CDOIQ Symposium 2024 | Melanie Krause | IRS

theCUBE.net

Clips
News
More from CDOIQ Symposium 2024

Melanie Krause

Chief Operating Officer

Internal Revenue Service

IRS implements AI, improving the taxpayer experience, thanks to funding boost

Many people find their yearly taxes to be an onerous task but with the aid of artificial intelligence and increased funding, the Internal Revenue Service is looking to improve the IRS user experience for its taxpayers.The IRS has long valued data science and analytics as an integral part of its operations, according to Melanie Krause (pictured), chief operating officer of the IRS.Melanie Krause, COO of the IRS, talks about transforming the IRS user experience.“We are responsible for a lot of information,” she said. “We received, in fiscal year ’23, over 270 million tax returns that we processed

play_circle_outline Melanie Krause's background as nurse turned COO of IRS

play_circle_outline Experience in government accountability office and VA office of Inspector General

play_circle_outline Enhancing IRS Compliance with Data Analytics and AI for Improved Taxpayer Interactions

play_circle_outline Organization structure at IRS and relationship between CISO and CIO

play_circle_outline Enhanced Funding Accelerates IRS Progress Towards Data-Driven Decision Making Culture

play_circle_outline Marginal revenue to cost in budget negotiations for additional staff.

Info
Transcript

Melanie Krause | IRS

Melanie Krause

Chief Operating Officer Internal Revenue Service

Dave Vellante and Paul Gillin talk with Melanie Krause, the COO of IRS. Melanie discusses her background as a nurse and how it prepared her for leadership roles. She shares her journey from a nurse to the chief data and analytics officer at the IRS. The IRS uses data and analytics to support their operations, focusing on compliance activities and improving taxpayer services. Melanie also highlights the importance of data security at the IRS. The organization has a clear structure, with various chiefs overseeing different functions. The CISO reports to the CIO... Read more

explore Keep Exploring

What skills and experiences have helped prepare you for leadership roles, such as your position as COO of the IRS? add

What was the speaker's role and approach to addressing data security issues at the IRS following a large data security event? add

What methods does the IRS use to leverage data and artificial intelligence to support compliance activities and taxpayer service activities? add

What is the structure of the organization and how is the leadership roles distributed within it? add

What makes data-driven decision making core to how the IRS operates in comparison to other federal organizations you have worked with before? add

What is the most effective way to fund a country according to decades of research at the IRS? add

bolt Powered by CUBE AI

Melanie Krause | IRS

search

>> Hi everybody. Welcome back to CDOIQ 2024. We're here in Cambridge, Massachusetts overlooking the Charles River, Dave Vellante with Paul Gillin. We're getting deep into day two. Melanie Krause is here. She's the chief operating officer of the IRS. Melanie, thanks for making some time. Welcome to theCUBE.

Melanie Krause

>> Thank you for the opportunity. Thank you.

>> So you're a nurse?

Melanie Krause

>> I am.

>> Tell us about, wait a minute. But you are COO of IRS. You were the chief data and analytics officer of the IRS, the first ever. How does that happen? You love data?

Melanie Krause

>> I do, I do. No, it was an interesting journey, but I would just say I love my profession and clinical training and then experience as a nurse actually positions a lot of people very nicely for leadership roles. So the types of skills that make you an effective leader are the same that make you an effective clinician. So empathy, being able to develop rapport quickly, being able to get up to speed quickly, being able to differentiate between what's important and what's really not that important. These are the things that make one successful leader. And so that has been, I think, what made me competitive for the initial role at IRS and helped me to be successful in that capacity.

>> That makes sense. Strong with people, you can prioritize, you can work well under pressure, but still not a lot of people make that transition.

Melanie Krause

>> No. So that's fair. So when I was in my undergraduate program, there was this really strong push to move more nurses into doctoral level programs. So there were opportunities to, as an undergraduate, begin my master's training and then as I was working as a nurse, to continue on in my doctoral training. And at Wisconsin you have a lot of flexibility to, with your mentor, determine a program that meets your needs. And so in addition to coursework and nursing, I spent a lot of time in educational psychology. The Department of Educational Psychology at the University of Wisconsin has a great applied quant methods program. Spent a lot of time with the industrial engineering folks and it was a great education. By the time I was ready for my comprehensive candidacy exams, I knew that life in the university probably wasn't right for me at this time. There was a push to identify a program of research, what is it that you are going to focus on to build the science? I hate that. For me, variety is the spice of life and I want constant new opportunities to learn something different. And so moving into the government was very interesting to me because it was an opportunity to work on applied program research, to evaluate federally funded programs, to identify opportunities for improvement from a process standpoint, from a cost savings standpoint. So I spent five years at the government accountability office doing just that and then I moved over to the VA office of Inspector General. And the transition to IRS was an opportunity to, again, something completely different and to work for this important agency, which is central to funding government programs that we all care about at a really exciting time. So when I started at IRS, we were in the wake of a large data security event. And so they needed somebody who would have a steady hand to work closely with our IT partners, to work closely with the business, to identify very quickly opportunities to mitigate risks to data security in a way that still allowed people to fulfill mission essential activities. We had already a great team. Research applied analytics and statistics was a wonderful team to inherit. And so my value proposition was identifying what the key business issues were that the service was facing, how we could plug members of our team into that, where we had gaps in the work underway that we could fulfill additional needs quickly. And then to ensure that we had a person at the table who was thinking about data always, at inception through to the completion of a program.

