Artificial Intelligence - Articles

All Content

Posted by: George Lewis, Walter Battle & Nicole Berkowitz on Mar 1, 2018

Journal Issue Date: Mar 2018

Journal Name: March 2018 - Vol. 54, No. 3

State of the Industry and Ethical Issues

“Artificial Intelligence is the future of all mankind. There are huge opportunities, but also threats that are difficult to foresee today. Whoever becomes the leader in this sphere will become the ruler of the world.”

— Vladimir Putin speaking to Russian students on Sept. 1, 2017

State of the Industry

Conceptual photo illustrating artificial intelligence.
Photos © iStock

What Is Artificial Intelligence?
Artificial intelligence (AI) is the science and engineering of teaching computers how to learn, reason, communicate and make decisions. It is about making intelligent machines, especially intelligent computer programs. Cognitive tools are trained vs. programmed — they learn how to complete tasks traditionally done by people, by looking for patterns in data, testing the data, and finding/providing results.

The Current State of the Art
Legal AI is still in its early stages in regard to its sophistication and potential. There are many opportunities, but also many challenges, to using AI in the legal profession. Challenges include: data privacy; law enforcement access to data; lack of regulatory framework and standards; and legal ethics. Experts are signaling that Legal AI will become a major disrupter in the legal industry, much like email was. Experts predict that overall spending on AI by companies will grow from $8 billion in 2016 to $47 billion in 2020, up almost 600 percent.[1] For comparison, the current size of the U.S. legal market is estimated to be $437 billion.[2] Research shows 75 percent of law firms spend somewhere between 0 and 4 percent of their total revenue on technology, compared to 5.2 percent for the average U.S. company.[3]

Current Uses for Artificial Intelligence[4]

Law Firms
Currently, law firms are primarily using AI in “search-and-find type tasks” in electronic discovery, due diligence and contract review. Various programs, such as Kira, can be used in contract review to identify particular contract clauses, such as noncompete provisions and change-of-control provisions. Users of Kira report reducing the amount of lawyer time required for contract review by 20 to 60 percent. Kira works with many U.S. and global law firms, but its flagship partnership is with Deloitte.

The Ross Intelligence program uses natural language processing and machine learning capabilities to identify relevant legal authorities. Ross runs its legal research bot off of IBM’s cognitive computing system, Watson. Statistics published by Ross show that the Ross program retrieves more relevant authorities than lawyers would through traditional Boolean or Natural Language searching. The Ross website lists a number of law firm users, including Latham & Watkins, Baker Hostetler, Bryan Cave, Simpson Thacher, and others.

Lawyers in smaller firms also have reported success using Ross. Ross is also developing a legal memo service, where the program produces a very rough draft of a memo and humans edit the memo and send it to the user the next day.

LexisNexis acquired Lex Machina in 2015, which was created in 2010 at Stanford University with support of companies like Apple and Cisco. In June 2017, LexisNexis acquired Ravel Law, a legal research and analytics tool that helps craft court arguments. Lex Machina and Ravel Law can pore through court decisions and data to predict the behavior of judges and lawyers. These can help law firms guide litigation strategy. Lex Machina’s website lists Kirkland & Ellis, Paul Hastings, Duane Morris, Cleary Gottlieb and others as customers. Lex Machina users can also set up alerts to notify them when new cases are filed, which can assist firms scouting for new business.

Premonition’s legal analytics software can rank arbitrators based on past decisions, help select expert witnesses based on their persuasiveness and past case results, and can help analyze the court, judge and opposing counsel based on their win rate and results. It can also help with tailoring client pitch data, finding local counsel and recruiting new litigators.

Tech startup Casetext’s automated CARA (Case Analysis Research Assistant) and its Brief Finder tool are geared toward assisting litigators. Casetext’s website allows users to securely upload briefs into CARA, at which point the AI researches Casetext’s entire database of U.S. law and finds relevant case law in milliseconds. Casetext has made its software available to federal district courts at no charge.

WordRake is a document proofing software, which not only corrects typos, but also makes semi-substantive editing suggestions. It recognizes when words are overused, edits sentences for brevity and clarity, and catches passive phrasing. It is like having a second set of eyes review your writing (and may be particularly helpful for solo practitioners), but it is not flawless. Lawyers must still review the suggestions made by the program, as there is the potential that some edits could change the meaning of the sentence.

Counsel On Call has recently acquired DSicovery LLC, a data management company. DSicovery uses the Catalyst platform, including Insight Predict, which utilizes continuous active learning on eDiscovery projects. It ranks documents based upon relevancy and prioritizes them. Then, as attorneys mark a review document as relevant, Predict learns and finds more documents like it, and moves those to the top of the review pile.

Inside Counsel
Inside counsel are also seeing the benefits of AI. For example, J.P. Morgan has created a program called COIN for “Contract Intelligence,” which interprets commercial loan agreements and cuts down on loan-servicing mistakes. Reviewing these agreements previously consumed 360,000 hours of work each year by lawyers and loan officers. Think Smart has a software program that allows companies to generate Non-Disclosure Agreements, store them in one place, and track their status.

