Is the Dawn of the Robot Lawyer upon us? The Fourth Industrial Revolution and the Future of Lawyers

The practice of law has been largely shielded from technological developments in the course of the past 50 years. While the ways in which legal professionals process and share information have evolved with new technologies — primarily with the emergence of personal computers, email and the internet — these technologies have not fundamentally transformed it. However, if media reports are to be believed, advances in technology in general — and the field known as "Artificial Intelligence" (AI) in particular — are on lawyers' doorsteps, and the legal industry is on the cusp of radical change. Fuelled by big data, increased computing power and more effective algorithms, AI has the potential to fundamentally transform the way in which legal work is done, the way in which law firms conduct business, and the way in which lawyers deal with clients. A number of technologies that fall under the AI umbrella, such as machine learning, natural language processing, deep learning and others, have already brought about the automation of many tasks that were, until recently, performed exclusively by humans because they required human intelligence. AI systems can also be used to perform many tasks that lawyers routinely perform, such as contract analysis, case prediction and e-discovery. And, according to proponents, these emerging technologies can do it cheaper, faster and more efficiently. This contribution examines the notion that recent advances in technology will "disrupt" the legal profession. It first describes the astonishing advances in technological progress, especially the recent rise of AI. It then considers the technologies and areas of legal practice most susceptible to this disruption. It concludes by envisaging what AI might mean for the legal profession, and how current technological trends might, in a relatively short period of time, transform the way in which legal services are delivered.


Introduction
In the Harvard Law Review in 1897, Oliver Wendell Holmes, Jr 2 wrote that: For the rational study of the law the black-letter man may be the man of the present, but the man of the future is the man of statistics and the master of economics.
The future that Holmes foresaw some 120 years ago has arrived. Humanity is fast traversing the chasm between a print-based analogue industrial society, with unreliable, slow and costly communication, and a technologybased internet society, with near universal, real-time and virtually costless communication. 3 The set of tasks and activities in which humans are strictly superior to computers is becoming "vanishingly small". 4 Machines today not only perform mechanical or manual tasks once performed by humans, but they are also performing thinking taskstasks long believed to be the exclusive purview of humans, in which human judgment was indispensable. From self-driving cars to self-flying aeroplanes, and from robots performing surgery on a pig, to artificially intelligent personal assistants, so much of what was once unimaginable is now reality. 5 And, unlike humans, artificially intelligent machines can work ceaselessly around the clock, without sleep or caffeine. 6 In considering the long-term future of the law, it is difficult to ignore the recent avalanche of interest in AI for lawyers. AI and its applications to law practice are grabbing headlines in the legal industry. Ever since the early success stories of IBM's Watson, the legal press has been abuzz with articles that debate whether AI is a threat or a hope, and whether AI will The introduction of AI and its impact in industry, education and society distinguishes the fourth industrial revolution from the third. 17 AI is the key transformative technology emerging in the early twenty-first centurythe wave of the future that will change forever the world of work, play, living and business as we know it today. 18 It is expected that within the next five years over 50 billion connected machines (the so-called "internet of things") will exist throughout the world. 19 According to the McKinsey Global Institute, a technology think tank, AI is not only contributing to the transformation of society, but compared to the first industrial revolution, the AI revolution is "happening ten times faster and at 300 times the scale, or roughly 3 000 times the impact." 20 In their book, The Second Machine Age, Eric Brynjolfsson and Andrew McAfee 21 contend that: Computers and digital advances are doing for mental powerthe ability to use our brains to understand and shape our environmentswhat the steam engine and its descendants did for muscle power.
