becoming more competitive, and this just makes it that much more competitive,” Levy said. “In essence what you’re getting here is equivalent to increasing the supply of lawyers in the market because it allows a given amount of lawyers to increase the amount of work they can do.”
Though the work performed by attorneys at all levels of seniority could be partly automated, any reduction in staffing is likely to be felt most by those at the bottom, Levy said.
“If there’s less business, who gets laid off?” Levy said. “That has to do not so much with skills and so on, it has to do with the hierarchy in the firm.”
Associates are already feeling squeezed by a contracting legal services market, and if that trend is exacerbated by technology it would put further pressure on BigLaw’s outdated business models, Furlong said. With law firms needing fewer full-time junior lawyers, the traditional pyramidal structure employed at large firms, with a few senior partners at the top supported by an army of associates at the bottom, could collapse, he said.
“This idea that you want to have one partner and four associates laboring under him or her, I think that’s going to run its course,” Furlong said. “It’s going to be very hard to convince a client you need four associates hammering away at keyboards when a machine can do some of that faster and cheaper and in some cases at a higher level of quality.”
Instead, Furlong envisions a future in which associates become more akin to independent contractors or temp agency workers who are hired by firms as they are needed for less money. If an associate is hired at a firm, it will be primarily for the purposes of grooming him or her into a future partner and firm leader, he said.
“I think ultimately we’re talking about redefining the whole structure of law firms,” said Robert Ambrogi of the Law Office of Robert J. Ambrogi, a former editor-in-chief of The National Law Journal who writes about legal technology for various outlets.
Although clients will not be willing to foot the bill for associates to do work that could be done by computers, that doesn’t mean associates won’t be needed at all, Ambrogi said. Artificial intelligence is not yet at the point where computers can be largely left to their own devices to complete tasks — and it is not clear it ever will be, Ambrogi said. There needs to be someone to check their work and to summarize results for review by senior attorneys, a role that could be performed by junior lawyers, even though not as many of them will be needed, he said.
Also, with computers handling the rote and repetitive tasks traditionally performed by associates, lower-level attorneys could pursue more highly skilled work earlier on in their careers, Ambrogi said.
“There’s always going to be a role for associates, but it’s a matter of market demand,” Ambrogi said. “Sophisticated clients are not going to pay for highly paid associates to be doing work that doesn’t even need a human doing it in the first place.”
There will still be a business need for associates at law firms because eventually someone will have to replace the attorneys who retire or leave a firm, said William D. Henderson of the Indiana University Maurer School of Law.
“We need associates because associates turn into practice experts,” Henderson said.
The minute a firm becomes made up entirely of the “50-year-and-older crowd,” it puts a “shelf life on your law firm,” he said. “You don’t want to do that.”
As artificial intelligence takes more work away from junior employees, not only will large law firms shrink, but midsize and boutique firms will be able to better compete with BigLaw, Ambrogi said.
Part of the business case for the megafirm is access to different tiers of experience and knowledge, packaging routine work along with highly skilled work, he said. With computers taking care of the scut work at the bottom level, that leaves the more sophisticated legal labor that can be provided by smaller firms just as well as larger ones, he said.
“As I look ahead in the future I see the big giant firms becoming fewer and more of that highly skilled legal work being done by smaller to midsize firms with more specialization and more use of technology,” Ambrogi said.
A New Model
To stay profitable in a changing legal arena, firms will have to change their compensation systems. As it stands, the billable hour places a hard cap on the amount of money that a firm can generate, Furlong said. In the past, that has not been a problem because firms have had large numbers of employees, but as the number of associates on one side of the equation goes down, the number of hours that a firm can bill will also drop.
One answer to what may replace the billable hour is a fixed-price model that charges a flat rate for services, Furlong said. That approach has already successfully been put into place by firms such as Seyfarth Shaw, Valorem Law Group and Littler Mendelson PC, he said.
“In 10 or 15 years’ time I don’t see how you can keep running a law firm without having done this,” Furlong said.
Some firms have based fixed-price pay models on so-called expert systems, a kind of artificial intelligence that replicates expert knowledge to offer legal advice according to users’ responses to a series of questions. Software company Neota Logic Inc. and others have developed expert systems tools firms can use to put together custom-built applications allowing clients to obtain on-demand guidance in a specific area, such as compliance with the Foreign Corrupt Practices Act.
Firms can charge a subscription to access the applications, which can answer the kinds of questions that clients typically ask over and over again, and the clients will come to the firm for higher-level analyses that experienced lawyers are good at and for which clients won’t mind paying hourly rates, Henderson said.
“This is where the law firms are realizing they can prosper in a world outside the billable hour because they still get to bill the higher-order work,” Henderson said. “This goes directly to the heart of reinventing the wheel of legal advice.”
Theo C. Ling, a partner at Baker & McKenzie LLP who has been the driving force in the development of the firm’s iG360 platform, a suite of expert systems and other collaborative tools, said the platform has allowed the firm to price its services more favorably to clients.