Clause is a program that creates “intelligent contracts” that can detect when a set of prearranged conditions are met or broken. The contract can be integrated into enterprise workflow processes. For example, if you violate a provision of a “smart contract,” such as by failing to ship goods on time, the program would automatically adjust the price in accordance with the contract.

Facilitating Access to Justice
There are also various bots that laypersons can use directly. For example, one bot called “DoNotPay” can assist people in fighting traffic tickets. It asks the user various questions about their case and generates a letter that can be filed with the appropriate agency.

DoNotPay’s founder Joshua Browder estimates that the DoNotPay bot has already saved people $9.3 million disputing 375,000 parking tickets.[5] Browder has recently added new functions to the bot, so it can now assist people in tackling issues in 1,000 legal areas for free. For example, it can assist people in demanding compensation from airlines for delayed flights, file paperwork for government housing assistance, resolve landlord disputes, make insurance claims, apply for asylum protection, and more.

Another program with implications for access to justice is Microsoft’s recent investment of $1 million into a partnership with Legal Services Corp. and Pro Bono Net. The goal is to create “a unified legal portal in each state that employs an automated triage process to direct clients to the most appropriate help and guide self-represented litigants through the legal process.”[6] The long-term goal is to be able to provide legal aid services via natural language processing, which would work like the Amazon Echo and Apple’s Siri.

Ross Intelligence provides its programs to select law schools, bar associations and nonprofits free of charge. Its website notes that Duke Law, Vanderbilt Law School and the New Hampshire Bar Association, among others, use its program. Lex Machina also provides free access to certain university faculty and students engaged in research on or the study of IP law.

Access to justice organizations that operate on the state and local level have collaborated to provide an online forum for receiving direct pro bono legal services, ABA Free Legal Answers. The Tennessee-specific version of “Free Legal Answers” directs users to a resource called a “Legal Wellness Checkup,” created by the Tennessee Alliance for Legal Services. With this resource, users can fill out a series of forms and learn about their legal rights and areas of risk that are associated with their information through personalized links to booklets, forms and videos.

Ethical Implications of Artificial Intelligence

Under ABA Model Rule 1.1, lawyers have an ethical obligation to provide competent representation, which includes keeping abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology.

Unauthorized Practice of Law
In 2007, the Ninth Circuit held that a software called the Ziinet Bankruptcy Engine, which offered automated bankruptcy assistance, constituted the unauthorized practice of law:

The software did, indeed, go far beyond providing clerical services. It determined where (particularly, in which schedule) to place information provided by the debtor, selected exemptions for the debtor and supplied relevant legal citations. Providing such personalized guidance has been held to constitute the practice law. [The] system touted its offering of legal advice and projected an aura of expertise concerning bankruptcy petitions; and, in that context, it offer personalized — albeit automated — counsel. We find that because this was the conduct of a non-attorney, it constituted the unauthorized practice of law.[7]

Cases like Reynoso v. United States, serve as a reminder that it is important to understand the distinctions between providing legal information and legal advice.

In 2012, the Tennessee Supreme Court’s Access to Justice Commission developed General Guidelines for Distinguishing Legal Information from Legal Advice. The purpose of the General Guidelines is to assist in differentiating between providing legal information and legal advice in order to help guide not only court clerks and attorneys, but all those who are regularly sought out by self-represented litigants for legal information.

The Commission’s Self-Represented Litigants Advisory Committee defines legal information as “facts about the law and the legal process …. Providing legal information is educating the litigant.”[8] The General Guidelines further state that it is important to remain impartial and never provide information to give an advantage to one side over the other.

Legal advice is defined as “advice about the court of action a self-represented litigant to take to further his or her own best interests …. Providing legal advice is guiding or directing the self-represented litigant, especially to unique facts of the person’s case.”[9] The General Guidelines provides a list of examples of legal information and actions that can constitute legal advice to highlight the distinction.

Examples of legal information include: explaining court rules and procedures — like telling the person where to find local court rules, providing citations to cases and providing forms and instructions. Actions that can constitute giving legal advice include: providing interpretations of rules, telling the litigant what to do and drafting the wording to be entered on forms.[10]

In Lola v. Skadden,[11] the Second Circuit held that contract attorneys performing discovery document review may not be engaging in the “practice of law,” at least in North Carolina.[12] The court held that “an individual who, in the course of reviewing discovery documents, undertakes tasks that could otherwise be performed entirely by a machine cannot be said to engage in the practice of law.”

If courts hold that AI constitutes the unauthorized practice of law, legislatures may liberalize laws to specifically exclude such products as the practice of law. For example, in Unauthorized Practice of Law v. Parsons Tech Inc.,[13] the district court found that Parson’s Technology d/b/a Quicken Family Law constituted the unauthorized practice of law by providing legal templates.

