Artificial intelligence (AI) is permeating every aspect of our lives: it’s telling us what to buy and when we need to buy it, when we need to get up from our desk to take some steps, which emails to read using filtering techniques, and what to watch next on our smart device or TV.
While AI can make our personal lives much simpler, it also has the capability to revolutionize our approach to eDiscovery—if we take advantage of its benefits. Here are some examples of ways AI can transform your eDiscovery processes, from start to finish.
AI and Early Case Assessment
AI can reveal critical relationships between documents and people as well as important matter timelines—even before you’ve collected a single document for discovery. Early case assessment tools powered by AI often offer visualization tools that highlight patterns in communications so you understand the context of the case quickly and make decisions about which custodians to interview, which documents to collect first, what keywords to use to search for data, and, ultimately, whether to proceed with litigation or an investigation or resolve the matter quickly. In addition to building a case strategy earlier in a matter, using technology early in a case can lead to significant savings later in discovery by reducing the volume of data collected, processed, and reviewed.
AI and Processing
Parties can use AI early in the EDRM to cull their data sets. In fact, parties have reduced their data volume significantly—sometimes by 90% or more—by applying technologies such as these to their data sets:
· deNISTing: removing files unlikely to contain relevant information, such as system files;
· deduplication: removing exact duplicates;
· near deduplication: grouping similar documents together, such as emails that have been forwarded;
· other natural language processing tools such as concept clustering: grouping documents with related ideas together; and
· email threading: collecting related conversations together to avoid redundant work and inconsistent reviews.
AI and Document Review
Technology-assisted review (TAR) expedites review. There are two types of TAR: predictive coding and continuous active review. In a nutshell, with predictive coding, reviewers code a sample of documents, called a “seed set,” that is fed to a computer algorithm. The algorithm analyzes the document coding and “learns” from it, applying its learning across the data set. With continuous active learning, there is no seed set; the computer observes as human reviewers code and begins feeding the reviewers the documents that it deems to be the most relevant based on their past coding. The computer continuously learns, improving its results as the review continues. Both forms of TAR save both time and money—and yield more accurate results than human review alone.
If you aren’t already using AI in your eDiscovery workflows, it’s time to start. Want to learn more about how you can improve the accuracy of your results and reduce your litigation budget with AI tools for eDiscovery? Get in touch.