Lady justice on digital background (Concept of artificial intelligence lawyer)

Beyond Science Fiction: Artificial Intelligence and the Legal Industry

An Ankura Research Report

January 29, 2019

Introduction

We asked more than 350 general counsel, outside counsel, and technology professionals at global companies with annual revenue from $100 million to over $5 billion about their use of artificial intelligence (AI) and machine learning tools. Respondents overwhelmingly agree that the tools are becoming more mainstream, but less than half are currently using these technologies in practice. Read more to learn about how, where, and why to consider leveraging AI tools in your practice.

Key Findings

  • A surprising 17% of respondents believe that computers will be able to do ALL of their current work within two years. If true, such a massive change is likely to have huge implications for both the legal profession and its clients.
  • AI is currently being used by 34% of respondents and rapidly becoming a mainstream legal tool in daily use.
  • Lawyers are using AI to help them make smarter decisions, not just save money.
  • Lawyers with access to the best AI are likely to gain a competitive advantage.

Technological Transformation of the Legal Industry

From works of fiction to meticulously cited MIT research reports, humans have been predicting for the last century that robots will take over the world. Recently, however, computers have started to learn how to solve problems with continuous active learning capabilities, through a process of trial and error and the continuous refinement of algorithms, mastering increasingly complex tasks.

It’s clear from our research that the legal industry is entering a period of profound technological change, which will transform traditional jobs, careers, and organizational structures, and allow for new positions, roles, and skills to be introduced into the legal organization.

AI is already able to undertake specific tasks such as document reviews and contract assembly and analysis, which until recently could only be achieved by teams of lawyers or contract attorneys. As computing sophistication increases, it is likely that computers will be able to undertake more of the repetitive legal tasks, leaving value-added legal work to the lawyers, perhaps in areas we have not yet been exposed to, such as the legal and ethical implications of AI.

What’s the Difference?

People tend to think about and use the terms AI, machine learning, and predictive analytics interchangeably.  While all three are connected, it’s worth reviewing the definitions to better understand the impact to the legal industry.

  • Artificial Intelligence is an umbrella term that covers many different data solutions, including machine learning, deep learning, image recognition, natural language processing, bots, and voice-to-text capabilities. At present, AI tends to be applied to homogeneous information in the legal world, which consists of highly consistent data classes.
  • Machine Learning is an important application of AI that uses statistical analysis to equip systems with the ability to automatically learn and improve from experience, without explicit programming, when given new data.
  • Predictive Analytics. Within machine learning, predictive coding is a specialty that involves taking a seed set of documents, with varying relevance or privilege, out of a much larger population of documents, reviewing the seed set, and then applying the results of that coding process to the broader population.

Escalating Speed of Technological Change

As an example of how quickly things can change, it’s worth remembering that not long ago, humans found it hard to imagine computers would ever beat them at board games like chess.  For lawyers, it’s now possible to imagine a world in which computers and algorithms will be able to analyze legal documentation for a new matter and provide predictions on the probability of different case strategies and outcomes.  While there are still some limitations to this type of analysis, the possibility is becoming more real.

For example, contracts currently prepared by human lawyers may start to become available within a much shorter time frame, and compliance processes that can be slow may start to be possible in real-time due to automation.

Although the augmentation and replacement of traditional legal jobs and institutions by AI is likely to be disruptive, it is also possible that far more people and companies will end up having access to even higher-quality legal representation than is currently the case.  While the development of new roles and responsibilities continues to evolve, there is an undercurrent of analysis and assessment on the impact to lawyers at all levels, with focus on pushing new curriculum requirements into law schools to better equip young lawyers with the proper tools to enter a world driven even more by technology than it is currently.

Regardless, all lawyers will be able to benefit from AI to help them make smarter and more strategic decisions, synthesize information, and provide high-value advisory services.

Human Counsel Replacement

17% of respondents (one in five) think that computers will learn to do ALL their current work within two years. This startlingly high percentage offers an important clue that real and profound changes are happening in the workplace and that staff feel vulnerable.

