Proctor : December 2017
33 PROCTOR | December 2017 by Milan Gandhi, the Legal Forecast There is a difference between retrieving information in response to a legal question – which, for example, ROSS Intelligence can do – and explaining the answer by “[reasoning] with the rules and concepts relevant to choice of law and legal subject matter.” 16 Even if a tool like ROSS could recognise and extract legal arguments, “[it] could not itself construct the explanation from first principles.” 17 According to Kevin Ashley, the next breakthrough in AI for the legal world will come with AI-generated “explanations and arguments in law”. 18 Choosing a humane future In 2016, Eric Loomis was convicted of a drive-by shooting and sentenced to six years in prison. His conviction was based, in part, on a trial court’s consideration of a software-generated report. 19 The software’s underlying algorithm, which remains a trade secret, predicted “a high risk of violence, high risk of recidivism, [and] high pre–trial risk.” 20 On appeal, the Wisconsin Supreme Court rejected Loomis’s arguments that judicial reliance on the software tool violated his due process rights. Loomis had argued that it did so by incorporating data about general groups into considerations of an individual’s sentence, and because the underlying methodology was kept hidden. 21 Loomis’s arguments – albeit unsuccessful – are a reminder to look beneath the surface of technology. It has been argued before lawmakers of the European Union that “being able to interrogate an AI system about how it reached its conclusions [should be] a fundamental legal right.” 22 Indeed, it must be borne in mind that the makers and beneficiaries of AI are human, and some of its worst failures may mirror our own. “The Guardian” newspaper reported, for example, that “[m]achine learning algorithms are picking up deeply ingrained race and gender prejudices concealed within the patterns of language use.” 23 The marriage of law and new technologies is a rightfully tantalising prospect. However, as we rely on machines to solve many of our human problems, including in the law, we have a duty to ensure that we do not, by failing to scrutinise it, reinforce our old problems, or create entirely new ones. Legal Technology Milan Gandhi is the national director of The Legal Forecast (TLF), a Centre for Legal Innovation advisory board member, and a research clerk at McCullough Robertson. Special thanks to Michael Bidwell and Benjamin Teng of The Legal Forecast for editorial advice and input. The Legal Forecast (thelegalforecast.com) aims to advance legal practice through technology and innovation. TLF is a not-for-profit run by early career professionals passionate about disruptive thinking and access to justice.