Pillay 2018 https://www.up.ac.za/news/post_2688680--artificial-intelligence-mythor-reality-for-the-fourth-industrial-revolution. The 1970s saw the advent of the third industrial revolution, which was characterised by electronics, the computer and nuclear energy. Personal computers and information technology in the form of the internet taking hold in working life ushered in an era of global access to information and the automation of work. Pillay 2018 https://www.up.ac.za/news/post_2688680-artificial-intelligence-myth-or-reality-for-the-fourth-industrial-revolution; Wisskirchen et al Artificial Intelligence and Robotics 12. 18 Botha date unknown http://www.foresightfordevelopment.org/featured/artificialintelligence-ii. 19 Wisskirchen et al Artificial Intelligence and Robotics 12. 20 Barfield "Towards a Law of Artificial Intelligence" 3. 21 Brynjolffson and McAfee Second Machine Age 7-8. 22 McKamey 2017 Appeal 47. WH GARVETT PER / PELJ 2020 (23) 5 The legal profession is at a crossroads. Just as the other professions are undergoing tremendous upheaval, so it must be with the law. The legal world of tomorrow will bear little resemblance to that of the present. It will change more in the next twenty years than it has during the past two centuries. 23 In his book, Tomorrow's Lawyers, legal technologist Richard Susskind identifies three drivers of change in the legal profession. 24 While the 2008 financial crisis can be seen as the precipitating event, developments in legal information technology are actually a root cause of many of the long-term changes in the legal services market. 25 This contribution focusses on the next frontier along the path of blistering technological advance, specifically AI. 26 At its core, law is information technology -"a code that regulates social life", 27 and we are in the middle of an information revolution. In our age, the ability to collect, store, process and analyse data continues to improve at an exponential rate. 28 This is the result of "Moore's Law", which is named after Gordon Moore, one of the co-founders of Intel Corporation. 29 In 1965 Moore predicted that the number of transistors that could be fitted onto a silicon chip would double every two years. His prediction has proved to be remarkably accurate. Over the past fifty years, computing speed, power and capacity have doubled approximately every two years, while the attendant costs have halved every two years. 30 If one compares the first microprocessor with today's models, performance has increased more than 3 500 times, energy efficiency has increased 90 000 times, and price has decreased by a factor of 60 000. 31 23 Susskind Tomorrow's Lawyers xvii. 24 According to Susskind, the three drivers fuelling change in the legal profession are the "more-for-less" challenge; liberalisation; and technology. Very briefly, the morefor-less challenge describes the increasing pressure on law firms to deliver more legal services for less money. It is intuitively difficult for humans to appreciate the immense power of such exponential growth. At first glance, doubling computing speed every two years may not seem that impressive, but these technological improvements have led to gargantuan changes over time. Imagine, for example, that motor vehicles had developed according to Moore's Law. That would mean that they would currently have a top speed of more than 482 000 km/h, fuel efficiency of more than 8 500 000 km/l, and a cost of approximately 4 cents. 32 Recent studies have confirmed that not only has Moore's Law remained remarkably accurate, but telecommunication and the storage of information have also experienced exponential growth. In 2008 a cell phone's computational power was one thousand times greater, and it was one million times less expensive, than the combined computing power at the Massachusetts Institute of Technology in 1965. 33 With dramatic improvements in computer software and hardware, and increasing access to large datasets, it is no wonder that Ray Kurzweil, 34 a Google AI expert and well-known futurist, believes that the average desktop computer will surpass human brain power by 2023, and reach "singularity" in 2045when the AI in a single computer will surpass the brainpower equivalent to that of all people combined.
The time has indubitably come for lawyers to reconsider some of their working practices. 35 Because increases in the power of computing are exponential and not linear, computers may be able to undertake complicated legal tasks relatively sooner than it initially took computers to perform simpler legal tasks.

What does it all mean?
Although AI is talked about in the media almost every day, there is still no generally accepted definition of the term. Individual definitions run the gamut from a super-intelligent, sentient, humanoid, world-conquering robot to an 32 Sprangers 2018 https://www.johnsonflora.com/recent-articles-media/artificialintelligence-and-the-law-revolution-or-evolution/. An attempt will be made to find how to make machines use language, from abstractions and concept, solve kinds of problems now reserved for humans, and improve themselves ... For the present purpose the artificial intelligence problem is taken to be that of making a machine behave in ways that would be called intelligent if a human were so behaving.
Thus, in the sense intended in this article, AI refers to a computer's ability to imitate human intelligent behaviour, especially human cognitive functions, such as the ability to reason, discover meaning, generalise and learn from past experience. 38 While AI has many attributes useful for its varied applications, at present two are most important in the legal domain. Firstly, "machine learning", the leading innovative force in AI, has proven incredibly efficient, as machines can perform in mere minutes tasks that would otherwise take a team of lawyers tens of hours. 39 Machine learning refers to the capability of AI systems to teach themselves and learn from experience. This means, in essence, that AI can do much more than blindly adhere to what it has initially been programmed to do; it can learn from experience and dataso-called "big data" that collectively is almost unimaginably vast in the human context, and far beyond what highly-skilled, experienced workers could reasonably construct on their ownto constantly improve its capabilities. 40 Machine learning should be envisioned as a spectrum that ranges from relatively simple algorithms to complex self-teaching systems that could eventually mirror the human brain in their complexity, if not their structure. The latter is termed "deep learning". 41 Deep learning relies on what is referred to as "neural networks", an interconnected group of nodes designed to mimic the activity of neurons in the human brain in order to recognise complex patterns in data sets. 42 36 Giuffrida Secondly, "natural language processing" is the capability of algorithms and software to interpret, understand and generate spoken and written human language, 43 and then to apply and integrate that understanding to perform human-like analysis. 44 Search engines, speech-to-speech translation, and artificially intelligent assistants, such as iPhone's Siri, are built with natural language processing technology for the user's benefit. 45

The story of AI through human games
A long-standing goal of AI was to develop an algorithm that learns, tabula rasa, superhuman proficiency in challenging domains. 46 As a powerful illustration of how this goal has been met, it is worth considering the remarkable success of AI in the arena of human games.
