Humans vs Robots

This article first appeared in Corporate Counsel and The American Lawyer's December issues under the title “Inside Legal's (Over?) Excitement with AI.” The full version is reproduced below.

Artificial intelligence will replace lawyers completely. It will kill off the billable hour and allow for instantaneous review of any agreement or document set. And all one has to do is turn it on and set it loose.

Sound familiar?

It should. Such proclamations on the death—and reinvention—of the legal industry at the hands of AI are frequently printed in advertisements, written in press releases and think pieces, and spoken at conferences and even client meetings. At times, they can seem almost omnipresent.

But those who have actually worked with AI know that such declarations take broad liberties with the truth—essentially making AI more exciting, consequential and disruptive than it really is in today's world.

Of course, hype has always been a part of any new technology. A Hype Model by research and advisory firm Gartner, which shows hype peaking when a new technology is introduced, plummeting when its limitations are discovered and rising to a plateau after it becomes more well known, is widely regarded as the way such expectations evolve.

So where is AI in the hype cycle?

“I would say that it is still overly hyped,” says Aaron Crews, chief data analytics officer at Littler Mendelson, who was the former general counsel at AI startup TextIQ and the former global head of e-discovery at Walmart Inc.

Crews notes that such hype can oftentimes come from legal tech companies promoting their AI products. But it also can come from legal service clients themselves, including legal departments, who buy into the exaggerations.

“The hype kinds of feeds the client conversations, and in some ways it creates a circular conversation” when vendors over promise what AI can do to meet a clients' unrealistic expectations, Crews says.

While many will agree there is still AI hype—which has had both positive and negative effects on legal's tech evolution—they also note that it's on the downtrend. As legal professionals get hands-on experience with AI platforms, they say, attorneys are beginning to better understand the technology's real limitations.

But few, if any, dispute that AI will fundamentally change the legal industry. Despite all the hype, Crews notes that AI will “be a driver for significantly reinventing the business of law.”

Many are simply coming to realize that the speed at which this reinvention will occur, and how it will occur, is too often overstated and likely still widely misunderstood.

|

Let's Get Real

While AI technology was once something only bold, well-funded law firms and legal departments could experiment with, today it's widely accessible as the engine that powers a host of contract review, legal research and work flow automation platforms. And, as more and more of these products find their way into legal offices, attorneys are realizing that the promised scope of AI's abilities might not be entirely accurate.

“I do feel like there is not as much hype as there was a couple of years ago,” says Brad Blickstein, principal at legal consultancy firm Blickstein Group, which has helped numerous legal departments implement technology in-house and streamline their operations. He adds, “There seems to be a growing awareness that AI itself is not a solution” but rather another tool at attorneys' disposal. And, like any tool, AI can't be used in every situation.

Take, for example, technology-assisted review—essentially the use of AI to automate document review. “Everybody said, at least early on, that TAR would radically transform how document review was done, dramatically change pricing and shift that balance in a positive way,” Crews says. “While I think there are folks who are deploying it and achieving that outcome, if you look at a lot of the legal space, that prediction has largely not materialized.”

The reason, he explains, is that the attorneys' use of TAR in pretrial e-discovery isn't currently supported by the majority of case law and courts. And in many cases where TAR use is allowed, Crews notes that “you have to turn over your seed set, including nonrelevant, nonresponsive documents”—a new level of transparency e-discovery practitioners may not be comfortable with.

For Crews, the reality of TAR's restricted use points to a broader problem with AI: So long as the technology can't realize its full potential in practice, it won't live up to its hype. “If I have a Ferrari that can go 300 miles per hour, but I live in place where the maximum speed is 60 miles per hour, then a conversation that says a Ferrari will get me somewhere two hours faster is absurd.”

To be sure, resistant courts and rigid case law aren't the only reasons AI can't run wild. Arup Das, CEO of managed legal services company Alphaserve Technologies, notes that past hype initially looked to define AI as a technology that one turns on and “boom your contracts get done, your document review is done or litigation [work] is done.”

But attorneys are realizing that such automation comes at a high upfront cost. Document and contract review products that leverage AI are essentially machine learning platforms, which need to be trained again and again to be able to identify relevant information in different data sets.

“When people do actually get some of these platforms, obviously then they realize not only is there training, but the training needs to be continuous,” Das explains. “That means they get into this vicious cycle of cost, because every time you need to train and make additions to the training data set, it's not free.”

|

Finding the Holy Grail

While the actual implementation of AI has helped attorneys separate some of the hype from reality, “the hype still remains on very sophisticated [AI] applications,” Das says.