Paul Gillin

>> Now in describing that, you don't talk about analytics, you don't talk about heavy data science knowledge or the technology. Is that because you had those resources on staff and what they really needed was a guiding hand, not-

Melanie Krause

>> An integrator.

Paul Gillin

>> Not a technology person?

Melanie Krause

>> So they needed somebody who was strong analytically. Again, somebody who understands the language, understands the issues, but at that role of the organization, what they needed most was an integrator. And that's, again, back to the clinical experience, that's what nurses do really effectively.

Paul Gillin

>> So have you picked up along the way? Have you burnished your technology skills along the way or has that really not been so important to your success?

Melanie Krause

>> So it is, but again, somebody who has, so I have strong training from the University of Wisconsin. I have a PhD conferred by the university. I am married to a person in information technology. I am professionally curious. And so having a curious mind and then the ability to, again, lift up the people, all the subject matter experts and demand experts who I work with, that, for me, has been the key to success more so than trying to be the expert in all things.

Paul Gillin

>> So what are the major analytical projects now underway at the IRS?

Melanie Krause

>> It's good question. So at IRS, we have a pretty long tradition of using data and analytics to support our work. And that's, frankly, an operational mandate. We are responsible for a lot of information. So we received, in fiscal year '23, over 270 million tax returns that we processed. We had more than 60 million taxpayers either call us for help or come into one of our taxpayer assistance centers. And so the volume of information that we were working with requires use of analytics, key use of data to drive decisions, because with that much work, one can't be successful without a strong technology, strong analytics focus. And so we have, for a long time, leveraged data to inform case selection. So for compliance activities, being able to quickly differentiate between tax returns that are likely to be compliant with tax laws from those who are not. This has also been really fertile ground for the use of artificial intelligence. So we've developed recommender systems that use unsupervised machine learning techniques to identify changes, patterns that then can be risk scored and then provided to the humans, the subject matter experts who work with us, who ultimately determine which tax returns will be examined. Equally, if not more importantly, we also use analytics and AI to support our taxpayer service activities. So beginning shortly after I joined IRS, and I can't take credit for this. This is work that was underway for years before I joined. We began incrementally delivering additional bots that help taxpayers to address simple things that they need help with, like establishing an installment agreement, which then frees up our people answering phones to help taxpayers with more complex issues or those who aren't able to negotiate the bots. We also use large language models to mine the information we have about calls that come in, so that we can identify what the major pain points are for taxpayers. What are the most problematic journeys? Which taxpayers are having to interact with us multiple times to resolve an issue? What happened? And then what are we going to put in place to address that? And then using the data that we have, have we been successful in fixing those issues? And so we have three main buckets of work, in terms of using analytics to support compliance activities, improving services for taxpayers, and then more internally-focused process improvements.

>> And when did you start in the IRS?

Melanie Krause

>> October 2021.

>> Okay. So prior to and leading up to COVID, there was a pretty big brain drain in the IRS. We saw that, as customers. It became very frustrating and the website clunky and the accountant just told me, "Just send a paper check. It's a nightmare." And then all of a sudden you guys got this big infusion of funding. I've seen a difference. It's actually quite easy to interact with. The UI still need some work, but if you're not embarrassed about UI, then you're not innovating. So that's okay. And so we've seen a difference. The few times I've had to call, I got really great assistance. I didn't have to call back six or seven or 10 times.

Melanie Krause

>> I'm so glad.

>> So we've started to see that, but help us understand, from the inside out, what's changed.

