The potential and capabilities of ediscovery software have been greatly enhanced by ediscovery AI. AI in ediscovery has improved the speed, accuracy, and reach of ediscovery efforts while decreasing the cost associated with complex litigation.
At Now Discovery, our Lumix platform makes the most of these advancements while maintaining the affordability and user-friendliness we’re known for across the industry.
eDiscovery means “electronic discovery.” It refers simply to the identification, collection, and production of electronic data in response to a request during litigation or investigation.
AI, or artificial intelligence, refers to the development of automated systems that can do the things we normally rely on humans to do. For example, an AI system might categorize images based on what they depict, convert speech to text, or make decisions based on algorithms.
In the context of ediscovery, AI can refer to several different technologies designed to improve the collection of data during discovery efforts or decrease the costs associated with that effort. It’s primarily used in the subfield of document review.
Machine learning refers to software that automatically improves as it’s used. Relying on previously processed data, machine learning software can make predictions about or classify data in an increasingly accurate way.
Because of this ability to classify data, machine learning ediscovery is frequently used in document review. Called TAR, short for technology-assisted review, ediscovery AI in document review dramatically speeds up and reduces the cost of reviewing large numbers of documents.
Machine learning is part of what makes it possible for users of our Lumix platform to assess, organize, and review tens, or even hundreds, of thousands of documents in relatively little time. The software accurately predicts the correct classification of an individual document so you don’t need to pore over each email and word document individually.
Predictive coding refers to the practice in ediscovery AI whereby a human operator selects a sample of relevant documents from a larger set and an AI identifies further relevant documents based on the characteristics of the first hand-picked collection of data.
It is the primary method used by ediscovery AI in document review. As discussed in the previous section, this form of document review is called TAR, or “technology-assisted review.”
As an example, a user of Lumix might manually identify 15 emails relevant to a particular issue out of a field of tens of thousands. Predictive coding allows our software to examine the 15 pre selected emails, narrow down the common characteristics of those emails, and use those characteristics to automatically locate all the other relevant emails in the entire group.
(If you’re curious about how ediscovery AI manifests in the Lumix platform, check out this short article about the Five Pillars of Lumix.)
Issues of artificial intelligence and technology-assisted review have begun to make their way into the courts. For example, in FCA USA LLC v. Cummins, LLC, No. 16-12883 (E.D. Mich. Mar. 28, 2017) the United States District Court of the Eastern District of Michigan (Southern Division) opined that TAR should occur before a keyword search is applied to the universe of available ESI.
In other words, rather than first apply a keyword search to all available ESI and follow that with a technology-assisted review, the producing party should apply TAR to all available ESI and, if necessary, apply a keyword search afterward.
In a word, yes. Especially in the subfield of ediscovery known as technology-assisted review, ediscovery AI can dramatically reduce the costs associated with document review, increase its accuracy, and decrease the time required to analyze the frequently large number of documents requiring review.
In addition to saving time and money, countless valuable person-hours belonging to junior counsel, students, and paralegals can be put to much more profitable use than reviewing thousands of documents of dubious relevance.
eDiscovery AI is not a magic bullet or secret elixir that will completely obviate the need for human intervention in the ediscovery process. While AI document analysis and review can be a tremendously helpful adjunct to human involvement, skilled counsel will ultimately need to piece together the relevant documents compiled by the review into a cohesive narrative.
You should also be cautious about ediscovery providers who treat ediscovery AI like a gimmick or an end in itself. At Lumix, we’ve enthusiastically incorporated AI into our software where it adds legitimate value, but not at the expense of affordability or useability. (Curious about just how affordable Lumix can be? Drop us a line and we’ll show you!)
We haven’t even begun to touch on many of the major topics that currently consume the ediscovery AI space. For example, we haven’t discussed the role that natural language processing plays in document review, or what automatic speech recognition (ASR) can do to simplify the review of audio and video data.
At the end of the day, you’ll need to remember that ediscovery AI is one of many tools in your ediscovery toolbox. It’s an especially powerful one, but it’s a tool nonetheless. It is not an end in itself. Rather, it’s a way of unlocking the potential of your human capital by allowing them to focus on tasks that computers can’t (yet) perform.
At Now Discovery, our Lumix platform makes extensive use of artificial intelligence to make the job of eDiscovery much faster, more reliable, and easier. We only rely on tested and proven AI techniques to help you discover the information you need. Try us for free today!