In 1997  For I.B.M., the showdown was not merely a well-publicized stunt and a $1 million prize, but proof that the company had taken a big step toward a world in which intelligent machines will understand and respond to humans, and perhaps inevitably, replace some of them.
Even more remarkable than Watson's victory in Jeopardy! in early 2016 an artificially intelligent machine, Google's AlphaGo, ruthlessly defeated the human grandmaster, Lee Sedol, in four matches against one in the ancient Chinese board game of Go. The game of Go had remained an inviolably human pursuit for centuries, and one of the most difficult challenges for AI. 52 This is a game with simple rules 53 but profound complexity 54there are more potential moves in Go than there are atoms in the entire observable universe of galaxies, stars, planets, animals, plants and people. 55 Given this virtually illimitable complexity, a computer was unable to gain a competitive advantage by employing brute force computations. 56 To play a decent game of Go, a computer had to be endowed with the ability to recognise subtle, 50 Katz 2013  On its surface, Go seems simple compared to chess. A chess match begins with two facing armies of sixteen pieces each, ranking from pawn to king, on a 64-square board. Each of the six kinds of pieces is allowed to move only in certain ways. In Go, on the other hand, there are few such complications. A game begins with a blank 19-by-19 grid. Each player has access to a bowl of either white or black stones which she then places on the board in turn. The goal of the game, which originated in China more than 2 500 years ago, is to capture a group of the opponent's stones by surrounding it and then removing the cluster from the board. The object is to build complex fence-like structures enclosing as much territory as possible. Out of this sheer simplicity, great beauty arisescomplex battles between black and white stone armies that span the entire board. In essence, with chess one starts with all the pieces on the board, while in go one starts with an empty board and then builds. Koch 2016 https://www.scientificamerican.com/article/how-the-computer-beat-thego-master/; Johnson 1997 https://www.nytimes.com/1997/07/29/science/to-test-apowerful-computer-play-an-ancient-game.html. 54 Lew 2018  complex patterns and to draw on the kind of intuitive knowledge that is the hallmark of human intelligence. 57 It is for this reason that Dr Piet Hut 58 of the Institute for Advanced Study at Princeton predicted in 1997: "It may be a hundred years before a computer beats humans at Gomaybe even longer." At the same time, an article in The New York Times declared: "When or if a computer defeats a human Go champion, it will be a sign that artificial intelligence is truly beginning to become as good as the real thing." 59 Then, in less than 20 years the unthinkable happened. At a press conference in the wake of his defeat to AlphaGo, Lee Sedol 60 said: "Today, I am speechless". In the popular press, AlphaGo's victory over the human grandmaster cemented its position as a remarkable scientific feat that came years ahead of scientists' predictions. 61 But its mastery is even more exceptional if one considers the deeper implications of the AI technology that made the computer's victory possible.
