NEW YORK — At Legalweek 2018, “artificial intelligence” is in the air. It's what all the lawyers want to know about, and what almost all of the dozens of legal technology vendors seem to be offering (better than their competitors, of course).

It's the distillation of a buzz that has long been building. But for all the hype, it's struck me throughout the various panels and presentations here that few in the legal community have a real understanding of what's being talked about. Part of the problem is definitional. As Catherine Krow, a former Big Law litigator who founded software company Digitory Legal, put it: “My caution in using this term is that it brings to mind Skynet.”

Here are a few of the things I've heard so far at the conference about artificial intelligence and the legal profession, and some of the unanswered questions.

|

AI excels at discreet tasks. And it's only as good as your data.

In one panel I attended at the LegalCIO track, a member of the audience asked: Will there be a moment in the near future when the best litigator is a computer? The answer was: “No, but …”

Andrew Arruda, CEO of legal technology company Ross Intelligence, drew a distinction between AI, which is skilled at automating a discreet task, and “general intelligence,” which is able to respond to varied circumstances in different or unfamiliar contexts. General intelligence, he offered, is still miles down the road.

The best litigator will still be the smartest and most cunning human, someone who knows how to use AI tools most effectively to his or her advantage.

Adding a caveat to that was Brian Kuhn, the global leader and co-founder of IBM Watson Legal. Kuhn envisions—and it sounds like IBM is implementing—the creation of “cartridges” of specialized legal information that can be deployed for various legal tasks. That's a mouthful, I know.

But imagine this: A firm that specializes in antitrust law “trains” an AI algorithm to interpret documents relevant to that practice area. Then, the firm sells that piece of trained software, allowing a firm weak in antitrust to gain capacity (and removing the need, perhaps, to bring on a bunch of antitrust partners).

Another point hammered home here: AI is only as good as the data it's trained on. Krow referred to this as the “garbage in, garbage out” problem. Arruda adds that it's not just having sufficient “Big Data,” it's whether that data is usable in its current form. “Clean data” is the new buzzword.

|

The legal technology industry is still trying to build lawyer trust.

There's a sense of frustration among vendors that lawyers, even when presented with hard metrics showing accuracy, still inherently mistrust AI and favor human review. One programmer vented to me on the sidelines of the conference, saying he's constantly wondering, “Why do you think a first-year associate can do it better?”

Part of the issue is legal ethics. Krow noted that when a law partner uses an AI tool, the advice they ultimately provide still has to be derived from their independent judgment. She pointed to Rule 1.1. in the ABA Model Code, relating to the “duty of competence.” Just like when she entrusted a task to a junior associate, Krow said, her mentality was “trust but verify.”

Nicole Eagan, the CEO of AI cybersecurity firm Darktrace, said building trust is designed into how the company deploys its tool. During the early days when Darktrace's software is monitoring network traffic and detecting anomalies, it gets the company's (or law firm's) IT people involved to see what the software is picking up and the actions it recommends. Once there's comfort that it's working correctly, the AI runs on its own, she said.

The pressure to build lawyer trust, both to explain the technology and drum up business, led to at least one high-tech showdown here, between Ross Intelligence and Casetext. Check my colleague Rhys Dipshan's Twitter feed for that.

|

AI will put pressure on law firms. But not in the way you think.

There will not be “robot lawyers.” But some legal technology companies are targeting in-house counsel with analytics tools that can crunch billing statements en masse, allowing them to better compare costs for the same legal services across firms. That's going to allow them to push back harder if a billing statement comes in sky high.

IBM Watson Legal has already rolled out a tool like this, mainly targeting the insurance sector and helping insurers contain legal spend. But the technology is being rolled out for other applications as well, and companies such as Bodhala are also in this space and allowing companies to hold outside counsel more accountable with data.

Follow me on Twitter @benghancock and check the hashtag #Legalweek18 for live updates from the conference.