Whether this statistic proves to be true or not, senior executives confirm that large teams of junior lawyers and paralegals, in certain regions or specific areas of law, are being replaced or heavily supported by AI software solutions.

Case Cruncher Alpha UK Outcomes 20% More Accurate

In the UK case of the Payment Protection Insurance mis-selling claims, AI tools developed by Case Cruncher Alpha are now approximately 20% more accurate at predicting the outcome of the claims than junior lawyers. This means that almost overnight, a lucrative legal business has been able to switch from relying on human judgment to being able to exploit low-cost judgments powered by AI.

Although there is a tendency to reassure existing staff that they will always be needed, there are definite signs that the need for human lawyers with the current skillsets is being augmented for different skills and roles within certain areas of legal work. For example, the ability of AI to undertake vast literature reviews and produce complex legal contracts using algorithms already suggests that smaller and more skilled human teams are needed to interpret and refine automated outputs, and that fewer junior staff are necessary for the actual construction of the contract itself.

Libby Jackson, partner and global head of practice for Alternative Legal Services at Herbert Smith Freehills, tells us: “How LegalTech, including AI, will affect the number of people in my business is a question I’m asked a lot. It’s something we’ve had quite fierce debates about in our leadership team too. The truthful and non-defensive answer is that we expect the volume of data to continue to increase. Data sources will continue to multiply and become more complex. Therefore, the distillation exercise that AI helps with will still require human interaction, to interpret what comes out, so I still expect to have many if not all of the people in my business.  It’s just they might be doing different things.  Humans can understand what the issues are and how these relate to their client’s legal case.”

Do you think computers will learn to do all, some, or none of your current work in the next 2 years? (by Region)

Breakdown of people who think computers will learn to do all, some, or none of your current work in the next 2 years? (by Region)

Human Counsel Augmentation

It’s more comfortable for respondents to consider technology as a tool rather than as a human replacement. In fact, 83% of respondents think that computers are likely to be doing SOME of their work within two years.

Stephen Allen, global head of Legal Services Delivery at Hogan Lovells, emphasizes the versatility of AI support: “We are using AI across the board; I have four priorities in terms of what I want it to do. One is automated document review — both in relation to litigation and for due diligence in contract reviews. The second is in relation to robotic process automation. The third is in relation to knowledge management trend analysis. The fourth is in relation to general big data analysis relating to the business of law.”

This all suggests that the 83% of survey respondents who think computers will do some of their work within the next two years can expect technology to pick up some of their most repetitive, frustrating, and time-consuming tasks. Unsurprisingly, no survey respondent feels AI will be doing NONE of their work within two years. This suggests the entire legal profession is going to be using some form of AI within a relatively short time horizon, if not between the time the survey was taken and its release.

“The first thing I try to do is to ask the client what their appetite for innovation is, and what technology they are using now. I’m ready to talk about AI, product development, and working with clients to evolve new solutions. But, more than once, the client has laughed, and said: “We only really use Word and Excel!”

But is this techno-optimism going to prove to be an exaggeration? Libby Jackson tells us that in reality, “The first thing I try to do is to ask the client what their appetite for innovation is, and what technology they are using now. I’m ready to talk about AI, product development, and working with clients to evolve new solutions. But, more than once, the client has laughed, and said: “We only really use Word and Excel!” However, with the trajectory of technical advancement expected in the coming years, the rise in the discussions around the impact on the legal industry will shortly follow with a wave of usage.

Industry Adoption: Replacement vs. Augmentation

At one end of the results spectrum—in the data-driven industries of energy, utilities, and waste management—almost half of the respondents (44%) think computers will be doing all their teams’ core functional work within two years. This dramatic finding suggests that data-rich activities, which are already highly automated, may quickly find that specialist AI tools can undertake a very large proportion of their work.

Libby Jackson explains that her firm is using AI to learn from their unstructured data and to establish valuable patterns, including “the way we price, the way we manage client relationships, and the way we offer value.”

At the opposite end of the spectrum, respondents from the agriculture, food, and beverage industry (6%) and the public sector (8%), suggest that it will be much more difficult to fully replace human workers when there is a heavy reliance on outdoor, manual, or unionized labor.