AlphaGo sometimes made unexpected moves that the human experts simply could not explain. At first, these moves appeared to be errors. But the AI system then used the surprising positions it generated to dominate the rest of the game. 62 Such intriguing moves are also sometimes made by human Go masters. They are known in Japanese as kami no itte ("the hand of God" or "divine moves"). Thus, here is a system that displays characteristics that a human would describe as "creative" or "innovative". 63 Similarly, decisions made by AI-augmented machines to balance an electrical grid or to manage a city's traffic flow are made beyond human processing capabilities. Yoshua Bengio, 64 a computer scientist at the University of Montreal and often called one of the "fathers of AI", refers to this kind of processing as "artificial intuition". 57 Johnson 1997 https://www.nytimes.com/1997/07/29/science/to-test-a-powerfulcomputer-play-an-ancient-game.html. 58 Alarie, Niblett and Yoon 2016 UTLJ 423. 59 Johnson 1997 https://www.nytimes.com/1997/07/29/science/to-test-a-powerfulcomputer-play-an-ancient-game.html. 60 As quoted in Murgia 2017 https://www.ft.com/content/cada14c4-d366-11e6-b06b-680c49b4b4c0. 61 Murgia 2017  This discussion about the advances of AI in human games serves to illustrate that many emerging technologies do not simply computerise and streamline pre-existing and inefficient manual processes. Rather than merely automate, these systems innovate. This means that they allow us to perform tasks that were previously not possible, or even imaginable. 67 According to Susskind, the message for lawyers is profoundwhen thinking about technology, the challenge is not simply to automate current working practices that are inefficient. The challenge is to innovate the practice of law in ways that we could not do or even foresee in the past. 68 These innovative technologies do not simply support traditional ways of working. As discussed below, they are disruptivethese "pervasive, exponentially growing, innovative technologies" will come to fundamentally and radically transform the way in which lawyers work. 69 Rather than a client paying a fixed amount to the lawyer to complete an assignment, irrespective of the lawyer's time expended, hourly billing has the intuitive appeal of paying only for the time the lawyer actually worked. 72 However, this approach, as has been well-chronicled, creates the potential moral hazard of lawyers billing as many hours as possible to maximise profits. 73 Beyond these concerns, the hourly billing model challenges clients' ability to retain legal representation. 74 Simply put, the cost of lawyering has become too high. Over time, accessibility issues have only worsened, as the cost of legal representation has far outpaced the rate of inflation. 75 Although the law is central to our lives, at present, effectively, only the very rich or the very poor (through legal aid) have the means to afford the services of lawyers.
Significant amounts of legal work that is administrative or process-based, whether at corporate legal departments or in private practice, are being undertaken by junior lawyers. This work is characterised by process rather than judgment, procedure instead of strategy or creativity; for example, document review in litigation, due diligence work in corporate transactions (such as mergers and acquisitions), basic contract drafting, and rudimentary legal research. Susskind 76 is of the view that this presents a great opportunity for change. The challenge is to identify work that can be routinised and undertaken more efficiently through computerisation.
In the aftermath of the 2008 recession, lawyers across diverse practice settingslarge firms, small firms, government, public interest organisations and corporationshave all explored ways to provide more cost-effective legal representation. Increasingly, lawyers in private practice find that clients are less willing to accept the indeterminacy of hourly billing. 77 Government and public-interest organisations, in response to budgetary austerity, have been compelled to explore ways to do more with less. The incentive to save costs and increase precision through technology is and will continue to be great. Imagine a computer application that reduces ten hours' worth of legal work into a mere instant. It is this kind of machine intelligence that is substantially "shaking up" the legal profession. 80 Lawyers who are prepared to change their practices or organisations to take advantage of the lower cost inputs that machines enable will be able to serve an expanding market of legal services for middle-class individuals and small businesses, meeting previously unfulfilled legal needs. 81 Legal services are not "monolithic, indivisible professional engagements" that lawyers must undertake in only one way. Instead, legal services can be "decomposed" (or "disaggregated" or "unbundled") into various tasks, which lawyers should undertake in as efficient a manner as possible. 83

Decomposing legal services
For better or for worse, a nontrivial subset of tasks undertaken by lawyers is subject to automation. The bundle of skills associated with the practice of law falls on a continuum, on which a number of basic tasks have already been displaced by computation, automation, and AI. 84  At this time and for the foreseeable future, current AI capabilities permit machines to approach, achieve or exceed only certain but not all human cognitive functions. At this point we can safely say that AI does not have the capacity and will probably not obtain the capacity in the next two or three decades to advise clients, appear in court and be responsive to a particular client's priorities and subjective needs. 88 Thus, although "knowledge" jobs in the law might fall victim to the advance of AI, "wisdom" jobs will notat least not in the foreseeable future.
That is why legal skills based on human judgment, inference, common sense, interpersonal skills and experience will remain valuable for the lifetime of any lawyer practising today. 89 As Phillip Marshall QC, a commercial barrister in London, stated in an interview in Financial Times: 90 In litigation, I can see that developments in [AI] may lead to improvements in the search process in disclosure and in the preparation of some documents. But where judgment is requiredfor example in written advocacy -I doubt there is much scope for it. The human touch and knowledge of the likely reaction of the tribunal is at a premium.
