publication of the International Legal Technology Association
Issue link: https://epubs.iltanet.org/i/535467
ILTA WHITE PAPER: JUNE 2015 WWW.ILTANET.ORG 26 The accompanying story was only slightly less alarmist than its sensational headline. It warned: "Computers are getting better at mimicking human reasoning…and they are claiming work once done by people in high-paying professions." However, computers are unlikely to replace lawyers or any other legal professionals anytime in the foreseeable future. Even the smartest machines cannot emulate the judgment of a trained professional or function without human guidance. On the other hand, intelligent machines are increasingly becoming critical "members" of legal teams. They are not replacing professionals on the team; machines are making professionals more effective and efficient. Nowhere is this more evident than in e-discovery, where technology-assisted review (TAR) is enabling legal teams to handle cases otherwise too vast or costly to get through. Having TAR or any other form of artificial intelligence (AI) on your team does not mean you have to take a computer to group lunches, but it does require you to re-examine your established tasks and workflows. What worked best in the past will change when AI joins your team, and some small accommodations for your electronic teammate can reap big rewards. A BRIEF BACKGROUND ON TAR TAR is machine intelligence that uses software trained by human feedback to find relevant documents quickly and cost-efficiently. First introduced to the e-discovery market in 2009, TAR's initial adoption was slow due to concerns about its defensibility. With its continued evolution and uniform acceptance by the courts, TAR is now part of the e-discovery mainstream. The first TAR processes (TAR 1.0) were designed to dramatically reduce review time and costs with the promise of transforming the economics of e-discovery, particularly in big data cases. The initial technology minimized the volume of data and intelligently analyzed content, reducing the need for manual review and providing a more accurate picture of the data. TAR 1.0 systems were limited to providing a static process. Because sampling for both training and measuring performance happened only once in the beginning of the review process, there was no easy way to accommodate rolling data uploads or to use reviewers' ongoing judgments to improve the accuracy of the algorithm. Next-generation versions of TAR (TAR 2.0) take cost savings and intelligent review further. TAR 2.0 allows for continuous learning throughout the review process. Review team judgments can be continuously fed back into the system, enabling the system to get smarter about the document population and improve accuracy. TAR 2.0 also allows additional documents to be added at any time. Continuously updated ranking produces better results, meaning fewer documents to review, fewer attorney reviewers, decreased review time and increased cost savings. WHEN MACHINE INTELLIGENCE JOINS YOUR PROFESSIONAL SERVICES TEAM As a starting point, it is helpful to separate tasks into those that humans do best (such as reading for comprehension or making relevance judgments) and those that machines do best (such as recognizing patterns across large volumes of data).