Harry Surden

Bio: Harry Surden is an academic researcher from University of Colorado Boulder. The author has contributed to research in topics: Technological change & Search cost. The author has an hindex of 8, co-authored 22 publications receiving 387 citations. Previous affiliations of Harry Surden include Washington and Lee University & Stanford University.

Machine Learning and Law

Abstract: This Article explores the application of machine learning techniques within the practice of law. Broadly speaking “machine learning” refers to computer algorithms that have the ability to “learn” or improve in performance over time on some task. In general, machine learning algorithms are designed to detect patterns in data and then apply these patterns going forward to new data in order to automate particular tasks. Outside of law, machine learning techniques have been successfully applied to automate tasks that were once thought to necessitate human intelligence — for example language translation, fraud-detection, driving automobiles, facial recognition, and data-mining. If performing well, machine learning algorithms can produce automated results that approximate those that would have been made by a similarly situated person.This Article begins by explaining some basic principles underlying machine learning methods, in a manner accessible to non-technical audiences. The second part explores a broader puzzle: legal practice is thought to require advanced cognitive abilities, but such higher-order cognition remains outside the capability of current machine-learning technology. This part identifies a core principle: how certain tasks that are normally thought to require human intelligence can sometimes be automated through the use of non-intelligent computational techniques that employ heuristics or proxies (e.g., statistical correlations) capable of producing useful, “intelligent” results. The third part applies this principle to the practice of law, discussing machine-learning automation in the context of certain legal tasks currently performed by attorneys: including predicting the outcomes of legal cases, finding hidden relationships in legal documents and data, electronic discovery, and the automated organization of documents.

Machine Learning and Law

Abstract: INTRODUCTIONWhat impact might artificial intelligence (AI) have upon the practice of law? According to one view, AI should have little bearing upon legal practice barring significant technical advances.1 The reason is that legal practice is thought to require advanced cognitive abilities, but such higher-order cognition remains outside the capability of current AI technology.2 Attorneys, for example, routinely combine abstract reasoning and problem solving skills in environments of legal and factual uncertainty.3 Modern AI algorithms, by contrast, have been unable to replicate most human intellectual abilities, falling far short in advanced cognitive processes-such as analogical reasoning-that are basic to legal practice.4 Given these and other limitations in current AI technology, one might conclude that until computers can replicate the higher-order cognition routinely displayed by trained attorneys, AI would have little impact in a domain as full of abstraction and uncertainty as law.5Although there is some truth to that view, its conclusion is overly broad. It misses a class of legal tasks for which current AI technology can still have an impact even given the technological inability to match human-level reasoning. Consider that outside of law, non-cognitive AI techniques have been successfully applied to tasks that were once thought to necessitate human intelligence-for example language translation.6 While the results of these automated efforts are sometimes imperfect, the interesting point is that such computer generated results have often proven useful for particular tasks where strong approximations are acceptable.7 In a similar vein, this Article will suggest that there may be a limited, but not insignificant, subset of legal tasks that are capable of being partially automated using current AI techniques despite their limitations relative to human cognition.In particular, this Article focuses upon a class of AI methods known as "machine learning" techniques and their potential impact upon legal practice. Broadly speaking, machine learning involves computer algorithms that have the ability to "learn" or improve in performance over time on some task.8 Given that there are multiple AI approaches, why highlight machine learning in particular? In the last few decades, researchers have successfully used machine learning to automate a variety of sophisticated tasks that were previously presumed to require human cognition. These applications range from autonomous (i.e., self- driving) cars, to automated language translation, prediction, speech recognition, and computer vision.9 Researchers have also begun to apply these techniques in the context of law.10To be clear, I am not suggesting that all, or even most, of the tasks routinely performed by attorneys are automatable given the current state of AI technology. To the contrary, many of the tasks performed by attorneys do appear to require the type of higher order intellectual skills that are beyond the capability of current techniques. Rather, I am suggesting that there are subsets of legal tasks that are likely automatable under the current state of the art, provided that the technologies are appropriately matched to relevant tasks, and that accuracy limitations are understood and accounted for. In other words, even given current limitations in AI technology as compared to human cognition, such computational approaches to automation may produce results that are "good enough" in certain legal contexts.Part I of this Article explains the basic concepts underlying machine learning. Part II will convey a more general principle: non-intelligent computer algorithms can sometimes produce intelligent results in complex tasks through the use of suitable proxies detected in data. Part III will explore how certain legal tasks might be amenable to partial automation under this principle by employing machine learning techniques. This Part will also emphasize the significant limitations of these automated methods as compared to the capabilities of similarly situated attorneys. …