Melanie Krause

>> And I would put in a plug for the session we had earlier today. So there's been some great sessions this week at MIT, CDOIQ. Earlier today I had an opportunity to talk about some of that work that is on the horizon and work that's already in flight. Reza Rashidi, one of our executive directors with research plan analytics and statistics, and then Maya Bretzius from the Transformation Strategy Office were here to talk about that. But we've been very intentional about demonstrating and then communicating the improvements associated with the investment that we received. And so we did publish just a couple of months ago the 2024 strategic operating plan annual update. And I think that does a nice job of categorizing and describing what you got from this. What you, as a taxpayer, got from the last year of money that we invested. The transformation work is helping. And so you'll see things that are very mission-focused. There's a very deliberate effort to show that the newest tranches of enforcement activities that are really centered on the higher end, so complex partnerships, the work looking at corporate versus personal use of jets. I mean the incremental delivery on the promise to shift our attention to the higher end and the use of analytics to support that work is described in that document. There's also a lot of things on the backend that we've had underway that are maybe a little bit less sparkly, but super important. Data security. I mean data security is and will continue to be my number one priority. And so there have been a lot of activities to strengthen controls. We have been working through opportunities to improve the way that we share data with state and local governments. Tax data is really important for establishing whether outreach activities within states and communities to get people into the benefits that they're entitled to, whether they're effective. And so moving out on efforts improve the way that we share data that is compliant with 6103, the statutorily required productions for tax data, but still empowers government agencies with the information they need to see whether their programs are effective is what we have underway.

>> Can you describe the organization? Is there a CISO reporting into you? Your previous role reports into you? What does that look like?

Melanie Krause

>> That's a great question. So we recently reorganized and so we have now a commissioner who's a political appointee with senate confirmation. There is a deputy commissioner, he's Doug O'Donnell, the previous acting commissioner in between the last commissioner and this commissioner. And then we have four chiefs. And so there is myself, the chief operating officer. I have all of the operational functions, with the exception of information technology. So I have research applied analytics and statistics, procurement, human capital, risk, privacy, government liaison and disclosure. I'm probably forgetting one.

>> Big organization.

Melanie Krause

>> Big organizations. And then there's the chief information officer, who is a peer. We have the chief tax compliance officer, who oversees the main compliance function, so tax-exempt government entities, small business, self-employed, large business, international, return preparer, et cetera. And then our largest team is under our chief of taxpayer services. So that's Ken Corbin. The chief tax compliance officer is Heather Maloy. And then the chief information officer is Rajiv Uppal.

>> And the CISO reports into the CIO, is that right?

Melanie Krause

>> Correct. But we have a very close working relationship and so on activities related to data security, we've established a data security executive steering committee that pulls together the key people who oversee and are accountable for different aspects of data security. So there's the chief data analytics officer who presently is chairing it, although it's a rotational responsibility, the chief information security officer, the chief privacy officer, the chief risk officer, they're all on this team that meets frequently to, again, identify in a continuous fashion, what are we going to tackle. And then to ensure that we are holding each other accountable to dedicating the resources and making progress on what will be a never ending quest of maintaining the security of the information that we have and that we've been entrusted with, but also ensuring that we are doing so in a manner that allows people to do the important work that they need to do. So for example, a CSR, one of our customer service representatives needs to be able to access information from multiple systems on a taxpayer's account quickly. So a person calls on, they don't want to be on hold forever. They want you to be able to answer their question quickly. And so the ability to seamlessly access those systems, but only what you need, is something that I think we all struggle with, which is that tension between data at your fingertips and zero trust principles that we all struggle with.

>> So we talked earlier about last year. You were the chief data and analytics officer here at this conference, CDOIQ 2023. You were representing that role, I think, this year. There's now a lot of talk about the chief data and AI officer coming together. What do you see that's different between this year and last year and how do you see AI fitting into all this complexity?

Melanie Krause

>> Yes. So our chief data analytics officer is also our responsible AI official. And so for us, the roles are inextricably intertwined. But again, there's a very close working relationship with the CISO and the CIO. Between last year and this year, I just see progress and delivery. So last year we spoke a bit about the high level ambition, some of the work underway to develop stronger governance guardrails to ensure that we're using AI in a way that is responsible and is conformant with the guidelines and expectations established by the office management and budget. And so this year, I was sitting listening to my colleague, Reza, speak about what we did. So not the framework, but this is how we have done it. Here is the approach that we use using data sheets and data cards and here's how we've moved through the backlog of production use cases. And people were interested in how we use AI and I could point them to the inventory that is compiled with the other Treasury Bureau use cases and is on treasury's public facing website. And then we talked about the future and Taxpayer 360 and the work that we have underway, too, for our customer service representatives. Integrate the information from the nine, 10, 11, 12 systems that they currently need to access to be able to answer a taxpayer's question, and the way that we are, for this filing season, beginning to leverage generative AI to assist those staff to get quickly to the internal revenue manual guidance that they need to be able to address, not just that taxpayer's current issue, but if we believe that there are potentially related issues that aren't on the taxpayer's radar yet, that we can address that with them while they're on this phone call with us instead of having to call back in a couple of weeks when that issue bubbles up too. So then there was a question, again, from the audience, well, when is this going to happen? And we could say we are incrementally delivering this filing season and then we will deliver additional functionality next filing season. And that although we're starting with capabilities that are more internally focused for our staff, as we learn as a learning organization, we can then extend these functionalities to ideally be the types of tools that we are providing directly to taxpayers to self-serve where they're comfortable and where they wish to self-serve.