Matthew Holmes, and Irish barrister, concurs that, in advocacy, the "human touch" will always remain crucial: 91 that litigators engage in are routine, repetitive and largely administrative. 93 It is in these tasks that no human lawyer stands a chance against the formidable processing power of a mainframe when it comes to sifting through voluminous data. 94

Disruptive legal technologies
In management theory, based on Clayton Christensen's influential book, The Innovator's Dilemma, 95 a distinction is commonly drawn between sustaining and disruptive technologies. Generally, sustaining technologies are those that support and enhance the way in which a business or a market currently operates. In contrast, disruptive technologies fundamentally challenge the functioning of a business or a market sector. An example of the former is computerised accounting systems, which sustain and enhance the work of accountants who previously laboured over paper ledgers. An illustration of the latter is digital camera technology, which famously disrupted and contributed to the eventual downfall of market leader Kodak, whose business was based on outdated chemical printing technology. 96 The introduction of information technologyelectronic databases, the internet and e-mailchanged the speed and mode of the delivery of legal communications, and made redundant the need for legal processes and services to take place anywhere in the physical world, instead occurring in cyberspace. However, it did not transform the fundamental nature of legal services or the practice of law, and was thus sustaining in its effect. 97 On the other hand, AIas it has had across near enough all industries and all aspects of societywill have a seismic impact on the practice of law. John McGinnis and Russell Pearce argue that machine intelligence will cause a "great disruption" in the market for legal services. 98 In fact, there already are a number of disruptive technologies in law. Individually, they are challenging the way in which certain legal services are delivered, and collectively, they will transform the practice of law A recent survey of managing partners of United States law firms revealed that more than 36% of law firms with 50 or more lawyers, and 90% of mega firms (with more than 1 000 attorneys), are either currently using, or actively exploring, the use of AI in their legal practices. 100 The majority of the "magic circle" firms in the United Kingdom have recently employed one or another AI product and are expressing great hopes for their investments. 101 Large South African law firms have also recently started to invest in AI solutions to improve efficiencies in certain key legal processes, particularly in mergers and acquisitions, private equity and compliance areas. 102 As a cover story in the American Bar Association Journal explained: 103 [AI] is changing the way lawyers think, the way they do business and the way they interact with clients.
[AI] is more than legal technology. It is the next great hope that will revolutionize the legal profession.
By the same token, however, this disruption has sparked a fearful murmur across the legal industry. According to Deloitte Insight in a 2016 study, about 114 000 legal sector jobs are likely to be automated in the next twenty years. 104 Deloitte claims that 39% of legal jobs are at "high risk" of being automated in the next two decades, while the McKinsey Global Institute puts the number at 23%. 105 In a 2015 survey by law firm consultant Altman Weil of law firm managing partners, 47% stated that, in the next five to ten years, they could envision a law-focussed AI tool replacing paralegals, and 35% could envision the replacement of first-year associates. 106 106 Zhao 2017 http://stlr.org/2017/10/03/artificial-intelligence-and-its-impacts-on-thelegal-profession/?cn-reloaded=1. 107 Remus and Levy 2017 Geo J Legal Ethics 536. WH GARVETT PER / PELJ 2020 (23)  17 recognise patterns. 108 Computers may still be ill-equipped to tackle many basic cognitive tasks, but at the tasks in which they do excelshifting through vast amounts of data, for examplethey dramatically surpass human abilities. Asking whether individual lawyers can be entirely replaced by machines is perhaps not the right question. Rather, the question should be whether one lawyer, augmented by machines, can perform the same work that five human lawyers used to do. 109 The answer is simpleit is already happening.