Artificial Intelligence and Law: An Overview

Abstract: Much has been written recently about artificial intelligence (AI) and law. But what is AI, and what is its relation to the practice and administration of law? This article addresses those questions by providing a high-level overview of AI and its use within law. The discussion aims to be nuanced but also understandable to those without a technical background. To that end, I first discuss AI generally. I then turn to AI and how it is being used by lawyers in the practice of law, people and companies who are governed by the law, and government officials who administer the law. A key motivation in writing this article is to provide a realistic, demystified view of AI that is rooted in the actual capabilities of the technology. This is meant to contrast with discussions about AI and law that are decidedly futurist in nature.

Technological Opacity, Predictability, and Self-Driving Cars

Abstract: Autonomous or “self-driving” cars are vehicles that drive themselves without human supervision or input. Because of safety benefits that they are expected to bring, autonomous vehicles are likely to become more common. Notably, for the first time, people will share a physical environment with computer-controlled machines that can both direct their own activities and that have considerable range of movement. This represents a distinct change from our current context. Today people share physical spaces either with machines that have free range of movement but are controlled by people (e.g. automobiles), or with machines that are controlled by computers but highly constrained in their range of movement (e.g. elevators). The movements of today’s machines are thus broadly predictable. The unrestricted, computer-directed movement of autonomous vehicles is an entirely novel phenomenon that may challenge certain unarticulated assumptions in our existing legal structure.Problematically, the movements of autonomous vehicles may be less predictable to the ordinary people who will share their physical environment — such as pedestrians — than the comparable movements of human-driven vehicles. Today, a great deal of physical harm that might otherwise occur is likely avoided through humanity’s collective ability to predict the movements of other people. In anticipating the behavior of others, we employ what psychologists call a “theory of mind.” Theory of mind cognitive mechanisms that allow us to extrapolate from our own internal mental states in order to estimate what others are thinking or likely to do. These cognitive systems allow us to make instantaneous, unconscious judgments about the likely actions of people around us, and therefore, to keep ourselves safe in the driving context. However, the theory-of-mind mechanisms that allow us to accurately model the minds of other people and interpret their communicative signals of attention and intention will be challenged in the context of non-human, autonomous moving entities such as self-driving cars.This article explains in detail how self-driving vehicles work and how their movements may be hard to predict. It then explores the role that law might play in fostering more predictable autonomous moving systems such as self-driving cars, robots, and drones.

Technological Opacity, Predictability, and Self-Driving Cars

Abstract: Autonomous or “self-driving” cars are vehicles that drive themselves without human supervision or input. Because of safety benefits that they are expected to bring, autonomous vehicles are likely to become more common. Notably, for the first time, people will share a physical environment with computer-controlled machines that can both direct their own activities and that have considerable range of movement. This represents a distinct change from our current context. Today people share physical spaces either with machines that have free range of movement but are controlled by people (e.g. automobiles), or with machines that are controlled by computers but highly constrained in their range of movement (e.g. elevators). The movements of today’s machines are thus broadly predictable. The unrestricted, computer-directed movement of autonomous vehicles is an entirely novel phenomenon that may challenge certain unarticulated assumptions in our existing legal structure.Problematically, the movements of autonomous vehicles may be less predictable to the ordinary people who will share their physical environment — such as pedestrians — than the comparable movements of human-driven vehicles. Today, a great deal of physical harm that might otherwise occur is likely avoided through humanity’s collective ability to predict the movements of other people. In anticipating the behavior of others, we employ what psychologists call a “theory of mind.” Theory of mind cognitive mechanisms that allow us to extrapolate from our own internal mental states in order to estimate what others are thinking or likely to do. These cognitive systems allow us to make instantaneous, unconscious judgments about the likely actions of people around us, and therefore, to keep ourselves safe in the driving context. However, the theory-of-mind mechanisms that allow us to accurately model the minds of other people and interpret their communicative signals of attention and intention will be challenged in the context of non-human, autonomous moving entities such as self-driving cars.This article explains in detail how self-driving vehicles work and how their movements may be hard to predict. It then explores the role that law might play in fostering more predictable autonomous moving systems such as self-driving cars, robots, and drones.

