Will law be artificial?

June 2018 | EXPERT BRIEFING | LITIGATION & DISPUTE RESOLUTION

financierworldwide.com

We are entering an era where knowledge of the law will no longer be limited to flesh-and-blood lawyers. Machines are becoming increasingly familiar with our statutes and case law. This trend will only expand as artificial intelligence (AI) and machine learning (ML) algorithms become more elaborate.

In basic terms, AI consists of supplying information to machines and software, which, in accordance with predetermined rules governing its use, try to reach conclusions. In essence, it is a way of mimicking human reasoning. On the other hand, ML algorithms are computer models designed to make sense of large amounts of data, by perceiving patterns in (and by extracting rules from) them.

Undoubtedly, there is a great potential for the use of AI and ML in legal matters. Contrary to what some might think, lawyers do have a knack for innovation. In particular, lawyers are always trying to find new ways to carry out their day-to-day tasks in a more efficient manner. An important part of a lawyer’s job is to analyse large volumes of information, which takes a considerable amount of time and is also expensive for clients. Therefore, finding a way to assess information in a faster and more accurate manner would mean freeing up time to think about the strategic (and more creative) part of a lawyer’s intervention in any given case.

At the same time, legal procedures are becoming increasingly digital, with parties filing their written statements and documents in proceedings by electronic means, and the courts publishing their orders and rulings on websites. Every day, millions of items of data are being generated around the (legal) world, which can be used to feed algorithms, making it easier to introduce AI and ML into legal matters.

Although at this stage we cannot be sure about the impact AI will have on the legal profession, and specifically on what lawyers do, it is still possible to make some predictions. For that purpose, all it takes is to compare what we know about the profession and the known capabilities of AI. With that in mind, a fair analysis would point out that a substantial part of what lawyers do will, in the short to medium term, be carried out by computers. Translations, legal research, drafting of agreements and (for the time being) simple procedural documents, all of this may be (and, in some cases, already is) carried out by a machine better and faster than by a human being. Additionally, ML can help to detect patterns in case law, and to come up with forecasts for cases brought to lawyers by clients.

In fact, there are already some examples. Academics from Washington University developed an algorithm to forecast Supreme Court decisions which revealed an accuracy rate of 75 percent (in predicting the Court’s rulings in 2002), and researchers from University College London put together an algorithm that predicted with a 79 percent accuracy rate the decisions of the European Court of Human Rights. Furthermore, the market is already brimming with applications and software specifically aimed at forecasting court rulings.

AI can be put to other uses in the legal environment. Some American states use software named COMPAS, which is an algorithm designed to assist criminal judges in assessing the likelihood of a defendant reoffending, which is an important factor when sentencing and when ascertaining whether parole may be applied for by the defendant. The algorithm takes into account several criteria, like the defendants’ gender, age, ethnic and social background, and criminal record. The reasoning behind the use of software like this is that an algorithm fed with large amounts of data may determine the risk of a defendant reoffending without being influenced by prejudices that invariably arise when a human judge is involved.

It should come as no surprise that this kind of software has stirred up some controversy. In the state of Wisconsin, where COMPAS is used, the state Supreme Court heard a case where the defendant claimed that the algorithm violated a number of essential legal principles, like non-discrimination or equal treatment, and the need to properly substantiate court rulings.

In any event, it is worth noting that, according to a study published recently, programmes like COMPAS are no more accurate than humans in forecasting the risk of reoffence of defendants. Two researchers at Dartmouth College (New Hampshire) carried out a study where they randomly selected 1000 defendants, from a database of more than 7000 people arrested between 2013 and 2014 whose cases had been evaluated by COMPAS. The researchers then put together short profiles for each of the defendants, listing their gender, age and previous convictions, and asked a group of 400 volunteers to try to predict the likelihood of the defendants being arrested again in the next two years. After the end of that period, the researchers analysed the volunteers’ answers and concluded that, while COMPAS had predicted with a 65 percent accuracy rate the cases where a defendant would be arrested again, the volunteers had got their forecasts right in 67 percent of cases – which is essentially the same in statistical terms.

The results of this study suggest that the usefulness of AI and ML in law is still at an early stage. However, it would be reckless to think that this will be the case for a long time. There are hundreds (perhaps thousands) of companies around the world developing apps and software aimed at increasing the efficiency of how lawyers execute some of the tasks that they have always carried out, or at simply rendering human intervention dispensable. Fields like due diligence, competition law and intellectual property law are obvious targets. And they will only be the first wave – others will follow.

The philosopher Jeremy Bentham amusingly (and aptly) said that “lawyers are the only persons in whom ignorance of the law is not punished”. With the arrival of AI and ML in the legal profession, even a relative ignorance of the law will no longer be an option.

Luís Pais Antunes is managing partner and Lourenço Noronha dos Santos is an associate at PLMJ Intellectual Property. Mr Antunes can be contacted on +351 21 319 7300 or by email: luis.paisantunes@plmj.pt. Mr Noronha dos Santos can be contacted on +351 21 319 7300 or by email: lourenco.santos@plmj.pt.

Luís Pais Antunes and Lourenço Noronha dos Santos

PLMJ Intellectual Property