FEATURE ARTICLE the AI “QA” (question answering) system that defeated Jeopar- ics platform and related applications, Lex Machina can assist dy champions Ken Jennings and Brad Rutter in 2011. In 2015, corporate counsel in determining how likely a judge is to grant AlphaGo, an artificial intelligence system powered by Google’s or deny a particular motion, estimate the approximate length DeepMind technology, defeated legendary player Lee Se-dol at of a particular litigation matter, and determine how likely a the abstract Chinese strategy game Go – and, this year, Alpha- particular judge is to find in his client’s favor. Lex Machina Go defeated reigning Go world champion Ke Jie. Tech-savvy even provides data on opposing parties and attorneys, informcompanies have begun integrating highly visible, user friendly ing counsel in key strategic decisions. Lex Machina initially AI systems, such as Apple’s Siri and Amazon’s Alexa, into main- focused on intellectual property litigation, but is expanding stream electronic products, enabling smart phones and other to provide similar sophisticated data collection and analytics gadgets to respond to certain natural language commands, an- for securities and antitrust litigation. Lex Machina is already swer questions, and respond to other verbal prompts. These in use at law firms and companies across America, and is not applications, while impressive, only scratch the surface of what alone in the AI-powered legal analytics marketplace. Competartificial intelligence technology can do. ing products exist, marketed by companies like Premonition, Given the increased availability of products integrating which provides AI solutions for legal professionals, as well as a AI technologies in recent years, diverse array of other fields, init should be no surprise that cluding insurance, finance, and COMPAS applied (and challenged) products marketed specifically lobbying. to law firms and legal profesAI technologies are not In Loomis v. Wisconsin, now before the US Susionals have emerged. One such just for large scale, high stakes, preme Court on Loomis’ petition for writ of product is called ROSS, and is multi-million dollar litigation certiorari, a Wisconsin court sentenced Loomalready in use at many promimatters. Another example of is to six years in prison based in part on a renent law firms, in-house legal AI’s impact on the legal world port generated by a computer program called departments, and law schools. is COMPAS, a computerized acCOMPAS, which uses proprietary software to Like the machine that beat Ken tuarial risk assessment system forecast the risk of recidivism. Having been deJennings at Jeopardy, ROSS is that uses both static and dynied access to the program’s secret algorithms, powered by IBM’s Watson technamic data relating to criminal Loomis argued the court’s use of the COMPAS nology. ROSS, marketed as “the offenders to assess their risk of report violated due process. The Wisconsin world’s first artificially intellirecidivism, violent conduct, and gent attorney,” draws upon a other behaviors, and to predict Supreme Court rejected that argument and comprehensive and constantly outcomes. This technology upheld the sentence, prompting Loomis’ petiupdated database of legal auis already in use by some state tion to the US Supreme Court. The cert. petithorities, and responds to legal courts and departments of cortion is on the Court’s June 22 calendar. questions presented in natural rections to aid in critical decilanguage. ROSS performs legal sions like sentencing, bail, case research tasks with speed, effiplans, and allocation of rehabiliciency, and comprehensiveness, eliminating many hours nor- tation and supervision resources. mally spent by lawyers on information retrieval and analysis, Rule 1.1 of ABA Model Rules of Professional Conduct inand enabling those lawyers to focus on tasks involving creativ- structs that “lawyers should keep abreast of changes in the ity and advocacy that ROSS is not capable of performing – at law and its practice, including the benefits and risks associatleast, not yet. ed with relevant technology.” Consistent with this guidance, LexisNexis also seeks to be seen as a leader in early AI prudent legal counsel should be aware of AI technologies as technologies directed at legal professionals, and boasts one of they develop, learn their benefits and their risks, and considthe most sophisticated legal predictive analytics tools ever cre- er whether such technologies might enhance their practices ated in Lex Machina. Lex Machina employs natural language today. AI has arrived in the legal field, and just as electronic processing and machine learning, automatically mines a vast discovery has revolutionized the practice of law over the past database of litigation data, provides relevant data-driven re- two decades, we can expect AI’s expansion and development ports, and predicts outcomes based on demonstrated tenden- to fundamentally change the way we provide services to our cies of particular judges and lawyers. Through its legal analyt- clients over the next 20 years. www.sacbar.org | July/August 2017 | SACRAMENTO LAWYER
17