Things are speeding up. In 1997, IBM Deep Blue defeated the reigning world champion chess player. In 2011, IBM Watson defeated the two best human Jeopardy players. In 2014, Google Deep Mind learned to play and win 46 old Atari arcade games. In 2015, Deep Mind learned to play Go, the most complex board game humans have invented for themselves. In 2016, Deep Mind beat the world-ranking Go master.
In 2017, thousands of engineers at Amazon, Apple, Facebook, Google, IBM, and Microsoft are building ever more sophisticated AI tools, available inexpensively in the cloud to anyone with a problem to solve. Scores of young companies also built new AI-based services for the legal market in 2017, joining the established legal research and e-discovery companies who have used AI techniques for years.
So, what is AI?
AI is not magic. It is just software, just algorithms—many of them well-known for many years—applied to data. What's different today is that the computing power necessary to drive the algorithms, is orders of magnitude cheaper, ubiquitously available in the cloud, and able to reach the vast stores of data that the algorithms need to feed themselves.
Facebook can tell your friends from your cats in photos because its algorithms, on thousands of computers, power through millions of images, most of them helpfully labeled by us as Cat or Friend. Google can translate English to French because it finds and compares millions of documents translated by humans from one language to the other. Siri can answer your questions (some of them anyway) because it analyzes more than a billion queries every day. IBM Watson can assist oncologists with diagnosis and treatment plans because it has digested the medical literature on cancer and been taught by hundreds of doctors over thousands of hours about the rules governing relationships among inputs (patient characteristics, test results, research data) and outputs (statistically relevant literature, probable diagnoses, potential treatments).
What is AI doing in law practice?
Although there may be no "AI Inside" label on the box, the techniques of AI are used in many areas of law practice, as illustrated in this map:
(Like all terrain, AI in the law changes daily: the mapmaker welcomes additions and corrections.)
Here we can take a quick tour of key attractions.
Legal publishers have applied natural language processing and other machine learning techniques to legal research for many years. The hard (very hard) work is practical implementation against good data at scale. Legal research innovators like Fastcase, RavelLaw, and ROSS have taken a fresh look at the algorithms and exploited them to create distinctive capabilities such as result relevance ranking, "bad law" flags, judge analytics, conversational interfaces, and visualization.
In October 2015, Thomson Reuters, publishers of Westlaw, announced a collaboration to use Watson across TR's information businesses. A rumored first product is a financial regulatory compliance information service.
Machine learning in e-discovery. New? No, but still important, challenging, and not yet as widely accepted by the profession as, say, typewriters. Technology-assisted review (TAR or predictive coding) has been proven to be faster, better, cheaper, and much more consistent than human-powered review. See, for example, Cormack & Grossman, Evaluation of Machine Learning Protocols.
Yes, it is assisted review, in two senses. First, the technology needs to be assisted; it needs to be trained by senior lawyers very knowledgeable about the case. Second, the lawyers are assisted by the technology, and the careful statistical thinking that must be done to use it wisely. Thus, lawyers are not replaced, though they will be fewer in number.
Lex Machina, after building a large set of intellectual property case data, uses data mining and predictive analytics techniques to forecast outcomes of IP litigation. Recently, it has extended the range of data it is mining to include securities and antitrust.
Good predictions depend on good data—well-structured (consistent across cases and jurisdictions, normalized, etc.), fine-grained (facts and findings expressed in detail), and big enough (not everything, but enough to be statistically valid.) Traditionally, law has not been data driven. We think about documents, not about data. No wonder that prediction is a hard problem, and that so far it has been solved in very narrow areas.
Self-Service Guidance & Compliance
Expert systems go beyond automation of documents to enable automation of legal guidance and processes. For example, the Internal Revenue Service offers a suite of Interactive Tax Assistants (created with Oracle's Policy Automation software) that answer questions about complex issues of federal tax. Riverview KIM augments matter intake for corporate law departments.
Neota Logic applies its hybrid reasoning platform, which combines expert systems and other artificial intelligence techniques, including on-demand machine learning, to provide fact- and context-specific answers to legal, compliance, and policy questions. (Disclosure: I am Cofounder and Chief Strategy Officer of Neota Logic.)
New Mexico Legal Aid and other not-for-profit legal services organizations use expert systems technology to recommend the most suitable agency for people in need, to assess their eligibility for agency services, and, when they are not eligible, to route them to self-help information.
General Counsel recognize 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. Natural language processing, machine learning, and other AI techniques are being tailored specifically to many aspects of the contract lifecycle, including discovery, analysis, and due diligence. For example:
- Kira Systems reports that contract review times in due diligence can be reduced by 20 to 60 percent.
- KM Standards can "identify common clauses, agreement structure, standard clause language, and common clause alternatives."
- LawGeex reviews draft contracts to identify clauses that "are rare, missing, or potentially problematic."
- RAVN will "read, interpret, and summarize" key information from contracts.
- Seal Software can crawl a network to discover, and then classify, all a company's existing contracts.
What should firms be doing about AI?
The tools and techniques of AI enable lawyers to:
- Serve more clients more effectively at lower cost.
- Create new revenue streams not dependent on the billable hour.
- Focus their time and expertise on work that requires the uniquely human and professional skills of empathy and judgment.
- Increase access to justice by meeting the legal needs of the poor and middle class.
These are central business and professional goals in the new normal of legal services. As the AI wave rolls forward across law practice—as it is doing (more rapidly) across all of business—law firms must:
- Learn, and keep learning, about AI and the companies bringing AI to law.
- long-proven solutions such as TAR in e-discovery.
- Create a culture and process for continuous experimentation in every practice area.
Back to Contents