Artificial Intelligence agsandrew/Shutterstock.com.

Artificial intelligence's (AI) potential as a game changer for legal is driving interest in the technology. But in the face of high implementation costs and an unfamiliarity with AI, deployment of AI is slow-going.

But for those already using AI, deploying the technology takes patience, understanding and, often, a whole new set of staff and processes. At the “Of Man & Myth: A Decade of Artificial Intelligence in the Law” session at Thomson Reuters' 16th annual Law Firm COO & CFO Forum, law firm and technology professionals explained what it takes to get AI projects up and running, and how to set oneself up to better use AI in the future.

Start Slow to Understand AI's Effects, and Cost

AI is still relatively new in the legal industry, and its effect on law firms' business and operations is not yet widely grasped. To understand the broader changes AI may bring, Andrew Shimek, president and chief operating officer at Neota Logic, advises law firms to take it slow rather than dive in.

“Think about, 'What are the ways that this is going to enhance your business model,' or 'What are the ways this is going to change your business model,'” Shimek said. For many law firms, thinking through how this will change their reliance on the billable hour or affect their staffing levels can be a challenge.

“That can be the hardest thing to understand up front,” he said. But starting slow with AI projects is not only useful to understand the broader changes it may bring, but also to realize the work it takes to get the technology up and running in the first place. After all, AI platforms require not just installation costs, but also considerable training before the technology can be used effectively.

T. Mark Flanagan Jr., U.S. chief operating officer at Dentons, noted his firm “usually first tests AI with clients sometimes to see if it's going to work for them.” An initial practice engagement allows clients to better understand that “there is no quick hit with this, it takes time to develop.”

Likewise, Neota also encourages clients “to start with some of the least sexy problems that are high volume,” such as nondisclosure agreement contract review, Shimek said. Such simple projects, he explained, require less complex training and can be launched relatively quickly.

Start With a Nonattorney User Base

One way to better understand AI is to test the waters with a quick pilot project. For Fenwick & West, who have a number of AI projects in-house, this meant experimenting with what the technology can do.

“In addition to traditional projects, we looked at smaller low-cost projects in our Fenwick Lab where we could fail fast and learn quickly,” said Scott Pine, chief operating officer at Fenwick & West. And a key part of this quick experimentation is keeping attorneys away from the pilot projects, at least initially.

Attorneys are “just more skeptical than the normal population,” when it comes to technology, he said. Moreover, platforms designed for legal services require “a lot of process” and effort to build. By excluding attorneys, one can have flexibility to “just play with and learn how can we use this technology” on a smaller scale first.

Instead of its lawyers, the law firm focused on the administrative staff as its target user base. They asked this staff what were the most frequently asked questions they get from attorneys internally, and aggregated all responses in a knowledge management (KM) database.

Leveraging the database, which held the top 100 questions, and AI and voice-command developer tools provided by Amazon and Google, the firm came up with a prototype solution that allows users to ask a question and receive an automated answer, via email or a voice-active device.

While the platform is still currently in development, “We're working to get 1,000 questions into this system by the first quarter of 2018, then we're going to roll it out only to administrative staff to just test it and see if it works,” Pine said.

From there, “if people think it actually delivers value then we'll roll it out to attorneys,” he added. “It will be relatively easy to hook it to have legal systems down the line.”

Get the Right People in Place

Implementing new technology in-house doesn't just mean installing new software and hardware. Experts need to bring the systems together, feed it the right data, and perform routine maintenance and upkeep.

And for AI projects, companies need to have developmental engineers working with IT to guide them. But that is just the start. As AI platforms depend on data intake to learn, those looking to implement AI need to streamline their data-collection processes and repositories before they can use the technology. AI contract solutions, for example, are only useful if a team already has relevant contracts in a digital format, and can access them from databases through established, well-documented processes.

Before AI can be installed, “vendors need us to map [out] that process,” Pine said. And to that end, “The folks [in a law firm's] knowledge management team need to understand how to map out process,” he said.

Having a well-equipped and knowledgeable KM team is only half the battle, however. Shimek noted that since some law firms may need to build centralized databases and processes from scratch, it also may be necessary to bring in business analysts to design more effective workflows in-house.

“They can help build out the framework for how you are going to apply user inputs and outputs for the data you have,” he said.

So far from just a simple “set-it-and-forget-it” technology, AI may require an entire retooling of a law firm's operations — and an entire new set of staff.