Dispelling the AI myths in the legal sector

Artificial intelligence (AI) research has a long history, but even before computers existed Hollywood has been implanting a fear of intelligent machines into the public psyche, with more than sixty films to date depicting stereotypical threats to humankind. But only in the last few years has AI featured as a regular news story, invariably illustrated by images of suit-wearing robots seated around a board table. As intelligent machines have become mainstream, all kinds of media outlets serve up stories - some of them post-truth - about how robots will replace humans across a wide range of sectors: thanks to their superior speed and intelligence many current jobs will become obsolete.    

It will happen fast - within the next generation, according to those seeking to grab our attention, leaving humans to find alternative employment, if they can. As soon as 2021, robots will eliminate 6% of all US jobs, according to market research company Forrester while the World Economic Forum (WEF) predicts a loss of 7 million jobs within four years.   

Look further ahead and some predict that up to half of all current jobs will disappear by 2050, possibly sooner. This dystopian vision is notably striking in the legal sector, where AI systems are anticipated to become the masters of document drafting leaving many young lawyers without a future. Bad news sells. Multiple commentators have fallen over themselves to suggest AI would scythe through lawyers in commercial law firms, mercilessly cutting their numbers.   

We’ve been here before, of course. Legal futurologist Richard Susskind has been predicting the end of lawyers for a generation or more, even if his original algorithmic rule-based programming approach to artificial intelligence, formulated in the 1980s, was long ago dismissed by many of the leading players in the AI community. 

According to today’s prophets of doom, the real winners in this Brave New AI World will be a new breed of legal technologists and in particular, the tech companies, such as IBM with its celebrated Watson, a question answering (QA) computing system which has become the most famous supercomputer in the world. In 2011, it won $1m first prize in the US game show Jeopardy, while its predecessor IBM system, Deep Blue, defeated Garry Kasparov a world Chess champion, and more recently, a comparable system, Google AI, beat Lee Sedol one of the world’s leading GO players.    

Beyond these notable gaming achievements, Watson has made a substantive practical contribution to the world of work. Since 2013, it has been used for diagnostic applications by US doctors: its utilization management decisions in cancer treatment are helping them to make more informed diagnoses based on analysis of historic medical data and body scans. Law firms are also beginning to use multiple AI applications, including Watson based systems, to enhance their legal research functions. This allows young lawyers to deliver much faster and more comprehensive analysis of contracts and precedents. 

Last year saw the first approval by an English court in the use of predictive coding in document review when an AI system was trained to evaluate large volumes of data to identify and classify relevant documents. In starting to make a real impact on legal practice, the beneficial applications of AI combine a spread of different technologies. Machine learning, underpinned by neural networks and powered by clusters of parallel computers gives us the ability to analyse vast datasets for patterns. Today’s AI is ever more capable in performing multiple complex, repetitive tasks analysing information much faster than humans ever could.   

A generation of lawyers have used laptops, smart phones and internet data sources as tools. AI is set to become an important supplement to these existing technologies. But in assessing what AI has already achieved for the legal sector and more significantly, what today’s approaches to AI have the potential to achieve, one factor separates fact from fiction: AI in all its current manifestations will remain a tool. A very useful and even sometimes surprising tool, but a tool nonetheless. It will not replace lawyers. Instead it will liberate them to spend more time on other work, which is less laborious and more creative or client focused. The knowledge worker generation will be advanced not replaced.  

So what is the evidence as to how big law firms are embracing and using AI? The big five UK law firms, better known as the magic circle, provide a good starting point. Linklaters was the first of their number to sign a deal with an AI service provider, UK-based RAVN.  Clifford Chance and Freshfields Bruckhaus Deringer have entered separate partnerships with Canadian AI software provider, Kira Systems. Meanwhile, Slaughter and May is working with Luminance Technologies to test AI technology which will streamline the due diligence process. Finally, Allen & Overy has created a service to assist banks cope with complex regulations: MarginMatrix. This automates document drafting with the drafting time reduced from three hours to three minutes. 

It might be reasonable to expect that such technological advances might provoke a reduction in lawyer headcount. But at the magic circle, this has not happened. Set against the background of these AI developments, the number of lawyers that they employ has either remained the same or increased slightly.   

The best predictor of future size – the number of trainees that they take on each year - has not decreased. Indeed, compared with five years ago, the number has increased. The reality is that while they are busy exploring and implementing AI systems into their offering to clients, big commercial law firms are not cutting their headcount. If anything, the reverse is true: overall training contract numbers in the UK increased by 9% in 2016 from 5,000 to 5,450, according to the latest research.   

So forget the scare stories, Skynet is not about to come online. There is no doubt however that machine learning techniques, neural networks and parallel processors, present many exciting possibilities. Just like their medical counterparts, tomorrow’s lawyers will increasingly make use of AI as a diagnostic tool to make their output more efficient and more accurate. But - and it is an important but - however fast and flexible AI becomes, it will remain a tool, not capable of delivering creative, independent judgment. These essential human elements will remain the preserve of a lawyer.   

Because however clever and capable our machines become, they can never become human. We are so much more than a series of algorithms. Our understanding of different situations and how to handle them cannot be replicated by computer code. Although AI has successfully evolved from basic data processing, to the extraction of information, and now arguably the simulation of strong subject knowledge, the distinctly human characteristics of wisdom and empathy can never be part of a computer’s skillset.   

Some may argue that if simulation becomes advanced enough, it will become indistinguishable from the real thing. Well, the UK Meteorological (Met) office has a £100m Cray supercomputer which can perform 16,000 trillion calculations per second, requiring a dedicated power supply of 2.7Mw and it uses the latest probabilistic techniques to produce advanced weather forecasts; but it never actually rains inside the computer.   

Some may also point to the recent “passing” of the Turing Test where human judges have to decide whether they are in typed conversation with a human or an AI. The most advanced modern techniques have been able to persuade just 33% of humans that they were in conversation with a 13 year-old boy. An impressive feat in many respects, but would you chose that as your legal representative?    

Lawyers should be excited and inspired by what the current generation of AI research can achieve. It will certainly save time and minimise laborious tasks. As they become increasingly invaluable over the next twenty years, AI systems will help lawyers to perform their job in many different ways. But AI will only ever be a useful tool: it will remain our servant rather than becoming our master.   

Image Credit: Enzozo / Shutterstock