Current index to legal periodicals

Abstract: American Bankruptcy Institute Law Review 17 Am. Bankr. Inst. L. Rev., No. 1, Spring, 2009. Boston College Law Review 50 B.C. L. Rev., No. 3, May, 2009. Boston University Public Interest Law Journal 18 B.U. Pub. Int. L.J., No. 2, Spring, 2009. Cardozo Journal of Conflict Resolution 10 Cardozo J. Conflict Resol., No. 2, Spring, 2009. Cardozo Public Law, Policy, & Ethics Journal 7 Cardozo Pub. L. Pol’y & Ethics J., No. 3, Summer, 2009. Chicago Journal of International Law 10 Chi. J. Int’l L., No. 1, Summer, 2009. Colorado Journal of International Environmental Law and Policy 20 Colo. J. Int’l Envtl. L. & Pol’y, No. 2, Winter, 2009. Columbia Journal of Law & the Arts 32 Colum. J.L. & Arts, No. 3, Spring, 2009. Connecticut Public Interest Law Journal 8 Conn. Pub. Int. L.J., No. 2, Spring-Summer, 2009. Cornell Journal of Law and Public Policy 18 Cornell J.L. & Pub. Pol’y, No. 1, Fall, 2008. Cornell Law Review 94 Cornell L. Rev., No. 5, July, 2009. Creighton Law Review 42 Creighton L. Rev., No. 3, April, 2009. Criminal Law Forum 20 Crim. L. Forum, Nos. 2-3, Pp. 173-394, 2009. Delaware Journal of Corporate Law 34 Del. J. Corp. L., No. 2, Pp. 433-754, 2009. Environmental Law Reporter News & Analysis 39 Envtl. L. Rep. News & Analysis, No. 7, July, 2009. European Journal of International Law 20 Eur. J. Int’l L., No. 2, April, 2009. Family Law Quarterly 43 Fam. L.Q., No. 1, Spring, 2009. Georgetown Journal of International Law 40 Geo. J. Int’l L., No. 3, Spring, 2009. Georgetown Journal of Legal Ethics 22 Geo. J. Legal Ethics, No. 2, Spring, 2009. Golden Gate University Law Review 39 Golden Gate U. L. Rev., No. 2, Winter, 2009. Harvard Environmental Law Review 33 Harv. Envtl. L. Rev., No. 2, Pp. 297-608, 2009. International Review of Law and Economics 29 Int’l Rev. L. & Econ., No. 1, March, 2009. Journal of Environmental Law and Litigation 24 J. Envtl. L. & Litig., No. 1, Pp. 1-201, 2009. Journal of Legislation 34 J. Legis., No. 1, Pp. 1-98, 2008. Journal of Technology Law & Policy 14 J. Tech. L. & Pol’y, No. 1, June, 2009. Labor Lawyer 24 Lab. Law., No. 3, Winter/Spring, 2009. Michigan Journal of International Law 30 Mich. J. Int’l L., No. 3, Spring, 2009. New Criminal Law Review 12 New Crim. L. Rev., No. 2, Spring, 2009. Northern Kentucky Law Review 36 N. Ky. L. Rev., No. 4, Pp. 445-654, 2009. Ohio Northern University Law Review 35 Ohio N.U. L. Rev., No. 2, Pp. 445-886, 2009. Pace Law Review 29 Pace L. Rev., No. 3, Spring, 2009. Quinnipiac Health Law Journal 12 Quinnipiac Health L.J., No. 2, Pp. 209-332, 2008-2009. Real Property, Trust and Estate Law Journal 44 Real Prop. Tr. & Est. L.J., No. 1, Spring, 2009. Rutgers Race and the Law Review 10 Rutgers Race & L. Rev., No. 2, Pp. 441-629, 2009. San Diego Law Review 46 San Diego L. Rev., No. 2, Spring, 2009. Seton Hall Law Review 39 Seton Hall L. Rev., No. 3, Pp. 725-1102, 2009. Southern California Interdisciplinary Law Journal 18 S. Cal. Interdisc. L.J., No. 3, Spring, 2009. Stanford Environmental Law Journal 28 Stan. Envtl. L.J., No. 3, July, 2009. Tulsa Law Review 44 Tulsa L. Rev., No. 2, Winter, 2008. UMKC Law Review 77 UMKC L. Rev., No. 4, Summer, 2009. Washburn Law Journal 48 Washburn L.J., No. 3, Spring, 2009. Washington University Global Studies Law Review 8 Wash. U. Global Stud. L. Rev., No. 3, Pp.451-617, 2009. Washington University Journal of Law & Policy 29 Wash. U. J.L. & Pol’y, Pp. 1-401, 2009. Washington University Law Review 86 Wash. U. L. Rev., No. 6, Pp. 1273-1521, 2009. William Mitchell Law Review 35 Wm. Mitchell L. Rev., No. 4, Pp. 1235-1609, 2009. Yale Journal of International Law 34 Yale J. Int’l L., No. 2, Summer, 2009. Yale Journal on Regulation 26 Yale J. on Reg., No. 2, Summer, 2009.

