An update on the intersections between artificial intelligence and the legal industry
Over the past few years, the topic of artificial intelligence (AI) has begun to appear with increasing frequency in the legal industry press, filled with both promise and threat. A 2016 report from Deloitte “said that ‘profound reforms’ will occur in the legal sector over the next decade, estimating that nearly 40 percent of jobs in the legal sector could end up being automated in the long term.” Although impact that dramatic and change that widespread are still somewhere in the future, some impacts and changes are starting now.
For example, JP Morgan revealed recently that it had replaced “360,000 hours of work each year by lawyers and loan officers” reviewing and working with contracts with an artificial intelligence “program, called COIN, for Contract Intelligence.” And, Ernst & Young analysts project that “artificial intelligence (AI) and machine learning will be at the center of many corporate deals” this year.
This update provides a brief overview of the topic, and reviews several recent news items to investigate the current impacts of these technologies on the legal industry.
WHAT WE TALK ABOUT WHEN WE TALK ABOUT AI
In the broadest possible terms, artificial intelligence refers to “intelligence exhibited by machines,” except that our colloquial thinking about what qualifies as intelligence continually changes. We general think of tasks that would require human intelligence, but as soon such a task is successfully automated, our thinking about it changes. A recent New York Times Magazine article on Google’s use of AI to reinvent its translation tools captured the difficulty with the term well:
The phrase “artificial intelligence” is invoked as if its meaning were self-evident, but it has always been a source of confusion and controversy. Imagine if you went back to the 1970s, stopped someone on the street, pulled out a smartphone and showed her Google Maps . . . Google Maps would almost certainly seem to her a persuasive example of “artificial intelligence.” In a very real sense, it is. It can do things any map-literate human can manage, like get you from your hotel to the airport – though it can do so much more quickly and reliably. . . .
Practically nobody today, however, would bestow upon Google Maps the honorific “A.I.,” so sentimental and sparing are we in our use of the word “intelligence.” Artificial intelligence, we believe, must be something that distinguishes HAL from whatever it is a loom or wheelbarrow can do. . . . The goal posts for “artificial intelligence” are thus constantly receding.
To speak more specifically, distinctions are drawn between strong and weak AI and between general and narrow AI. Strong AI would be AI that simulates human cognition and reasoning; nothing approaching a strong AI has yet been created. Weak AI is AI that simulates human behavior (e.g. playing chess, tagging photos, etc.) but does so by other approaches than simulated human cognition. Recent developments in neural nets and other machine learning approaches fall in between these two camps, using learning techniques modeled on human ones but without attempting to model human cognition. “Not exactly like the brain, but inspired by it.”
An additional distinction is drawn between general artificial intelligence, which would be AI capable of general reasoning that could be applied in a variety of contexts, and narrow artificial intelligence, which is AI capable of accomplishing only a specific narrow task (e.g. facial recognition, language translation, etc.). All the tools currently in use and development in the legal industry are narrow artificial intelligences developed to accomplish specific tasks, but that does not mean their impact is not going to be significant soon.
MEET ROSS, YOUR NEW VIRTUAL ASSOCIATE
IBM’s Watson became a household name after its Jeopardy victory in 2011, but since then the so-called “question answering machine” has been deployed in a variety of real-world applications across seventeen different industries, most prominently in healthcare. Now, it’s coming to join the legal industry:
Those in-the-know when it comes to legal technology are likely aware that IBM is in partnership with a variety of organizations to utilize Watson, too. Among those are Thomson Reuters, which in October 2015 announced its intentions to apply Watson in its own technology deployment, and LexisNexis, for powering its Lexis Search Advantage. Also utilizing the IBM technology for legal research was ROSS Intelligence, which initially focused on bankruptcy law but is currently being developed for other practice areas as well as business intelligence.
A team of college students initially created Ross as a “prototype legal assistant” for the IBM Watson Cognitive Computing Competition in 2015. After winning second place, multinational law firm Dentons arranged financing for the team through its Nextlaw Labs technology incubator. ROSS Intelligence has since added other prominent customers like Latham & Watkins.