Clients can use the platform to answer routine questions, freeing up the time of attorneys who are not tied up on the phone, but it has not led to a layoff of associates, Ling said. In fact, because of the level of clients’ interest in the platform, one of his biggest worries in the short term is that he won’t have the staffing to keep up with demand.
“It’s quite likely in the next six months we are going to need to onboard more than 100 global clients, which raises questions about whether we can keep up” at current staffing levels, Ling said.
A move away from the legal industry’s stubborn adherence to the billable hour, however, will likely take many years as old ways of doing business are phased out and new ones are developed. Some capital-intensive firms may even be able to subsist on old models for a decade or more as they service clients who are loath to change the way they do business, Henderson said.
“For everyone else ... they’re going to have to figure out a better way,” he said.
Some are more sanguine about the prospects for BigLaw. Ron Friedmann, a partner at legal industry consulting firm Fireman & Co., cut his legal technology teeth at Wilmer Cutler & Pickering, one of the first firms to have almost every lawyer using a computer, he said. But through the rise of the PC and then the internet in the 1990s, it is never clear that the firm was rewarded for adopting the technology, he said.
The same will probably hold true for artificial intelligence, Friedmann said. The technology may result in incremental changes to the legal profession, but a momentous shift in the legal services market due to the implementation of artificial intelligence is not likely, he said.
For one thing, it has not been shown that any billable hours lost to automation cannot be made up elsewhere, such as with an increase in volume, Friedmann said. He pointed out that the 13 percent loss in employment that Remus and Levy predict in their paper is premised on all other things remaining the same. In reality, any market shift from the adoption of artificial intelligence is likely to be dwarfed by other factors, he said.
“The volume of transactions may go up, or it may go down,” Friedmann said. The recent decision by the U.K. to leave the European Union, for example, may put a damper on deals, or it may result in a surge of demand for legal services because of uncertainty, he said. “Those types of effects can easily overwhelm 13 percent over five years.”
Jean O’Grady, author of the blog Dewey B Strategic and past chair of the private law library special interest section at the American Association of Law Libraries, said artificial intelligence has been subject to a lot of hype when its existing applications don’t move the ball forward much compared with technology that has come before.
That is not to discount the advantages artificial intelligence can provide to lawyers who know how to use it correctly, but intelligent attorneys and lawyer expertise will always be needed to properly apply it, she said.
Artificial intelligence is “not something that can stand alone and deliver adequate answers,” O’Grady said. “It’s not a replacement for lawyers, it’s at best an adjunct to lawyers.”
Heart of the Profession
Though artificial intelligence may hasten a reinvention of the law firm business model, it will not replace the core of lawyers’ work. Remus and Levy write that the limitations of artificial intelligence and other automated technologies shine a light on the value of human lawyers and provide insight into the true meaning of legal professionalism.
The title of “The World’s First Robot Lawyer” may belong to an artificial intelligence program called DoNotPay that recent news reports say has already helped users contest some 160,000 parking tickets in New York and London.The automated “chatbot” program asks users a series of questions to figure out if the tickets can be appealed, such as whether there were clearly visible parking signs at the time of the violation, and gives users advice on contesting the tickets based on their answers, according to a June 28 report in The Guardian. In the nearly two years since its debut, the program has enjoyed a 64 percent success rate and has overturned more than $4 million in total fines, The Guardian said.The program’s creator is 19-year-old Stanford University student Joshua Browder. The London native is also developing a bot to help refugees apply for asylum that will use the IBM Watson artificial intelligence platform, and he is looking into developing applications to help attorneys create their own automated legal advice bots, according to tech news outlet Venture Beat.
The legal tasks that computers are lousy at and that humans can perform well — communicating with clients, counseling them, watching out for their interests, advocating for them in the courtroom — are at the beating heart of what lawyers do.
A computer may be able to gauge whether the Internal Revenue Service is likely to audit a given tax strategy, but only a human can give meaningful advice to a client about whether such a strategy is wise.
The legal profession may be profoundly changed by the adoption of new technologies, but lawyers may also find themselves freed of the tedium of performing boring and repetitive tasks and able to pursue more meaningful work that is ultimately more rewarding.
Some even see a future in which lawyers are happier because of artificial intelligence. Dennis Kennedy, an information technology lawyer, legal technology author and adjunct professor at Washington University School of Law, said he has long thought that a lot of the dissatisfaction in the legal profession stems from the amount of tedious work that lawyers are forced to do. An attorney doing document assembly, for example, may spend hours proofreading and formatting a document, he said.
If artificial intelligence can strip away such mundane tasks, however, then attorneys can spend more of their time doing the creative work from which they derive satisfaction, he said.
“From my point of view I see law as a creative profession, and moving in this direction is really attractive to me,” Kennedy said. “If my day’s work is running an artificial intelligence program that helps my client settle a case and save millions of dollars, and they’re happy to pay me whatever I want and take the next few days off, I think I’m going to be happy about that,” he said.
Or, as Arruda puts it: “Artificial intelligence doesn’t spell the end of law. Artificial intelligence spells a new beginning.”