Inferring Personal Information from Demand-Response Systems

Abstract: Current and upcoming demand-response systems provide increasingly detailed power-consumption data to utilities and a growing array of players angling to assist consumers in understanding and managing their energy use. The granularity of this data, as well as new players' entry into the energy market, creates new privacy concerns. The detailed per-household consumption data that advanced metering systems generate reveals information about in-home activities that such players can mine and combine with other readily available information to discover more about occupants' activities. The authors explore the technological aspects of this claim, focusing on the ways in which personally identifying information can be collected and repurposed. Their results show that, even with relatively unsophisticated hardware and data-extraction algorithms, some information about occupant behavior can be estimated with a high degree of accuracy. The authors propose a disclosure metric to aid in quantifying the impact of data collection on in-home privacy and construct an example metric for their experiment.

Machine learning, social learning and the governance of self-driving cars:

Abstract: Self-driving cars, a quintessentially ‘smart’ technology, are not born smart. The algorithms that control their movements are learning as the technology emerges. Self-driving cars represent a high-stakes test of the powers of machine learning, as well as a test case for social learning in technology governance. Society is learning about the technology while the technology learns about society. Understanding and governing the politics of this technology means asking ‘Who is learning, what are they learning and how are they learning?’ Focusing on the successes and failures of social learning around the much-publicized crash of a Tesla Model S in 2016, I argue that trajectories and rhetorics of machine learning in transport pose a substantial governance challenge. ‘Self-driving’ or ‘autonomous’ cars are misnamed. As with other technologies, they are shaped by assumptions about social needs, solvable problems, and economic opportunities. Governing these technologies in the public interest means improving soci...

Predicting the Behavior of the Supreme Court of the United States: A General Approach.

Abstract: Building upon developments in theoretical and applied machine learning, as well as the efforts of various scholars including Guimera and Sales-Pardo (2011), Ruger et al. (2004), and Martin et al. (2004), we construct a model designed to predict the voting behavior of the Supreme Court of the United States. Using the extremely randomized tree method first proposed in Geurts, et al. (2006), a method similar to the random forest approach developed in Breiman (2001), as well as novel feature engineering, we predict more than sixty years of decisions by the Supreme Court of the United States (1953-2013). Using only data available prior to the date of decision, our model correctly identifies 69.7% of the Court’s overall affirm/reverse decisions and correctly forecasts 70.9% of the votes of individual justices across 7,700 cases and more than 68,000 justice votes. Our performance is consistent with the general level of prediction offered by prior scholars. However, our model is distinctive as it is the first robust, generalized, and fully predictive model of Supreme Court voting behavior offered to date. Our model predicts six decades of behavior of thirty Justices appointed by thirteen Presidents. With a more sound methodological foundation, our results represent a major advance for the science of quantitative legal prediction and portend a range of other potential applications, such as those described in Katz (2013).

A General Approach for Predicting the Behavior of the Supreme Court of the United States

Abstract: Building on developments in machine learning and prior work in the science of judicial prediction, we construct a model designed to predict the behavior of the Supreme Court of the United States in a generalized, out-of-sample context. To do so, we develop a time-evolving random forest classifier that leverages unique feature engineering to predict more than 240,000 justice votes and 28,000 cases outcomes over nearly two centuries (1816-2015). Using only data available prior to decision, our model outperforms null (baseline) models at both the justice and case level under both parametric and non-parametric tests. Over nearly two centuries, we achieve 70.2% accuracy at the case outcome level and 71.9% at the justice vote level. More recently, over the past century, we outperform an in-sample optimized null model by nearly 5%. Our performance is consistent with, and improves on the general level of prediction demonstrated by prior work; however, our model is distinctive because it can be applied out-of-sample to the entire past and future of the Court, not a single term. Our results represent an important advance for the science of quantitative legal prediction and portend a range of other potential applications.