SCOTUS Marriage Equality 2015 58021
Photo by Ted Eytan (Flickr/Creative Commons)

Will Artificial Intelligence Favor Conservative Legal Theorists?

  • Future of Legal Automation Holds Multiple Scenarios, Paper Argues
  • If computers replace human lawyers anytime soon, it is more likely to benefit politically conservative legal philosophies, such as the one embraced by Supreme Court Justice Antonin Scalia, than liberal philosophies – purely due to idiosyncrasies in the way the dominant technology works, a new paper argues.

    That’s one of the ideas put forward in a draft version of “Incomplete Innovation and the Premature Disruption of Legal Services,” written by Brian Sheppard, an associate professor at Seton Hall University School of Law.

    “My argument is that if lawyers are going to be replaced by computers … some of the more conservative [legal] theories may become more popular again because computers can do them quite well,” he said in an interview with Big Law Business.

    By legal philosophies or theories, Sheppard said he meant the different principles of adjudication that lawyers use when interpreting the law and the Constitution. His paper doesn’t identify any one philosophy with a political leaning, but Sheppard told Big Law Business that generally speaking, some conservative theories will likely be easier for computers to emulate because they are more black-and-white, with little subjectivity.

    One of the best-known philosophies, Originalism, holds that the meaning of the U.S. Constitution was fixed at the time of its enactment, and is now closely identified with Scalia, Sheppard said. While it’s often viewed as conservative, it could also be viewed as a liberal bent under certain circumstances, he added.

    “There’s no philosophical reason why these theories are conservative,” Sheppard said, about Originalism, adding, “Liberals would love Originalism if the country had been on a steady conservative slide since the day it was founded.”

    The bulk of the paper is devoted to Sheppard’s theory that bespoke legal services — such as negotiation, the creation of high-stakes legal documents and advocacy — will be more scarce in the future as a result of what he calls “premature disruption,” caused by technological innovation, business model innovation or both.

    In explaining that theory, Sheppard devotes a chunk of his paper to “natural language processing,” the technology used in various artificial intelligence tools including Google, Apple’s Siri, and IBM’s Watson, and which he believes will play a critical role in any disruption that occurs.

    Loosely defined, natural language processing is the interaction between computers and human language, such as a computer’s ability to conduct textual analysis or understand speech. A particular branch of this technology, called the syntactical approach, has progressed the furthest, according to Sheppard.

    Used in Siri and other products, the syntactical technique “is agnostic as to meaning; it essentially searches for matching sequences of characters and sorts based on distance between matching sequences within documents or based on match frequency,” the paper explains. Thus, it struggles with “elementary interpretative tasks,” but it can parse texts and extract meaning.

    He situates various legal philosophies along a curve based on how easily they can be emulated using syntactical natural language processing.

    At the bottom of the curve, the legal philosophy that could be most easily emulated with existing technology is Naïve Textualism, or the idea that statutes can be understood by looking up their words in a dictionary.

    Further along the curve, already partially available for automation, Sheppard placed the legal philosophy known as Originalism, most famously embraced by Scalia, which holds that the U.S. Constitution’s meaning was fixed at the time it was enacted.

    To the extent that Originalism envisions the law as having a fixed meaning in the past, Originalism can be more easily automated, the paper states. But Originalism also calls for an understanding of “the intent” of the people who wrote the Constitution, and thus requires a computer to understand complex contextual dimensions, such as the “perceived problems, goals, hopes or ideals” of original legislators, which makes it difficult to fully automate with the current technology.

    Living Constitutionalism, which involves evaluating principles according to their practicality, how well they promote democratic legitimacy and the evolving values of the present time, sits at the highest spot on the curve and is the least computer-friendly, the paper states.

    Eventually, natural language processing could emulate such philosophies, with more gray areas, but that technology is still years, if not decades away from being deployable, according to Sheppard.