ATT Editor’s Note: Robert (Rob) Kovacev is a partner in tax controversy and litigation with Miller & Chevalier in Washington, D.C., and chair of the ABA Tax Section’s Tax Policy & Simplification Committee. Rob has over two decades of experience representing clients in Tax Court and in federal and district appellate courts. Before entering private practice, Rob was a senior litigation counsel in the Tax Division of the U.S. Department of Justice, where he was lead trial counsel in some of the largest and most complex tax cases in the nation. He is a recognized leader in the emerging field of robotics and artificial intelligence, and writes and speaks frequently on these topics.
DM: It is a great pleasure to speak with you today, Rob. You recently presented to the Tax Section’s Administrative Practice Committee on the IRS’ use of algorithms in audit selection and you shared some significant concerns around the lack of meaningful human oversight over such technology. This is such an important and timely topic that I am privileged to interview you and help raise awareness.
Before we get to that though, this column serves as a way for people to get to know the interview subjects not just on a professional level, but also on a personal level. In keeping with this tradition, can you tell us a little about yourself such as where you were born and where you grew up?
RK: I was born in Whittier, CA, and grew up in Yorba Linda. I lived there until I was 18 and I am probably the least likely Californian that you have ever seen. When we first moved to Yorba Linda, it was basically a small town on the edge of the LA metropolis. The McDonald’s still had a hitching post and the wind carried the scent of oranges from nearby groves. But it was a great place to grow up. Plus we lived a few minutes away from Disneyland, and I could go pretty much any time I wanted.
DM: Can you walk us through your path towards Washington, D.C., and working for the Department of Justice (DOJ)?
RK: After I graduated from law school, I clerked for Judge Blane Michael in the U.S. Court of Appeals for the Fourth Circuit. That was a fantastic experience. Then I worked at Quinn Emanuel in Los Angeles for three years before moving to Washington, D.C., to work as an associate at another law firm there. I went to the Tax Division at the DOJ in 2006.
DM: Did you have an early experience that sparked your interest in law and then eventually tax law?
RK: In high school I had a summer internship at the Riverside County District Attorney’s Office. Although much of the work was menial, we did get to go to court with the attorneys to observe trials and motion hearings, which I enjoyed immensely. That experience convinced me to pursue law but the focus on tax law came later. I studiously avoided tax courses in law school and (like many other law students) waited until the second semester of my 3L year to take a tax course. I loved it and was then kicking myself that I had not taken tax earlier and more tax classes, but by then it was too late. After law school, I worked in civil litigation first, but I knew I wanted to go into government. The Tax Division was hiring so I jumped on the chance to go to DOJ, and the rest is history.
DM: Can you tell us a bit about your government experience and how that impacted your career path?
RK: The DOJ Tax Division was a terrific place to work and I enjoyed my work there immensely. In one day I went from being an associate carrying partners’ briefcases at a deposition to having a stack of 30 active cases with deadlines coming up, and I was the only lawyer who was in charge of all those cases. I remember that by Friday of my first week I had my first solo hearing on a summary judgment motion in front of a federal district judge. That would never have happened for me as an associate in a big law firm.
It really opened my eyes to the fact that when you are in government, you get opportunities that you just cannot get in the private sector. And you get them when you are much younger. There are some skills that you can learn by watching other people and studying, but there are some things which you really have to do in order to master them. For example, with cross examining a witness, the rule of thumb is you have to have 50 cross examinations under your belt before you can say you actually know what you are doing, and my own experience confirms this.
DM: What was mentorship like at the DOJ Tax Division?
RK: When you have a heavy caseload and you are in government, there is no pyramid of partners, associates, and staff working on cases. It is basically you, except for the largest cases which may have two or three trial attorneys assigned. However, with each section you definitely have certain mentors and senior advisors. Pat Mullarkey was the Chief of the Northern Trial Section of the Civil Section of the Tax Division when I was there. He had been at the Tax Division for over 50 years and he had pretty much seen and done everything. He always had insight into just about any issue that would come up and that was really valuable to me.
