Will AI Research Make Lawyers Stupid?

Artificial intelligence and machine learning are the future of the law, or at least of legal research, we're told. Hi-tech algorithms will soon reduce time spent flipping through irrelevant caselaw, allowing lawyers to pinpoint the best research in just seconds. There are a host of traditional research companies and legal tech startups vying to be on the forefront of this developing technology and AI-powered research has already been embraced, if cautiously, by a few of the country's biggest firms.

But not everyone is convinced that research powered by machine intelligence will actually yield intelligent results.

AI Research and Machine Learning

If you're unfamiliar with how AI-powered research and machine learning works, here's a quick, very simplified illustration.

Say, for example, that you search "appurtenant easement California." (Boolean searches are best, but we're all getting lazier and relying on natural search these days, it seems.) That search pulls up your results and you can get to work.

But you're not the only one who has queried that same or similar phrase. Machine learning allows the search algorithm to learn from your and others' results, and how you interact with them. Did you all end up on similar cases? Expect those ones to go to the top. Did you highlight and save similar holdings? Maybe the search results can jump straight to those.

So forth and so on until the search leads to the best answers, cutting down your work significantly.

But ...

How Intelligent Is Machine Intelligence?

There are a few problems with this process, which Seton Hall law professor Brian Sheppard recently detailed for the ABA's Legal Rebels. Here's his take:

Determining whether a lawyer understood and prioritized cases correctly is no simple matter; legal problems are not math problems. For one, identifying whether the top-ranked case was the best one for a particular lawyer's case requires knowledge of the facts of the lawyer's case. Facts are tough for research companies to access.

The "top ranking" results in your legal research platform might be the most popular, but that doesn't mean they're the best for your case. And if the search algorithm decides that a case isn't relevant, you may never know. It simply won't show up, even if it could actually be on point.

Secondly, clicking on a case, spending time on a particular issue, or printing off a stack of papers isn't necessarily a sign that the research is the best. If lawyers are giving machine learning programs "bad data," then the resulting algorithmic changes could be skewed -- and both lawyers and programmers can have trouble interpreting those algorithms.

"To some extent, this situation leaves both lawyers and research companies fumbling in the dark," Sheppard writes. "Lawyers don't have a complete picture of what is happening, and research companies are relying on the lawyers to teach their machines."

Searching for a Solution

The rush to embrace AI in legal research, with the goal of providing the quickest answer in the simplest fashion, could have two major consequences, Sheppard posits:

On the one hand, it could increase productivity, potentially leading to reduced lawyers' fees. On the other hand, it could increase error, which might occur when overaggressive, secret algorithmic choices cut out vital cases.

So, should we toss out the AI baby with its data-driven bathwater? Return to searches like "/n proximate caus! /10 (auto! or vehicle)"? Just go back to paper-based legal reporters?

Probably not.

Thankfully, these problems aren't unique to legal tech. They're the same issues that afflict research everywhere, from Google to the financial industry to academic databases. That means that many of the brilliant minds working on building AI are also working on making sure the results are good: appropriately simple or complex, depending on the audience or industry, and resistant to manipulation and bad data.

And, thankfully, lawyers have good minds, too. The lazy associate who picks the top result and stops there has always been at a disadvantage; so long as research platforms allow lawyers to dig deep, the most committed attorneys will still be able to find the best information, even if it takes a bit of work.

Related Resources:

• Will Smart Machines Make Us Stupid? AI Experts Weigh In (The Wall Street Journal)
• Millennials Don't Know How to Google. Do You? (FindLaw's Technologist)
• Microserf Proposes Solving Tech Problems With Medieval Law (FindLaw's Technologist)
• Clients Want Artificial Intelligence and They Want It Soon-ish (FindLaw's Technologist)