Immediately after the appeal was filed in Parsons Tech, the Texas legislature enacted a statute providing that the practice of law does not include the design, creation, publication, distribution, display or sale of computer software or similar products, as long as the products clearly and conspicuously state that they are not a substitute for the advice of counsel.[14]

Supervisory Obligations
Model Rules 5.1 and 5.3 require supervisory attorneys to “make reasonable efforts to ensure that the firm has in effect measures giving reasonable assurance that,” under 5.1, “all lawyers in the firm conform to the rules of Professional Conduct,” and under 5.3, that the conduct of non-lawyers employed by, retained by, or associated with the lawyer, “is compatible with the professional obligations of the lawyer.”

One of the comments to Rule 5.3 of the ABA model rules specifically cites to a document management company hired “to create and maintain a database for complex litigation” as an example of a non-lawyer outside the firm to whom the duty to supervise would apply. The duty to supervise may also include understanding how third-party AI services work and vetting vendors for things like security to maintain the confidentiality of client information. States such as Iowa, New Jersey and North Carolina already require lawyers to vet vendors in this way.


Both the ABA Model Rules and the Tennessee Rules of Professional Conduct state in their respective preambles that “as a public citizen, a lawyer should seek improvement of the law, access to the legal system, the administration of justice, and the quality of service rendered by the legal profession.”[15] As members of a learned profession, lawyers should cultivate their knowledge of the law beyond the use of their clients, work to reform the law, and strengthen legal education. Furthermore, “[l]awyers have a responsibility to devote professional time and resources and use civic influence to ensure equal access to our system of justice for all those because of economic or social barriers cannot afford secure adequate legal counsel. A lawyer should aid the legal profession in pursuing these objectives and should help the bar regulate itself in public interest.”[16]

As the field of AI grows more sophisticated and mainstream, lawyers have a responsibility to work within the profession and with laypeople alike in exploring how AI can better serve their clients, provide access to justice and strengthen legal education.


  1. Steven Norton, “Artificial Intelligence Looms Larger in the Corporate World,” The Wall Street Journal (Jan. 11, 2017, 2:38 p.m.),
  2. “The Size of the U.S. Legal Market: Shrinking Piece of a Bigger Pie — an LEI Graphic,” Legal Executive Institute (Jan. 11, 2016),
  3. Trevor Martin et al., Legal Technology, AGC Partners Insights (Apr. 2017),
  4. The resources identified in this article should not be considered an exhaustive list of artificial intelligence programs targeted to the legal sector.
  5. Jose Vilches, “DoNotPay ‘robot lawyer’ bot can now help with 1,000 different areas of law in all 50 U.S. states,” TechSpot (July 14, 2017, 10:45 a.m.),
  6. Richard Acello, “Legal Services Corp., Microsoft plans voice response portals for legal advice,” ABA Journal (September. 2017),
  7. Reynoso v. United States, 477 F.3d 1117 (9th Cir. 2007) (citations omitted).
  8. Tenn. Sup. Ct. Access to Justice Comm’n, General Guidelines for Distinguishing Legal Information from Legal Advice (2012),
  9. Id.
  10. Some states, like Arizona and California, have created a class of individuals called certified legal document preparers, who may prepare legal documents but may not provide legal advice. See Arizona Code of Judicial Admin. § 7-208; Cal. Bus. & Prof. Code § 6400 et al. However, if a company and its software behave too “intelligently,” there is a risk that it may be accused of providing legal advice rather than legal information.
  11. Lola v. Skadden, 620 F. App’x 37, 45 (2d Cir. 2015).
  12. Id.
  13. Unauthorized Practice of Law v. Parsons Tech Inc., 179 F.3d 956 (5th Cir. 1999) (per curiam),
  14. Tex. Gov’t Code Ann. § 81.101(a)(1998).
  15. ABA Model Rules of Prof’l Conduct Preamble & Scope § 6 (2016); Tenn. Sup. Ct. R. Prof’l Conduct Preamble & Scope § 7 (2017).
  16. Id.

W. Preston Battle IV W. PRESTON BATTLE IV, was a litigation associate in Baker Donelson's Memphis office when he contributed to this article. At the time of publication, Preston will be clerking for the Hon. Thomas L. Parker, United States District Judge for the Western District of Tennessee. Battle graduated from the University of Memphis Cecil C. Humphreys School of Law in 2016.



Nicole D. Berkowitz NICOLE D. BERKOWITZ is a litigation associate in Baker Donelson's Memphis office where she focuses on intellectual property and commercial litigation. She is a graduate of Vanderbilt University and Washington University School of Law. Before joining Baker Donelson, Berkowitz clerked for Hon. Jon P. McCalla, United States District judge for the Western District of Tennessee.



Buck Lewis GEORGE T. “BUCK” LEWIS is a shareholder at Baker, Donelson, and is the Larry Wilks Distinguished Practitioner in Residence at UT College of Law, is a former Tennessee Bar Association president, a former chair of the Tennessee Supreme Court Access to Justice Commission, chair of TBA Appellate Practice Section, and chair of the American Bar Association Pro Bono and Public Service Committee. He is a graduate of the joint J.D./M.B.A. program at the University of Tennessee.