Mills, Michael; Uebergang, Julian --- "Artificial intelligence in law: An overview" [2017] PrecedentAULA 22; (2017) 139 Precedent 35

---

ARTIFICIAL INTELLIGENCE IN LAW

AN OVERVIEW

by Michael Mills and Julian Uebergang

This article analyses recent developments in artificial intelligence (AI) and its impact on the legal profession. It examines how AI is being implemented in legal areas such as e-discovery, legal research, compliance, contract analysis, case prediction and document automation.

INTRODUCTION

There has been much commentary about AI and its applicability to the business of law. Widespread publicity of research conducted by leading technology companies demonstrating computers’ ability to perform tasks previously undertaken by humans has generated a sense of fear that professional services roles could be replaced by robots.

AI is a term that dates back to the 1950s and has evolved into a key technology pillar for firms such as Google, IBM and Netflix. But there is a sense of mystery as to what AI is and how it can be used in practice.

There are practical examples where law firms and in-house counsel are using AI to create efficiency, both in terms of time and accuracy within a legal process, thereby creating commercial value for firms and clients alike. Importantly, legal expertise is vital to the successful deployment of AI technology, so robots will not be becoming lawyers any time soon.

WHAT IS AI?

AI is a term that is often associated with a fictional view of the future, often manifested through the eyes of Hollywood. It is important to note that AI is not just another buzzword that has gained traction through the evolution of the technology age. In fact, it dates back to 1956 when an American computer scientist, John McCarthy, first referred to the subject area as ‘the science and engineering of making intelligent machines’.

Prior to computing power becoming more accessible through the 1980s and 1990s, technology advances struggled to keep pace with academic pursuits in the area of AI.

In 1997, an IBM-owned supercomputer, Deep Blue, defeated the reigning world champion chess player, Garry Kasparov, using machine-learning techniques to determine the next best move, based on the current configuration of a chess board. In 2011, another IBM product called Watson won Jeopardy against two former winners using several AI techniques, including natural language processing (NLP), to produce the most probable responses to a question by parsing 200 million pages of structured and unstructured data.

In 2015, from AI research at Google’s Deep Mind subsidiary came a program called AlphaGo, which won five straight games against the top-ranked Go master in Europe.^[1] What is most remarkable about AlphaGo’s victory is that AlphaGo was not ‘taught’ how to play Go. Instead, its multilayer neural network learned how to play, and then how to win, by playing millions of games and observing the winning strategies.

These stunning technological advances in software that do what humans do, but better, elicits questions about how far such techniques can go and some anxiety as to the impact of such technology on employment prospects across numerous sectors. In an editorial accompanying publication of the AlphaGo research, the journal Nature wrote:

‘As the use of deep neural network systems spreads into everyday life – they are already used to analyse and recommend financial transactions – it raises an interesting concept for humans and their relationships with machines. The machine becomes an oracle; its pronouncements have to be believed.’^[2]

Two key components characterise these AI initiatives. First, the data that enables the machine to find an answer to a problem or question, whether that be a large data repository as in the case of Watson, or access to the results of millions of prior scenarios as in the case of AlphaGo. The second ingredient is the algorithms that are deployed to analyse the data to produce the desired result.

At its core, AI is the ability to mimic functions that humans associate with other human minds, such as problem-solving and learning. Central to any AI solution are algorithms to replicate the human processes, and accessibility to data to enable the algorithms to produce the desired results.

THE BUSINESS OF AI

The perception that the IBM and Google forays into AI were successful has resulted in commercial organisations developing a renewed focus on the benefits of these advances. Driving forces behind this shift in mindset include reduced infrastructure costs through the rise of cloud computing and the increased availability of, and access to, big data.

A review of AI in business was published recently by Deloitte, ‘Demystifying AI’,^[3] which suggested the term ‘cognitive technologies’ to encourage focus on the useful technologies that are emerging from the broad field of AI. The paper suggests that various sectors including banking, retail, oil & gas and media and entertainment, are benefiting from their adoption of AI technology to streamline the efficiency and accuracy of their business processes.

A practical example of the benefit of AI technology, is evident in approaches to tackle fraud detection, for example. Online payments company, PayPal, has deployed solutions such that ‘machines have freed the human detectives up to identify new types of fraud patterns, which they can then inform the AI machine about’.^[4] This enables machines to complement human expertise to deal with a global problem. Importantly, machines are not replacing the human experts.

We have also seen pure play technology companies like Facebook and Netflix using AI to good effect to aid the user experience. Facebook has deployed facial recognition software to link the profiles of users to images that are uploaded by their ‘Facebook friends’. Netflix uses an AI algorithm to predict desirable content based on the past viewing habits of its users. Both products’ algorithms rely on the platform learning from data that has been generated through historical use, to aid future use.

Historically slow to adapt to change, there is growing evidence that the legal sector is adopting AI technology to better enable the delivery of legal services, resulting in increased efficiency and accuracy.