Paul Gillin

>> I wanted to ask you, we've been talking a lot today about data culture and the difficulty of establishing a culture where data is central. Where are you in that odyssey?

Melanie Krause

>> So IRS is different than other federal organizations where I've worked in that data and data-driven decision making is core to how we do what we do.

Paul Gillin

>> .

Melanie Krause

>> And again, I think it's really a function of a couple of things. We have a lot of information that we have to work through in order to be effective. And so I think just by necessity, our senior leaders have had to get comfortable with using data long before I joined the service. Also, as a tax administrator, there is a lot of interest in why and how we make decisions. And so I think with that external oversight, there's also a need to a priori think through the facts and the assumptions and the rationale and document the rationale for the decision using facts, using data, to be able to then later explain and then iterate.

>> Last question, the business case for the funding that the IRS got seemed like a no-brainer because there's a lot of fluff in the system and you guys can now get that in. That's found revenue. Correct me if I'm wrong, so obviously it was funded by Congress.

Melanie Krause

>> Yep.

>> So it went to law, right. So it's administration-independent, correct? I mean your charter, that funding, can you help us understand the life of that and any risks associated with that? Because it seems to be working.

Melanie Krause

>> So I love that question. Thank you for asking that.

>> You're welcome.

Melanie Krause

>> So we were given $80 billion in three buckets, the largest bucket being enforcement dollars. And so those are dollars that we do need to spend on compliance activities, and then smaller buckets for improving services for taxpayers and then operation support, which is a lot of things in my universe. And that is tenure funding, so that's exciting. But he who giveth, taketh away. We have had two rounds of rescissions and so we are down to about $60 billion at this point for the Inflation Reduction Act investment. And so that is still a healthy investment and we are working to deliver value. And so what we've been doing to demonstrate what we're doing is to try to be very transparent with the annual updates showing how much money we spent, in broad terms, what we spent it on, and then what you, as taxpayers, should see. Anecdotally here as well as other meetings, I've been hearing like what you shared at the beginning, that it feels better. So in addition to it feels better to a lot of people, we have examples of the value that is being delivered through this investment in tax administration. We are also, earlier this year, we had a white paper that looked at, again, the business case for investing in tax administration. And you're right that a lot of the focus for the Inflation Reduction Act was on revenue generation through compliance. And so the way that that works, and probably a more complicated way than you would want to get into now-

>> No, I'd love to hear it.

Melanie Krause

>> Is the marginal revenue to cost. And so you have baseline budget and the negotiation is around how much additional revenue would be collected through additional staff. And so it's really focused on compliance activities. However, what we know from decades of research at IRS is that the most effective way to fund a country is to support voluntary compliance services. Most, vast, vast, vast majority of taxpayers, they want to get it right.

>> Yeah, absolutely.

Melanie Krause

>> But the tax code is very complicated. There are people who are taxpayers who don't have great mastery of English, and so there's a lot of barriers to being compliant. And so what we're working on now is working through what the fact base is for the investment in services, to help people to get it right the first time, which we can't as easily fund through enforcement dollars. That this, too, is really important for the good of government, to focus not just on compliance activities, but on lifting up all of us who are trying to get it right. We're also looking at what the impact is on technology investments. So what is the business case for investing in technology that makes it easier for people to get it right, makes it easier for them to pay what they know that they owe to, again, support good of government, support sound tax administration.

>> Right. And your point about marginal cost, you're putting in infrastructure, people, processes and technology that are going to give you operating leverage down the road. And so $80 billion or $60 billion can turn into, hopefully, $100 billion or $200 billion or whatever it is over time.

Melanie Krause

>> For the economists out there, it's shifting to less costly treatment streams. And so you're helping people get it right upfront. And then for all the people who do have challenges and have compliance issues that we need to address, you're shifting them from an expensive audit, which is expensive for us, painful for the taxpayer, to things that are a lighter touch where we can get quickly into compliance and then reserve those heavier hitting resources for where we truly have a big compliance issue that we need to work through.

>> Yeah.

Paul Gillin

>> .

>> Well said. Well, congratulations on your career ...

Melanie Krause

>> Thank you.

>> And all the fine work you're doing ...

Melanie Krause

>> Thank you for having me.

>> For our country. Thank you for your service and thanks for coming to theCUBE.

Melanie Krause

>> Thank you for having me. Appreciate it.

>> You bet. Okay, keep it right there. We'll be back with Paul Gillin and Sanjeev Mohan to wrap up CDOIQ 2024. This is Dave Vellante from theCUBE. We'll be right back.