E-discovery
The most mature incursion of technology into the practice of law is technology-assisted review, i.e. using software with specifically programmed algorithms to organise, analyse and search enormous and diverse data sets for e-discovery or other record-intensive investigations. 110 Perhaps predictably, lawyers were initially reluctant to give a computer control of a task that could have grave consequences if performed poorly, and insisted on having humans do the work of discovery. 111 However, Maura Grossman and Gordon Cormack, in their seminal 2011 article, debunked the myth that manual human review of discovery documents is the most accurate form of review. Instead, these scholars found that "technology-assisted review can (and does) yield more accurate results than exhaustive manual review, with much lower effort". 112  corporations as litigants. The volume of discovery could reach into the millions of pages. 116 In the days of traditional discovery, the process was highly labour-intensive, time-consuming and expensive, requiring a small army of lawyers to manually read through boxes of documents in search of information relevant to the litigation. 117 However, because most legal documents currently originate in electronic format, through e-discovery, 118 the discovery process can now be conducted with a fraction of the time, expense, and need for lawyers, and with even greater accuracy. 119 Benjamin Alarie et al point out that United States lawyers now spend less than 5 per cent of their time on basic document review. 120

Document analysis
Lawyers have traditionally spent much of their time ploughing through documents, not least in preparation for litigation or in due diligence for transactions. However, in the course of the past decade, law firms have turned to computer programmes to analyse lengthy documents, reducing the time and cost of undertaking due diligence and discovery. 121 These emerging systems are proving increasingly impressive, whether in analysing document sets or summarising or extracting key provisions from contracts. 122 For some years now, in terms of precision and recall, properly primed computer systems have been able to outperform junior lawyers in reviewing voluminous documents and isolating those of relevance, 123 and to decrease the amount of time spent in discovery by as much as 75%. 124 The machine learning capabilities of these AI-augmented systems are disruptive for law firms that have profited from employing human beings to wade through roomfuls of documents (whether on transactions, litigation projects, or contract reviews). As Susskind rightly points out, regardless of how low human labour costs might be, a system of this kind, once implemented, will always be more cost-effective. and RAVN, are generating great interest across the spectrum in the legal profession. 125 The gains in efficiency can be striking. 126 The international investment bank JPMorganChase recently announced that it had developed and implemented learning software COIN (COntract INtelligence) to review in mere seconds what had previously taken a team of lawyers and loan officers some 360 000 hours to review, resulting in considerable cost-savings. 127 As if the magnitude of this computing power is not impressive enough, consider the accuracy of the work performed by LawGeex, whose AI contract review platform answers the elementary question, "Can I sign this?" According to the company, in February 2018 its AI reviewed five non-disclosure agreements to an accuracy of 94% in 26 seconds, compared to a team of 20 "carbon-based lawyers" with decades of experience, who achieved an average accuracy score of only 85% and on average completed the review of the same five contracts in 92 minutes. 128 Thus, these AI applications are capable of vastly outperforming human lawyers in reviewing documents in mere seconds, they are less error prone, and they do not get tired, or need to sleep, eat, or rush home to have dinner with the family. 129 A similar system, RAVN, was developed in the United Kingdom and first publicly adopted by the law firm Berwin Leighton Paisner in London in 2015. This software assists solicitors with due diligence in real estate transactions by verifying property details against the UK land registry records. 130 In the past the firm would have pulled together a small team of junior lawyers and paralegals at short notice, then put them in a room to extract the data from thousands of pagesa process that would have taken weeks. [O]nce the program has been trained to identify and work with specific variables, it can complete two weeks' work in around two seconds, making [it] over 12 million times quicker than an associate doing the same task manually.

Predictive analytics
The inimitable Oliver Wendell Holmes, Jr. expressed the value of prediction in legal work: "[P]rophecies of what the courts will do in fact, and nothing more pretentious, are what I mean by the law." 133 Holmes' words remind us that litigation is fundamentally about "bargaining in the shadow of the law" -Do we have a case? What is our likely exposure? What is the likelihood of negotiating a settlement? How much is this going to cost? What will happen if we leave this particular provision out of this contract? How can we best staff this particular legal matter? 134 These prognostications, however carefully reasoned, are often biased, both by the clients' desire to win and their lawyers' own direct and indirect experiences. The development of AI tools will facilitate such prediction. 135 Machine learning is proving powerful enough to use big data to provide more objective predictions of how courts will decide discrete legal issues 136 and enable lawyers to provide more impactful advice to clients in connection with dispute resolution issues. 137 So-called "predictive analytics" is one of the most important trends in AI in the last decade, and is now also coming to the law. 138 Predictive analytics utilises machine learning techniques to detect patterns and correlations in large quantities of data, and on the basis of this analysis, to predict what will happen in situations that have not yet occurred. 139 In 2016 the Lord Chief Justice of England and Wales warned jurists that AI will be better at predicting legal outcomes of cases than the "most learned Queen's Counsel, as soon as it has better statistical information". 140  In another breakthrough development, in the autumn of 2016 a group of computer scientists, relying on AI, reached the same verdicts as judges of the European Court of Human Rights (ECtHR) in four out of five cases. 142 The ECtHR hears cases in which parties allege that their rights under the articles of the European Convention of Human Rights have been violated and not remedied by their country's courts. The researchers claim that the textual model that they used "can reliably predict ECtHR decisions with high accuracy, i.e., 79% on average". 143 This was the first time that AI successfully predicted the outcomes of a major international court by analysing case text. 144 The authors of the study note: 145 This can be useful, for both lawyers and judges, as an assisting tool to rapidly identify causes and extract patters which lead to certain decisions.