DM: Was it challenging to make the transition back into private practice after you left the DOJ?
RK: Oh yes, it was challenging and that is because the focus changes. When I was at the DOJ, there was no shortage of cases. They just showed up on my desk, usually many of them, and I did not have to ask for them. In private practice, I learned very quickly that no one is showing up with cases; instead, you have to go get them. So I had to learn how to go out and talk to potential clients and how to develop business. I had to go to conferences and apply a host of different skills. When I was a trial attorney at DOJ, I did not have to worry about where my next case was coming from. Of course there are different challenges that come with having individuals, corporations, and family groups as clients versus your client being the IRS. There are pluses and minuses to both.
DM: Can you tell us about your current firm, Miller & Chevalier, and why you decided to join the firm?
RK: Miller & Chevalier is one of the preeminent tax firms in the country. We only have one office, in Washington, D.C., so all the tax partners are in one building, which I love because I can just walk down the hall if I have a question. We have meetings and we just talk about tax issues, not necessarily even about a case, but just about some legislation that is coming down the pike or a fascinating issue that has come up in that respect. It is a lot like DOJ, where you would just walk down the hall and people would have their door open and you could just talk about things that were going on in the tax world or about your cases. We can have those types of conversations very easily with each other and that is one of the things I treasure most about Miller.
DM: I am really excited to ask you about your focus on the legal aspects of emerging technology and the regulation of AI and robotics. Can you walk us through how your expertise grew in that direction?
RK: Not long after I left the government, Bill Gates began promoting the idea that robots will displace humans in the job market: he suggested we need to have a tax on robots to replace the tax revenue from jobs lost to automation. Suddenly everyone was talking about robot taxes, but I noticed that what was absent from the discussion was anyone who actually knew anything about tax. I decided to write a couple of articles on the subject from a tax practitioner standpoint, not just as a politician or a business person or an economist. Because of that article, I was invited to speak at a robotics law conference at Stanford University.
While I was there, I met people who were working on AI and algorithms and discussing many of the legal issues that arise from the use of algorithms, algorithmic decision-making in governments, and other issues like that. I had never heard about such issues so I started researching and authoring articles on them. I published articles in popular newspapers but also in law reviews and tax publications, which were all helpful in establishing and sharing my expertise in this area.
DM: Can you explain briefly what your concerns are with the IRS’ use of algorithms?
RK: To be clear, I am not against the IRS using algorithms or AI or machine learning in the right way. It is a very efficient way to spot compliance issues, given the substantial amount of data in IRS files. With that use, however, come certain concerns. Whenever you have any algorithmic or automated decision-making process, you have to ask questions. Is it being done fairly? Is it being done transparently? And is it being done without inherent biases in the data?
We have started to see areas where the use of algorithms has created racial and income disparities. One non-tax example is in Wisconsin where the state was using an algorithm for criminal sentencing decisions. The algorithm was designed to create a recidivism prediction: a judge would take the prediction into account in determining whether to give a harsher or lighter sentence. However, it was discovered that the algorithm decided a particular defendant was more likely to re-offend merely because the defendant was Black. It was not how the algorithm was designed, but it was how it evolved as it looked at the data. The recidivism index in Wisconsin has led to a tremendous amount of litigation and controversy.
This kind of bias shows in another area. The Institute for Economic Policy Research at Stanford University, in conjunction with members of Treasury’s Office of Tax Policy, recently released a study of audit rates for individuals, broken down by race. That study shows that Black taxpayers were between three and five times more likely to be audited than taxpayers of other races. You have to wonder why that would be so. Based on analysis of IRS audit data, the study attributed a significant part of this disparate audit rate to an IRS algorithm. It is quite easy for these biases to sneak in without anyone intending them to happen and without anyone realizing they are happening until it is too late.
