CBA Record April-May 2019

LEGAL ETHICS

John Levin’s Ethics columns, which are published in each CBA Record, are now in-

dexed and available online. For more, go to http://johnlevin.info/ legalethics/.

BY JOHN LEVIN Big Data, Artificial Intelligence,

and Legal Ethics T he New York Times of March 1, 2019 contained an article entitled “Is Ethical A.I. Even Possible?” The discussion revolved around the issue that “activists, researchers, and journalists voice concerns over the rise of artificial intel- ligence, warning against biased, deceptive and malicious applications.” Coincidently, the January-February 2019 edition of Har- vard Magazine contained a feature article entitled “Artificial Intelligence and Ethics.” That discussion referenced the fact that machine learning now entails “statistical methods [that] allow a system to ‘learn’ from data, and make decisions, without being explicitly programmed.” In effect, the program gathers “information that the algorithm uses to construct a model of the world.” It also seems that algorithms that gather large amounts of data, and then use that data to construct a model and make decisions based on that model, may incorporate the biases and prejudices con- tained in the real world into the decision making process. In a well-known example, a commercially developed program that predicted whether a criminal would re- offend was marketed to help judges make unbiased sentencing decisions. However, it was shown that this program was biased against blacks. As the Harvard Magazine article pointed out: “Even a thoughtfully designed algorithm must make decisions based on inputs from a flawed, imperfect, John Levin is the retired Assis- tant General Counsel of GATX Corporation and a member of the CBARecord Editorial Board.

GOT ETHICS QUESTIONS? The CBA’s Professional Responsibility Commit- tee can help. Submit hypothetical issues to LorettaWells, CBA Government Affairs Director, at lwells@chicagobar.org. SMALL FIRM RESOURCES One-stop shopping for resources on staring your own firm, marketing, business networking, law office technology training, low cost office man- agement consulting, and savings on insurance andbusiness expenses. Find the Portal under the Resources tab at www.chicagobar.org. system is analogous to a machine, and if the machine failed through no fault of the lawyer, the lawyer did not breach the Rules. One the other hand, two Rules deal with the responsibilities of supervisory lawyers: Rules 5.1 and 5.3. These Rules require that supervisory lawyers take reasonable efforts to make sure that lawyer and non- lawyer employees comply with the Rules of Professional Responsibility. (See January 2018 ABA Journal: “Supervision is key to effective employment of paralegals.”) So the question we will have to face is, whether artificial intelligence is more like a machine and we are primarily responsible for maintaining it properly, or whether it is more like an independent actor and we are also responsible for supervising and monitoring its behavior. I suspect that by the time the bar gets around to addressing the problem of A.I., A.I. will have become so sophisticated that we would have no choice but to treat computer assistants more like paralegals than word processors. At that point it will no longer be science “fiction.”

unpredictable, idiosyncratic real world.” So where is the ethical problem for lawyers? A lot of recent discussion has focused on the use of artificial intelligence in the practice of law. Possible uses include discovery, document review, contract man- agement, client screening, and review of employment applications. And over time, the uses will grow. Our professional behavior is governed by the Illinois Rules of Professional Conduct. The Rules contain a variety of requirements and prohibitions that could be violated if we turned decision making over to an algorithm that based its decisions on statistical norms. While the algorithm can be corrected in each instance of viola- tion, who is responsible for finding the flaw and making the correction –that is, who is responsible for the ethical behavior of the artificial intelligence? At this point you might ask whether this is science fiction. I suggest that we are going to face practical questions on how to deal with artificial intelligence in the very near future. Not surprisingly, nothing in the Rules deals with professional responsibility of artificial intelligence. There are, however, two analogous situations. The first is set out in Comment 18 to Rule 1.6: Con- fidentiality of Information. The problem is to determine the lawyer’s responsibility to secure the confidentiality of client data held by third party vendors or in cloud storage. The comment concluded that the “unauthorized access to, or the inadvertent or unauthorized disclosure of [confidential information] … does not constitute a violation [of the Rules] if the lawyer has made reasonable efforts to prevent access or disclosure.” In other words, the storage

48 APRIL/MAY 2019