Machine learning and data analytics are being employed to predict how courts will decide particular cases. These technologies analyse not only how the facts of particular cases fit into the legal landscape but also how individual judges have decided cases in the past, and how legal doctrines have evolved over time. Some systems can already make better predictions than expert lawyers.
Lex Machina, for instance, can predict trends and outcomes in United States patent litigation more accurately than intellectual property litigators. 146 Major technology companies such as Apple, Cisco, Genentech, Intel, Microsoft, and Oracle funded Lex Machina's development of a massive and extensive dataset, with more than 130 000 cases featuring in excess of 6 million docket entries and direct access to more than 4 million documents. 147 It applies natural language processing to the vast data set to find trends that 141 Susskind law firms can use to their advantage. For example, the software can determine which judges tend to favour plaintiffs, summarise the legal strategy of the opposing lawyers based on their case histories, and determine the arguments most likely to persuade individual judges. 148 The system is now expanding to other types of complex litigation. 149 Advances in predictive analytics make it possible for parties to litigate more effectively, by predicting legal outcomes with greater accuracy and consistency, and at a lower cost. 150 With the advent of predictive analytical tools, authors anticipate significant changes in the types of lawsuits that will be brought before the courts. Benjamin Alarie, the Osler Chair in business law at the University of Toronto, believes that such technologies will change the very nature of litigation. He forecasts that advances in technology would facilitate the settlement of disputes, and that the judge would only have to decide matters that give rise to the most complex of legal questions and that require concrete legal developments. 151

Legal research
Conducting legal research can be tedious, monotonous and timeconsuming. But performing timely and comprehensive legal research by combing through texts and precedents has been an important part of legal work for centuries. 152 Machine intelligence will not only perform more of this kind of work than lawyers, but will also perform it more efficiently. Just as computers have progressively replaced humans in doing complex calculationsin fact, a hundred years ago people who made such calculations were called "computers"so will machine intelligence replace the legal search function of lawyers. 153 The practice of law is on the verge of being disrupted by systems that answer legal questions in an apparently intelligent manner. One iconic automation disruption in legal technology in this regard has undoubtedly been ROSS Intelligence, marketed as the "world's first artificially intelligent 148 Koebler 2017  For the time being, experience like mine is something people are willing to pay for. What clients don't want to pay for is any routine work.
In terms of the actual quality of legal services in the wake of the wide-spread adoption of AI, however, an acute problem is posed by technology in all fields. If lawyers delegate more and more "grunt work" to AI, their legal skills might atrophy through lack of practice, and they will tend to rely even more heavily on AI. This very fact will make it difficult for lawyers to spot and correct errors when AI makes mistakes. Of course, the best way to develop and maintain legal skills is to use them. 166 There is some merit in junior attorneys performing thankless and menial tasks, because that is, in many ways, how lawyers train to become competent and experienced. 167 For example, junior attorneys should learn how to draft contracts, tedious as the work might be, and should do it involve thousands of gigabytes of information and long periods of lawyers' time spent on relatively mundane activities. Taylor 2016 Modern Law Magazine 22. 163 Beleuz 2017 https://medium.com/@sabinabeleuz/rise-of-the-planet-of-therobolawyers-6332ead1f489. 164 Beleuz 2017 https://medium.com/@sabinabeleuz/rise-of-the-planet-of-therobolawyers-6332ead1f489. 165 As quoted in Lohr 2017 https://www.nytimes.com/2017/03/19/technology/lawyersartificial-intelligence.html. 166 Mason 2016 Modern Law Magazine 11-12. 167 Millan 2017 https://www.canadianlawyermag.com/article/artificial-intelligence-3585/; Richie and Duffy "Artificial Intelligence in the Legal Field" 5. WH GARVETT PER / PELJ 2020 (23)  25 repeatedly to be able to have a solid grasp of the skill. We should thus be circumspect and exercise good judgment with the tasks that we delegate to machines and those that we do not. 168 For example, what happens when we eliminate the early/mid-tier of attorneys by automating routine work? We may then set upon a path of eliminating the early-to-mid-career development of future lawyers by automating the work that these attorneys typically undertake. 169

Conclusion