DM: What do you think would be the best remedy for the issues and challenges associated with automated processes and flawed algorithms? How do we ensure the issues we are seeing with the Service’s initial investments in AI and data analytics are not repeated going forward?
RK: There has to be a commitment to ensuring fairness and avoiding bias. In addition, there has to be transparency. In our voluntary compliance system, taxpayers need to have confidence the IRS is enforcing the tax laws fairly and equitably and without bias.
In other words, what we need is both greater transparency and oversight, two things that are currently missing from the Service’s Data Analytics program. Transparency gives the public the confidence that the IRS is trying its best to be fair and avoid bias. Oversight is necessary to ensure that the IRS is actually taking those steps to avoid bias. That oversight could come from Congress, TIGTA, or perhaps an oversight board that is dedicated to reviewing the technological tools used by the IRS to ensure that the tools are being monitored for potential biases and fairness issues.
DM: Do you perhaps have an example of where such oversight is being applied in an effective way?
RK: Yes. Europe is taking the regulation of AI much more seriously than the U.S. Member States in the EU are starting to engage in government agency regulation of AI. As an example, the Netherlands was utilizing an algorithm to spot child care benefits fraud that was disparately flagging immigrants from certain countries for committing fraud. The difference, though, is that EU has more of a mechanism in place to address these issues.
The Biden administration recently issued an executive order specifically directed at requiring government agencies to ensure that any algorithmic or technological tools being used are cognizant of the concerns around fairness and bias. This is a great first step but ultimately what is needed is active oversight. In my opinion, the best way to ensure the executive order is followed is through an oversight board charged with that responsibility. Creating a mechanism for oversight would bring the U.S. closer to the level of oversight that is being implemented in other nations.
DM: You have been very active in the Tax Section and currently chair the Tax Policy & Simplification Committee. Can you tell us how you first got involved in the Tax Section and do you have any guidance for new members on how to make the most of the opportunities that come with Tax Section membership?
RK: I credit Lisa Zarlenga of Steptoe & Johnson LLP because when I left government, I worked with Lisa at Steptoe and she is incredibly active in the ABA Tax Section. She encourages her colleagues to be active in the Tax Section and she first inspired me to get involved.
In terms of how to make the most of the opportunities that come with Tax Section membership, if you are a new Tax Section member, it is easy to get lost. You may come to the annual meeting and find there are thousands of tax lawyers who all seem to be good friends and know each other, and you are there on your own. That can be very intimidating, but the best way to become integrated is to volunteer for things. Comment projects are always a great way to get involved because you can get in on the ground floor of interesting policy issues that are relevant to your practice.
Attend committee planning meetings, volunteer for assignments, and just be active so people get to know you. If you have an idea for a panel, let committee leaders know: people are always looking for ideas and for new faces to present. There are many opportunities, but you have to go and actively participate. I am so glad that we now have in-person meetings again. The discussion is better, and in-person meetings provide an opportunity to exchange ideas with people you do not see on a daily basis—people from around the country, even around the world.
DM: If you could look into a crystal ball and see the future of tax administration and the practice of tax law, how would you describe that vision?
RK: There will be more technology and that means that there will be more technologically oriented people. The IRS is hiring data scientists, not just tax lawyers and CPAs. I think over the next 10 years or so we will see the technological aspects of tax becoming much more important. Lawyers and other practitioners will need to be aware of these technologies and what they can accomplish and be able to use them as tools. No one is talking about a computer replacing a tax lawyer, but certainly they can be valuable tools if you know how to use them. The challenge is we need to learn about them so that we know how to use them.
We will also see the effect of the Inflation Reduction Act money as it comes through and is utilized to fulfill the IRS’ Strategic Operating Plan. $80 billion is a lot of money. Over the past 10 or so years, the IRS has been starved for resources, so there are a lot of improvements on the customer service side as well as the enforcement side that they just have not been able to accomplish. Now they have a window of opportunity to do that. It could revolutionize tax administration in this country, which will obviously have huge effects on the private sector for tax professionals as well.