AI for eDiscovery: Avoid Common Pitfalls

December 7, 2023

|

Modern AI can unlock new levels of accuracy and efficiency in eDiscovery, especially during document review.

But in order to realize these benefits, legal teams must choose and implement AI tools with care. For instance, a tool may sound impressive or have the latest features, but does it actually change your work for the better? Once you’ve adopted a tool, how do you ensure you take full advantage of it?

Making smart choices about AI starts with understanding basic terms and applications.

From there, be on the lookout for three common pitfalls:

1. Choosing tools with appealing features but low ROI

2. Avoiding expert consulting

3. Being too rigid to try new workflows

Pitfall 1: Choosing tools with appealing features but low ROI

There’s no doubt that ChatGPT has changed the world’s impression of AI. The speed, detail, and sense of authority you get from generative AI is truly astonishing and has spawned a new generation of tools aimed at increasing work productivity.

But legal teams must carefully consider the use cases for those tools before diving into the generative AI pool. Many off-the-shelf or free generative AI products are great at creating content but were not meant to be used in a legal context—which is how attorneys have found themselves inadvertently citing “hallucinated” case law made up by generative AI. For eDiscovery and other legal use cases, teams should ensure that AI solutions are created securely and trained on specific data to get better and more accurate results.

And before onboarding any secure AI, legal teams should still stop and ask: What’s the ROI for this tool? Will it help substantially increase productivity or solve a persistent challenge?

For instance, a generative AI chatbot used in eDiscovery may have some alluring features. But will it help you reduce eyes-on review? Can it reduce risk by accurately identifying all sensitive data in a dataset?  

Sentiment heat maps for discovering hot documents and toxic language are another good example. Their visualizations may be appealing, but the out-of-the-box categories are often too high level to be helpful in finding sensitive or hot documents in specific matters. Other solutions—such as custom-built key document identification and toxic classifiers—are less flashy but deliver more substance.

If a tool doesn’t help you move the needle on your specific eDiscovery goals, it isn’t worth it—no matter how “powerful” and interactive it claims to be.

Pitfall 2: Avoiding expert consulting

Legal teams are often attracted to tools they can use out of the box, because they worry that tools coming with teams of hands-on experts also come with large price tags. But plug-and-play tools often have limited features and capabilities.

The truth is, experts can help you harness the most beneficial capabilities of AI, and relying on their expertise can actually reduce review timelines and costs. A quality technology provider will pull in data scientists, linguists, and AI experts to support projects at key moments—implementing tools, refining search parameters and classification models, and making other adjustments that maximize the AI’s performance and results. They can also explain workflows and results (including providing expert testimony) and provide insight into how the tools can be used to maximize investment across matters and legal portfolios.

While these experts make the team bigger, you don’t pay for the whole team the whole time. Instead, you get exactly who you need, exactly when you need them.

Experts help push your AI’s accuracy, efficiency, and insights to the max—saving time and money during document review.

Pitfall 3: Being too rigid to try new workflows

Legal teams may prefer to keep all their old, familiar review workflows, but AI often presents an opportunity to achieve better results with modified workflows.

For example, attorneys may prefer a “family-level” review process, where reviewers decide how to code a document based on the content of other documents in the same family. But AI-based tools get better results with a “four-corner” review process, in which each document is analyzed individually.

By using a four-corner process, modern AI will recognize a blank attachment as junk and code it that way—even though an attorney coded it as responsive and privileged during family-level review because it was attached to a responsive and privileged email. Without this process, you are not fully reaping the benefits of AI, particularly the ability to apply learnings across matters via the model.

Legal teams that adapt to review process changes will maximize the value of technology, including modern AI—and be glad they left their old workflows behind.

Understanding commons pitfalls when implementing new technology can help teams have more successful outcomes, especially with AI. Rather than learning by trial and error, which can lead to high costs and delays, keeping these three issues in mind from the outset of adopting AI will help you to unlock greater accuracy and efficiency in document review. As new AI tools and products continue to enter the market, this can serve as vital framework for evaluating what solutions you need and what’s just hype.

Learn more about strategically implementing AI for your matters on our AI and analytics page.

About the Author

Sarah Moran

Sarah is an eDiscovery Evangelist and Proposal Content Strategist at Lighthouse. Before coming to Lighthouse, she worked for a decade as a practicing attorney at a global law firm, specializing in eDiscovery counseling and case management, data privacy, and information governance. At Lighthouse, she happily utilizes her eDiscovery expertise to help our clients understand and leverage the ever-changing world of legal technology and data governance. She is a problem solver and a collaborator and welcomes any chance to discuss customer pain points in eDiscovery. Sarah earned her B.A. in English from Penn State University and her J.D. from Delaware Law School.