AI may well be a revolution in human affairs, and become the single most influential innovation in history. 170 It is not merely a futuristic vision; 171 it is here, and it is already rapidly being deployed in all major sectors of the economy and society, including national defence, medicine, finance, manufacturing, transportation, the media, and arts and entertainment. 172 Given current worldwide economic conditions and the burgeoning, exponential increase in the power of technology, it is unimaginable that the legal profession will remain substantially unchanged over the next decade. Indeed, although humans are notoriously bad at predicting the future, it can confidently be said that the least likely future is that little will change in the world of law. 173 The initial applications of AI to legal practice are just the early beginnings of what will be a radical technology-based disruption of the practice of law. 174 The Fourth Industrial Revolution heralds an era in which changes in the practice of law will be pervasive, irreversible, and transformational. 175 Susskind believes that by 2036: 176 [T]o pick a date ... that will be mid-career for young lawyers of today, it is neither hyperbolic nor fanciful to expect that the legal profession will have changed beyond recognition. 168 Millan 2017  This contribution accepts as its premise the idea that technology has, and will continue to have, a significant effect on the legal profession. 177 Emerging technology will continue to be disruptive, but also holds the promise of making lawyers more productive, and enabling more lawyers to perform services that were traditionally the domain of large law firms or viewed as economically non-viable. In so doing, emerging technology can democratise the legal profession in a way that benefits lawyers and clients alike. 178 The innately conservative legal system is traditionally risk-averse and tends to be slower to adopt new technologies. However, there is every sign that a combination of technological advances and market pressure will push law firms into the AI age. 179 Highly capable systems will assume a steadily increasing share of law firm billable hours, be applied to an ever-expanding set of legal tasks, and require knowledge and abilities beyond the existing skill set of most lawyers practising today. 180 This will be the context and backdrop of the careers and working lives of tomorrow's lawyers. 181 However, it is important to steer the dialogue away from alarmist sentiments about "robot lawyers" and into the realm of how lawyers can leverage this new wave of technology in their day-to-day workflows. 182 It is not that computer systems will replace all legal work by 2025. But from now on it will increasingly become commonplace across the legal profession for successful legal businesses to convert their business processes "from human handcrafting to ever more sophisticated and capable technologybased production". 183 The wisest course seems to be for lawyers to embrace AI now, so that it can be a tool as opposed to an impediment. In future, aided by artificially intelligent machines, lawyers will be able to play a valuable advisory role for their clients, as the more structured and routine work can both be expedited and performed to a higher degree of accuracy Much of the debate surrounding the impact of AI on the practice of law has centred on automation, and not augmentation. This distinction is key. Automation will initially have the greatest impact on areas that are relatively formulaic and process-or volume-driven, such as due diligence, document review and legal research. 185 With augmentation, the AI systems facilitate the lawyer's work. While the automation of certain tasks may lead to some job losses, a lawyer's tasks also include several other elements, such as strategy, creativity, judgment and empathy that are required in advising, negotiating, and oral advocacy, none of which can yet be automated. 186 In other words, AI will not be using data to formulate complex adversarial arguments in the near future. Human elementsintuition, judgment, emotion, memory, and perceptionare all integral to that process. 187 AI will augment the work of lawyers in three ways. Firstly, lawyers will be able to do more in the same period of time. Their increased efficiency could enable them to perform additional work on behalf of existing clients, or perhaps help more clients. Secondly, because AI will enable them to perform certain tasks in minutes that historically took hours, smaller firms can perform as capably as larger firms, because AI increasingly obviates the need for the massive labour power emblematic of large law firms. Thirdly, lawyers may be able to broaden, rather than narrow, their areas of specialisation. Unshackled from the daunting task of staying abreast of a growing corpus of materials, lawyers will be able use AI not only to maintain areas of expertise, but also to develop new ones. 188 At the moment, AI creates an advantage for law firms and their clients who can afford to pay for the service. 189 But for consumers at every level, advances in machine intelligence are good news indeed. AI has every potential to democratise the legal services sector. Machine intelligence will help lawyers to increasingly deliver low-priced services, and by fuelling the development of self-help applications that are directly targeted at nonlawyers. 190 Most legal needs of low-and middle-income citizens go unmet because they cannot afford the cost of legal services. These legal needs 185 include matters as varied as counselling on how to establish a small business, and drafting a prenuptial agreement. Technological advancement in legal practice will enable the underserved middle class and even the poor to access legal services that they can afford. 191 Navigating these issues is not easy. However, it is a task that we must take on if we are committed to maintaining not only a legal system that is efficient and cost-effective, but also a legal profession that is properly equipped to deal with social conditions in what is a time of extraordinary technological innovation. 192 In the short term, the development of new technologies will empower lawyers to work more efficiently, deepen and broaden their areas of expertise, and provide greater access to justice and more value to clients. In the longer term, it is unclear exactly how dramatic the impact of artificially intelligent tools will be: "Such predictions are difficult for us to make. We are, after all, only human". 193