AI AND THE LAW

AI is a broad subject area with many branches, and many significant connections and commonalities among them. AI is at work in the law in a number of areas including legal research, e-discovery, compliance, contract analysis, case prediction, and document automation. Some of the technologies and their alignment to the branches of AI are depicted in the diagram below:

As identified by PayPal, a fundamental benefit of these AI techniques is to enable lawyers to devote more of their time to more valuable (and interesting) work. Searching documents in discovery and due diligence, answering routine questions, sifting data to predict case outcomes, drafting contracts – all are faster, better, cheaper, and are becoming so with the assistance of intelligent software.

Electronic discovery

Technology-assisted review (TAR, or predictive coding) uses NLP and machine learning techniques against the gigantic data sets of electronic discovery. OpenText (which recently acquired Recommind), Equivio (now part of Microsoft®), FTI Ringtail and other technology vendors have pioneered what is now a billion-dollar industry.

The Victorian Supreme Court recently handed down the first decision of an Australian court to specifically approve the use of predictive coding technology. McConnell Dowell Constructors (Aust) Pty Ltd v Santam Ltd & Ors (No 1)^[5] involved a high-value claim relating to the design and construction of a natural gas pipeline in Queensland. Approximately 4 million documents were identified as being potentially relevant to the issues in the dispute. This was subsequently reduced to 1.4 million documents using TAR. In his findings, Vickery J’s support for the use of TAR added Australia to the list of countries, including the UK and the US, where this technology had previously received judicial endorsement.

Central to the issue of technology not replacing lawyers is the term ‘assisted review’. In the context of electronic discovery, the expertise is provided by senior lawyers with substantial knowledge of the case. They train the technology by reviewing a subset of the data and then let the algorithms determine similar documents from the initial analysis. Importantly, the lawyers are complemented by the technology, and the careful statistical thinking that must be undertaken to use it accurately. Thus, lawyers are not replaced, though they will be fewer in number.

Performed correctly, TAR is both powerful and reliable. Of course, it is important to use the technology correctly. An understanding of the principles, and even some of the statistical mathematics is essential, especially when appearing in court to argue that the outcomes are defensible and consistent with the standards of the Civil Procedure Act 2010 (Vic) and comparable rules in other courts.

A reference point for a better understanding is TAR for Smart People, a book by John Tredennick,^[6] one of the pioneers of e-discovery (and legal technology generally). TAR for Smart People is a superb guide to a critical and often misunderstood topic.

In scale and impact on costs, TAR is the success story of machine learning in the law. It would be even bigger but for the slow pace of adoption by courts, law firms and their clients.

Expertise automation

Neota Logic applies its hybrid reasoning platform, which combines expert systems and other AI techniques, including on-demand NLP and machine learning, to provide fact- and context-specific answers to legal, compliance, and policy questions.

King & Wood Mallesons recently deployed a Neota Logic application to assist international clients to determine whether a proposed deal requires Foreign Investment Review Board (FIRB) approval. This involved an expert, familiar with the FIRB legislation, creating a web-based application that replicates the decision-making they would undertake to make a determination with respect to the legislation.

Other firms in Australia have deployed Neota Logic to create similar applications which combine a lawyer’s expertise with web-based applications.

Contract analysis

General counsel recognise that their high priorities of risk management and cost reduction are served by understanding and managing the rights, obligations, and risks in a company’s contracts, and rationalising the processes by which contracts are initiated, negotiated, drafted, and managed through their life cycle, from execution to expiration.

NLP, machine learning, and other AI techniques are being applied to many aspects of the contract life cycle, including discovery, analysis, and due diligence.

There are a number of start-up organisations pioneering this space, for example:

• Kira Systems uses machine learning to enable extraction of key clause data from repositories of contracts. Reports indicate that contract review times in due diligence can be reduced by 20 per cent to 60 per cent using this technology.

• KM Standards can ‘identify common clauses, agreement structure, standard clause language, and common clause alternatives’ in a set of contracts.^[7]

• RAVN’s cognitive computing platform, the Applied Cognitive Engine, will ‘read, interpret, and summarise’ key information from contracts.^[8]

• Seal Software can crawl a network to discover, and then classify, all of a company’s existing contracts.

• Beagle, which recently announced a joint venture arrangement with Corrs Chambers Westgarth, enable ‘automatic contract analysis that learns!’^[9]

Contract analytics is well on its way to being a success story for machine learning in the law.

Legal research

Lexis Nexis® and Westlaw® have applied NLP techniques to legal research for over 10 years. No doubt others do as well. After all, the core NLP algorithms were all published in academic journals long ago and are readily available.

Recently, Canadian start-up ROSS Intelligence has been applying IBM Watson’s Q&A technology to legal research on bankruptcy topics, after winning a finalist spot in an IBM Cognitive Computing Competition among 10 universities. After building and training the data set, ROSS invites users to evaluate search results and feeds those evaluations back to the engine to continue tuning (the essence of machine learning) in the manner of recommendation engines at Netflix and Amazon® as well as Google® feedback loops, based on what we do with the search results we’re shown.