AI-powered prediction tools—such as those that claim they can predict the outcome of a case—are commonly discussed as a revolutionary use of AI. For now, these tools are limited in scope.

“Because the amount of data you would need to actually train a set of algorithms to do that kind of work and, because of the number of inputs that go into making predictions … those things are hard to grab and quantify,” Crews says. He adds that he believes those predictions will come, ”but I think that's the holy grail right now. And where that tech is going, I think that is a bit of a ways off.”

Likewise, the use of AI to essentially profile people to determine whether they should be hired or promoted is discussed as a real possibility. But in practice, it's a different story. Using AI programs “to figure out what makes a star lawyer, what lawyers are going to leave, what lawyers are going to make partner … it's very difficult, because you need a lot of data, data you don't have,” Das says.

And, even if one obtains all the necessary data to build such profiling systems, there's still the problem of AI bias. “There is a lot of stuff that we are now realizing. … If you don't look at patterns carefully, your outcome will be weighted with bias,” Das says. As an example, he notes that Amazon had to recently give up on an experimental AI-powered recruiting tool built to help it screen potential candidates after it was found to be biased against women job applicants.

Of course, it's not hard to understand why some believe such sophisticated uses of AI are possible. The term AI, after all, elicits images of hyper-intelligent sentient machines. So the biggest proof AI hype still exists may be that, well, it's still called AI.

“If you use the word AI and machine learning interchangeably—it's hype, because there is no AI, there are no robotic lawyers,” Das says. He adds, “I call [AI platforms] decision-aiding tools. They're not decision-making tools.”

|

The Good, the Bad and the Truth

By pushing some to believe AI is a far different technology than it actually is, though, the hype may have stifled the legal industry's ability to implement change.

The excitement for AI, for example, has made some legal professionals want AI for AI's shake. “The biggest issue is that people are running around [saying], 'Our law firm or legal department needs to use AI,' but they're not solving problems,” Blickstein says. He adds, however, that such an attitude is slowly changing as attorneys get their hands on the technology.

For some legal professionals, the hype has also led to the notion that AI is the single most important innovation to have. But Ron Friedmann, a partner at Fireman & Co. who has worked with law firms and legal departments for over 20 years to improve their operations with technology, says that perception is simply not true.

“I think there was so much focus for a period on AI that management forgot that there are a lot of basic tools that, if better used and better adopted, would gain benefits that would equal or surpass that of the AI tools,” Friedmann says. “If more lawyers learned how to touch type and how to use [Microsoft] Word properly, you would probably see more productivity from that than the limited use of AI.”

Some attorneys may have already realized this, but there isn't much to suggest a backlash of diminishing investment and interest in AI. If anything, one big positive effect of the hype has been to make attorneys more willing to explore how technology adds value to their work.

“I've been around legal technology a long time. I never remember a time when the bosses of law departments and law firms were going to their tech and operations people and saying, 'I heard about this cool technology. Please look into it,'” Blickstein says.

Without being excited by hype, many of these legal teams would not have had an impetus to explore AI use cases or, by extension, become more open leveraging technology in their operations.

|

It's Not Artificial Hype

Once you move beyond the hype, AI inevitably loses some of its luster. “We are far away from robot lawyers,” Friedmann says, explaining that current AI use cases in legal are “fairly narrow.”

But this doesn't mean AI isn't groundbreaking. After all, the level of automation and efficiency AI can bring to work like document review and legal research is unprecedented. So it's little surprise that law firms and legal departments are pushing ahead in looking for ways leverage AI across their operations.

Of course, in some ways, law firms and legal departments have little choice but to leverage AI. It's one of the best tools to help them become a leaner, most efficient operation—something the legal market is demanding more with each passing year. Blickstein notes that, while the billable hour still reigns in the market, for law firms that use AI for things like contract review, “the billing model does not make sense for them” anymore. They have found a way to adapt to client demands for alternative fee arrangements, and AI has played a critical role in that evolution.

Yet it also has done more than that. In addition to helping attorneys become more technology conscious, it is teaching them how to more critically think and talk about technology. And, mostly importantly, to approach it with a discerning eye that separates hype from fact.

It is a lesson some are still learning, but it is one that is vital, if legal professionals ever want to get to the real excitement of where the rubber meets the road.