Baker & Hostetler also recently announced their deployment of ROSS within their bankruptcy practice and indicated other firms would be making similar licensing announcements soon. ROSS will perform a variety of tasks for the practice similar to some of those performed by junior associates and paralegals:
When asked a question, Ross postulates hypotheses, researches and then generates a response, backing up its conclusions with references and citations. It narrows down research results by selecting only the most highly relevant answers, and presents the answers in a more casual and understandable way.
The AI also reportedly learns from its own experience; the more you interact with it, the more it gains speed and knowledge.
A TECHNOLOGICAL ARMS RACE
Beyond just the deployment of ROSS as a virtual team member at several firms, other law firms are also investing heavily in a range of AI tools for a variety of applications. Some recent examples include:
- “DLA Piper will use artificial intelligence technology by Kira Systems for due-diligence document review in mergers and acquisitions.”
- “Magic circle law firm Clifford Chance has for the second time in four months deployed artificial intelligence (AI) in a standalone product aimed at assisting clients to comply with complex regulatory changes. . . . The law firm advertised the product by saying that for a fixed up-front fee, clients could ‘access complex legal advice at a fraction of the price of individual analysis.’”
- “An Adelaide-based tax lawyer has created an artificial intelligence (AI) that’s become such an expert in Australian tax law he predicts tax agents will be gone within five years. . . . Tania Waterhouse, principal director of Waterhouse Lawyers – Ailira’s first client in Sydney – said that her firm’s tax practitioners could ‘find exactly what they are looking for in a matter of seconds.’”
- “Law firm MinterEllisonRuddWatts has invested $2 million in a joint venture exploring the potential use of artificial intelligence for legal services.”
- “A start-up using artificial intelligence (AI) to filter news and information, offering services that include being able to track how firms are perceived in the media and also give them intelligence tailored to their clients’ businesses . . . said it has Allen & Overy, DLA Piper and Mishcon de Reya on its books, among other big legal names.”
- “Littler Mendelson director of data analytics Zev Eigen said that using AI and predictive coding for hiring can help firms get away from traditional standards of success, like law schools or grades obtained in law school, which tend to be poor indicators of professional performance.”
REPORTS OF OUR DEATH MAY BE GREATLY EXAGGERATED
Finally, one recent study finds that artificial intelligence and machine learning technology may not pose as much threat to attorneys’ jobs as has been feared. The study evaluated data on attorneys’ time usage drawn from analysis of law firm invoices. The study defined categories of legal work, determined their portions of total legal work, and then assessed the likely impact of AI technology on each category.
- “Document management, fact investigation, legal writing, advising clients and other communications or interactions, court appearances and preparation, and negotiation”
- Approximately 55% of invoiced hours
- “Light” AI technology impact
- “Case administration and management, document drafting, due diligence, legal research, and legal analysis and strategy”
- Approximately 40% of invoiced hours
- “Moderate” AI technology impact
- “Document review – defined as reviewing documents for purposes of discovery in litigation or government investigations”
- Approximately 4% of invoiced hours
- “Strong” AI technology impact
So, 95% of legal work may only lightly or moderately impacted, and only 4% of legal work is likely to be strongly impacted. Certainly, a sea change is coming for document review and those who specialize in it, but it may be premature to sound the alarm about robots coming for the legal industry.
As we can see from these recent news stories, artificial intelligence tools are making real inroads in the legal industry: as support tools and services for law firms; as tools created by law firms for clients; and, as virtual research associates. As we noted above, these are all narrow, task-specific AIs, but the range of tasks for which such tools have been created or trained is growing rapidly. While AI may not be strong enough or general enough yet to replace most legal work, it is clearly going to cause dramatic changes in the narrow areas where it currently excels, reducing research times from days to minutes, providing superhuman consistency and reliability in contract work, and automating the review of large data and document collections.
MATTHEW VERGA, JD
VP, Marketing Content
Matthew Verga is an electronic discovery expert proficient at leveraging his legal experience as an attorney, his technical knowledge as a practitioner, and his skills as a communicator to make complex eDiscovery topics accessible. A ten-year industry veteran, Matthew has worked across every phase of the EDRM and at every level from the project trenches to enterprise program design. As VP, Marketing Content, for Advanced Discovery, he leverages this background to produce engaging educational content to empower practitioners at all levels with knowledge they can use to improve their projects, their careers, and their organizations.