In between these two digital and physical extremes, high-value businesses such as legal firms (16%), financial services and insurance (14%), and healthcare and pharmaceuticals (12%), which handle large volumes of documentation and currently rely on human judgment, may find that computers can greatly disrupt certain tasks while medical specialists or lawyers remain central to overall decision-making.

Do you think computers will learn to do all, some, or none of your current work in the next 2 years? (by Industry)

Breakdown of people who think computers will learn to do all, some, or none of your current work in the next 2 years? (by Industry)

Robert Wickstrom, global director of Compliance and Consulting at Paul Hastings, mentions some common limitations on the uptake of technology. “There is a natural skepticism about allowing a computer algorithm to identify information to be transferred out of a company for any purpose, especially without a lawyer or subject matter expert first looking at the documents. Users are looking for flexibility and automation to aid with this human interaction.,” he says. “Given today’s development landscape, there is no reason that using analytics in e-discovery should be any more complicated than using analytics on your favorite search engine or social media platform.”

Eventually, it’s likely that algorithms will gain acceptance if they can offer access to cost-effective insights that are based on analysis of the people or outcomes.

For now, though, humans are better than AI in their ability to make genuine experienced-based judgments, and there should always be a place for the logical human brain. Stephen Allen makes this important point: “AI doesn’t replicate what a human does. AI tools enable you to work in a different way and can extract data from those documents that you ask it to — map the data and visualize it. It saves humans from having to look stuff up.”

Most positively, it is this divergence in innate capabilities that increases the potential for humans and technology to work symbiotically and to bring out the best in each other, which is where the industry will likely end up in some balance moving forward.

Replace Manual Effort, Due Diligence, and Cybersecurity

Within AI, one of the fastest-growing fields is machine learning, which involves computers learning what succeeds without ongoing instruction.

Please rank the following in the order you think machine learning could most help your business with legal decision-making.

Breakdown of people who think machine learning could most help your business with legal decision-making.

When it comes to how machine learning is being used by businesses to help legal decision-making, we find that it is primarily being used to replace manual effort and to play a part in automating regular day-to-day legal work.

The results show that machine learning is being used to enhance due diligence by pulling together documentation and putting content into formats that can be analyzed, improving cybersecurity by more quickly spotting attempted breaches and alerting necessary individuals, and by identifying deviations from normal human behavior when accessing data.

Stephen Allen tells us that in some situations it is not always worth using AI. He says: “If you’re reviewing less than 1,000 documents, it’s currently more cost-effective to use humans. Between 1,000–1,500 documents, using AI and humans to review documents works out at roughly the same price. Once you get above 1,500 documents, the cost-saving of using AI can grow to be between 40%–60%.”

Computers Best Legal Scholars when Predicting US Supreme Court Decisions

According to the journal Science, a study published in May 2017 shows that computers can do a better job than legal scholars at predicting Supreme Court decisions, even with less information. The study draws on a rich archive of historical data on legal judgments to predict the behavior of any set of justices at any time. Researchers used the Supreme Court Database, which contains information on cases dating back to 1791, to build a general algorithm for predicting any justice’s vote at any time. They drew on 16 features of each vote, including the justice, the term, the issue, and the court of origin. Researchers also added other factors, such as whether oral arguments were heard. From 1816 until 2015, the algorithm correctly predicted 70.2% of the court’s 28,000 decisions and 71.9% of the justices’ 240,000 votes, the authors report in Public Library of Science ONE. Previous research has found that even knowledgeable legal experts are only about 66% accurate at predicting cases.

AI in the Mainstream

More than one-third of survey respondents are already using AI. This result confirms that AI has entered the mainstream and is achieving important levels of usefulness for many end users. With this elevated level of adoption, the remaining 66% of respondents should consider the benefits of AI tools to remain competitive.

Does your business already use artificial intelligence to assist with legal issues?

Breakdown of people whose business already uses artificial intelligence to assist with legal issues

Respondents’ biggest single reason for adopting AI is to help humans make smarter decisions (30%). The quantity and complexity of information that lawyers regularly manage is so detailed that any tool that can help consider more available facts and speed up access to the most relevant information is invaluable.