In October 2015, Thomson Reuters, publishers of Westlaw, announced a collaboration to use Watson across its information businesses. Although nothing was said publicly about Thomson Reuters’ specific plans for Watson, one could speculate that the vast trove of legal content in Westlaw and the army of subject-matter experts in the company could together do impressive things to improve legal research. Watson needs big data and training, at least initially, by people: Thomson Reuters has both.

On 1 February 2016, at a private ‘innovation summit’, Thomson Reuters allegedly teased the legal industry with hints that Watson would provide a beta service for financial services regulation by the end of the year.

Take note of the timeline: even a company with the immense resources of content and expertise of Thomson Reuters, even in partnership with IBM, needs more than a year to get to beta with an AI legal research product. Why? Because neither AI nor Watson is magic. It takes time, human expertise, and painstaking effort to assemble useful data sets, analyse the content, train the algorithms, and test the results. The broader the targeted topic, the greater the effort.

Outcome prediction

Lexis Nexis’ Lex Machina, after building a comprehensive set of intellectual property (IP) case data, uses data mining and predictive analytics techniques to forecast outcomes of US IP litigation. Recently, it has extended the range of data it is mining to include court dockets, enabling new forms of insight and prediction. For example, the Motion Kickstarter application 'makes it easy to draft winning motions by helping attorneys compare the arguments and motion styles that have been successful before a specific judge’.^[10]

The LexPredict application has built models to predict the outcome of US Supreme Court cases, at accuracy levels challenging experienced Supreme Court practitioners. Premonition says they are using data mining and other AI techniques ‘to expose, for the first time ever, which lawyers win the most before which judge'.^[11]

While Lex Machina has not yet been launched in Australia, it seems inevitable that predictive tools can leverage big databases of law firm case and billing data to offer outcome predictions as well as cost and rate benchmarks.

THE FUTURE

Traditionally, experimentation with technology, AI or otherwise. has not been the realm of the law firm. The ‘more-for-less’ demand on in-house counsel is, in turn, pressuring the legal establishment to focus on efficiency, including the use of technology. It seems that lawyers are no longer just in the business of managing client relationships and risk, they are also being asked to create efficient and accurate solutions, manage costs, and use technology to create a competitive advantage when servicing clients.

As we have highlighted, AI technology is being used to good effect in many areas of legal service delivery.

Cognitive technologies in the law are riding a wave of ever-smarter algorithms, infinite scaling of computer power by faster chips and cloud services, intense focus by companies led by seasoned data experts, and an ever-greater demand from clients for cheaper, faster, better services.

Michael Mills is the co-founder & Chief Strategy Officer of Neota Logic, a provider of intelligent software. EMAIL [e-mail].

Julian Uebergang is Managing Director APAC of Neota Logic. EMAIL [e-mail].

This is a reprise of a Thomson Reuters Legal Executive Institute White Paper, ‘Artificial Intelligence in Law: The State of Play 2016’, first published online on 24 March 2016 <http://legalexecutiveinstitute.com/artificial-intelligence-law-state-play-2016/> .

---

^[1] D Silver, A Huang, CJ Maddison, A Guez, L Sifre, G van den Driessche, J Schrittwieser, I Antonoglou, V Panneershelvam, M Lanctot, S Dieleman, D Grewe, J Nham, N Kalchbrenner, I Sutskever, T Lillicrap, M Leach, K Kavukcuoglu, T Graepel and D Hassabis, ‘Mastering the game of Go with deep neural networks and tree search’ (2016) 529 (7587) Nature pp484-9.

^[2] Editorial, ‘Digital Intuition’ (2016) 529(7587) Nature p437.

^[3] D Schatsky, C Muraskin, R Gurumurthy, ‘Demystifying artificial intelligence: What business leaders need to know about cognitive technologies’ on Deloitte University Press (4 November 2014) <https://dupress.deloitte.com/dup-us-en/focus/cognitive-technologies/what-is-cognitive-technology.html>.

^[4] P Crosman, ‘How PayPal Is Taking a Chance on AI to Fight Fraud’ on American Banker (1 September 2016) <https://www.americanbanker.com/news/how-paypal-is-taking-a-chance-on-ai-to-fight-fraud>.

^[5] McConnell Dowell Constructors (Aust) Pty Ltd v Santam Ltd & Ors (No 1) [2016] VSC 734.

^[6] J Tredennick et al, TAR for Smart People: How Technology Assisted Review Works and Why It Matters for Legal Professionals, Catalyst Repository Systems Incorporated, Colorado, 2015.

^[7] KM Standards, Our Services (2014) <http://kmstandards.com/services.html> .

^[8] RAVN Systems, Applied Cognitive Engine <https://www.ravn.co.uk/products/applied-cognitive-engine/>.

^[9] Beagle Inc., Artificial Intelligence Contract Analysis (2017) <http://beagle.ai/> .

^[10] Lexis Nexis, Legal Analytics Apps, <https://lexmachina.com/legal-analytics-apps/>.

^[11] Premonition, <https://premonition.ai/>.