Earlier this year, Legaltech News sat down with two professionals from McDermott—Chris Adams and Martha Louks—that have used AI in e-discovery on its impact, practical uses, and more. Below is a transcript of the conversation, first published in LTN's June print edition.

It seems like artificial intelligence (AI) is the hot buzzword in the industry right now. Is this really something more than just TAR 2.0?

Chris: In my book, it is. We're not just talking about a way to speed up linear review, or even to improve TAR [technology-assisted review] workflows—we are looking at technology that will truly impact the way we evaluate, categorize and utilize data. Unlike TAR, AI isn't merely taking coding decisions and promulgating them to the data set; it's about revealing features in the data that we may have never considered before, bringing powerful new tools to how we analyze that data and build case knowledge. But this is really Martha's area of expertise, so I'm sure she's going to have deeper insight than I will.

Martha: Essentially, AI is reading vast quantities of data in a tiny fraction of the time that it would take a whole team of people to go through the same data. It bubbles up patterns, anomalies and key concepts in the documents. This means you have immediate knowledge about your data in a way that wasn't possible before. Because the technology combing through the data is more powerful than ever, it has really moved predictive coding forward—we can train the system using this enriched data to get results in a fraction of the time compared to previous TAR approaches. Beyond predictive coding, AI is helping us peer inside the infamous “black box” to get at what's most important in a case more quickly.

What other advantages do you see from using an AI tool during discovery?

Chris: What I really focus on here is how we can utilize AI to bolster fact development from very early on in the discovery process. AI can help us more quickly uncover the type of information that can impact everything from motions practice to deposition preparation to settlement negotiations—and the faster our legal team can get to that data, the faster they can analyze it and make important strategic decisions that impact the entire litigation. That's huge.

Martha: Agreed. Just to give a recent real-world example, we had a case where we were on the hunt for hot documents related to a particular topic. A traditional keyword search turned up hundreds of documents that were mostly false hits. We then switched to AI to run a different kind of search across the same data. We looked for high-pressure emails that had a negative sentiment, and revealed a much smaller set of docs, about a quarter of which were actually hot docs. Needless to say, the attorneys doing this work were pretty happy with the results.

How do you see AI technology impacting the e-discovery industry? Do you still see barriers to entry for adoption?

Chris: I think it finally reshapes the landscape in terms of how data review will be conducted. For years, we've been hearing about how TAR would change the equation from large teams of contract reviewers conducting massive data review projects to a small team utilizing technology to accomplish the same thing. The advantages there are obvious—both in terms of cost and the quality of review. But the problem has been that, from a time standpoint—which can often be just as important a factor as cost in litigation—TAR took a long time, and often was abandoned as a tool because the clock ran out. So while some have pointed to practitioners and courts not being comfortable with technology as a reason that TAR wasn't adopted more often, my experience was that time was a much bigger enemy in terms of adoption than lack of understanding.

Martha: Time has definitely been an issue in the past. TAR always saves time, but there's a difference between saving time by reducing the total number of hours needed to get a review done versus reducing the number of days a project requires. There used to be this tension between scaling up by adding more reviewers to get the work done quickly and using TAR, but with the advancement in technology, we don't really have that problem anymore.

Q: Much has been made of the potential cost savings of using AI during the review process. Are you seeing that?

Chris: Absolutely. If you look at the substantial impact in terms of the time savings in reaching stability in data sets in a TAR workflow through the use of continuous active learning alone, it's already been incredibly impactful for us. It used to take up to a couple of weeks or longer to reach stability in a data set, where the system has decided that no further rounds of coding are needed to train the algorithm. On our last project, this was done in what, Martha—four days?

Martha: Actually, it was more like 2.5 days. It used to take anywhere from 10 days to three weeks to train the system, with subject matter experts reviewing around 12,000 to 15,000 docs, but we're now seeing results in a couple of days after looking at only a couple thousand documents. Not only are we able to reduce costs by defensibly eliminating a large number of documents from review and production, as we always have with TAR, but we're also able to save money by reducing the time required for subject matter experts to train the system.

Q: What impact, if any, will AI have on litigation practice inside law firms?

Chris: Martha and I were just speaking about this earlier—the possibilities here are pretty enormous. For now, we are focused on how this technology can take what has become a bifurcated process—with case and fact development occurring in one silo, and data review in another—and bring them together. When we put this new technology in the hands of the practitioners and allow them to learn first-hand about the facts of the case while effectively training the system to find relevant data to meet discovery demands, the downstream efficiencies are enormous. There's no better way I can think of to substantially impact litigation costs.

Martha: Absolutely. By leveraging AI technologies, our attorneys get better information about their case much more quickly. This drives better decisions and, depending on what they're able to learn from the data, this could have a big impact on case strategy and even help them identify avenues for better settlement negotiations. That's a big win for our clients.

Chris Adams is chief strategic counsel of McDermott Discovery, and leads the in-house e-discovery team. He was previously the head of discovery consulting and strategic consulting at Consilio. Martha Louks is McDermott Discovery's director of technology services and focuses on emerging technologies and how they can be integrated into the practice of law to increase efficiencies and improve results for clients. Both work in McDermott, Will & Emery's Washington, D.C., office.