Other considerations such as increased efficiency (25%), saving money (24%), and gaining a competitive advantage (21%) remain relevant reasons for legal departments using AI.

What do you feel would be the primary purpose for your legal department using artificial intelligence?

Breakdown of people who feel would be the primary purpose for your legal department using artificial inteligence

Overall, almost half of respondents say their companies are already using AI to provide predictive analytics (27%) or data management (21%) capabilities, reflecting the sheer quantities of complex information that lawyers now must deal with, and the difficulties associated with logically storing useful information and accessing it in a timely fashion.

When broken down by region, it becomes apparent that along with predictive analytics, information security is an internationally important use of AI. This reflects the global nature of the danger posed by cyber threats and the need for businesses to automate their responses, so that they can react rapidly and appropriately 24/7.

Where do you feel your company is getting the most value from artificial intelligence? (Yes respondents)

Yes respondents of where do you feel your company is getting the most value from artificial inteligence

It is apparent that the U.S. and UK have the most diverse and mature markets for artificial intelligence tools, with the U.S. apparently having a greater use for the digital forensics that help to identify online thefts or frauds, than the UK.

Conclusion

The use of tools to improve predictive analytics, decision-making, and due diligence is likely to become increasingly common as the amount of available information grows and the number of use cases diversifies.

Even advanced technologies like AI are simply tools and it is likely that highly skilled humans will remain vitally important for the foreseeable future — especially when it comes to determining strategies, identifying opportunities, and professionally delivering information for consumption by other humans.

While clients still expect a human to represent the matter, behind the scenes AI is likely to become vital in a wide variety of legal administrative, discovery, litigation, forensics, and analytical roles, in turn forcing human co-workers to demonstrate expertise, value for money, flexibility, and adaptability.


Survey Demographics

The analysis we provide in this report has been derived from an international survey of 351 senior legal and technology professionals, from across a broad range of industries, with annual revenues ranging from $100 million to over $5 billion.

Approximately 60% of the respondents to this report’s survey come from the US, with 30% coming from the UK and 5% each from Canada and the Middle East.

Please indicate the location of your workplace?

Breakdown of survey takers location of your workplace?

Approximately 15% of the survey’s respondents come from the financial services and insurance industries, with a further 14% coming from the telecommunications and IT community, 11% from legal firms and 11% from retail. Smaller but notable percentages of the respondents come from manufacturing (10%), energy, utilities and waste management (9%), healthcare and pharmaceuticals (9%), the public sector (7%), construction (6%), agriculture (5%), and the media and leisure industries (3%).

Please confirm the industry of your business.

Please confirm the industry of your business.

Respondents were asked a screening question about the size of their global annual revenues, and only asked further questions if the answer was in the $100 million to over $5 billion range.

Can you please indicate the global annual revenue for your organization?

Can you please indicate the global annual revenue for your organization?

General Counsels make up over a third of our respondents (34%), but a variety of other legal and technology roles associated with the delivery of high-tech legal services have also been well sampled, as can be seen below.

Can you please confirm your job title that you hold at your organization?

Can you please confirm your job title that you hold at your organization?

Other legal professionals, such as Heads of Legal, Senior Counsel, and Staff Attorneys (9%) and e-Discovery Counsel, Lawyers, Associates, and Directors, play important roles in the uptake of the latest legal technologies, so we think it is important to consider the views of different levels of seniority and specialist roles likely to be both heavy users and recipients of new technologies.

On the technology side, professionals such as CTOs (16%) and the Heads of Machine Learning and Analytics (5%), directly or indirectly responsible for the delivery of legal services within business, also offer a valuable perspective.

The responses of these two important groups has made it possible to compare and contrast the views and experiences of those responsible for the practical adoption and delivery of LegalTech with those that are more focused on the legal requirements of end users.

Overall, approximately 55% of our respondents consider themselves to have a direct role in the legal work their company undertakes, while 45% consider themselves to have an indirect role, functioning more as facilitators.

Are you directly or indirectly involved in legal work in your company?

Are you directly or indirectly involved in legal work in your company?